public personnel management - ipma-mc · public personnel management subscription order u.s. -...

Public Personnel ManagementVolume 40 No. 1 Spring 2011

Contents

Competency-Based Management—An Integrated Approach to Human Resource Management in the Canadian Public Sector . . . . . . . . . . . . . . . .1By Arieh Bonder, Carl-Denis Bouchard and Guy Bellemare, PhD

Does Public Service Motivation Affect the Occupation Choices of Public Employees? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11By Leonard Bright, PhD

Diversity and Representation in the U.S. Federal Government: Analysis of the Trends of Federal Employment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25By Sungjoo Choi, PhD

Sexual Harassment Training: Person versus Machine . . . . . . . . . . . . . . . . . . . . . .47By Melissa K. Preusser, Lynn K. Bartels, PhD, and Cynthia R. Nordstrom, PhD

The Impact of Management Work Habits on Public Sector Performance: A Study of Local Government Managers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63By Jonathan P. West, PhD, and Evan M. Berman, PhD

PublicPersonnelManagementIPMA-HR’s quarterly journal focusing on trends, case studies,and the latest research by top human resource scholars andindustry experts.

Online Access now provided with all print subscriptions to Public Personnel Management!

Visit http://www.ipma-hr.org/and click on “Publications”to access Public PersonnelManagement online.

Call (703) 549-7100 for assistance.

A subscription form for Public Personnel Managementappears on page vi to pass along to your colleagues!

Copyright ©2011

International Public Management Association for Human Resources1617 Duke StreetAlexandria, VA 22314tel: (703) 549-7100fax: (703) 684-0948Internet: www.ipma-hr.orgE-mail: [email protected]

“Competency-Based Management—An Integrated Approach to Human Resource Management inthe Canadian Public Sector,” Arieh Bonder, Carl-Denis Bouchard and Guy Bellemare, PhD, 40(1), 1

“Diversity and Representation in the U.S. Federal Government: Analysis of the Trends of FederalEmployment,” Sungjoo Choi, PhD, 40(1), 25

“Does Public Service Motivation Affect the Occupation Choices of Public Employees?” LeonardBright, PhD, 40(1), 11

“The Impact of Management Work Habits on Public Sector Performance: A Study of LocalGovernment Managers.” Jonathan P. West, PhD, and Evan M. Berman, PhD, 40(1), 63

“Sexual Harassment Training: Person versus Machine,” Melissa K. Preusser, Lynn K. Bartels, PhD,and Cynthia R. Nordstrom, PhD, 40(1), 47

Index to Volume 40 - by Title

Arieh Bonder, Carl-Denis Bouchard and GuyBellemare, PhD, 40(1), 1

Leonard Bright, PhD, 40(1), 11

Sungjoo Choi, PhD, 40(1), 25

Melissa K. Preusser, Lynn K. Bartels, PhD, andCynthia R. Nordstrom, PhD, 40(1), 47

Jonathan P. West, PhD, and Evan M. Berman, PhD,40(1), 63

Index to Volume 40 - by Authorw

ww

.bid

dle.

com

Contact Us for a TVAP Demo Today800.999.0438 ext. 180 · [email protected]

Ask about our other EXPERT series products.

EXPERT: Test Validation & Analysis ProgramThe TVAP software automates the multiple processes necessary to validate and analyze written tests. TVAP uses a content validity approach to validate KSA-based tests necessary for job success.

To help ensure that your testing processes are legally-defensible, TVAP includes the following calculations and analyses:

Rater analysis and bias evaluationItem analysis (including Differential Item Functioning, or “DIF”)Test-level analyses (descriptives, reliability, conditional SEMs)

Adverse impact analysesTop-down score banding using conditional SEMs

Guidelines for Contributors

As a leading, peer-reviewed journal in public sectorhuman resources, Public Personnel Managementparticularly encourages manuscripts from apractitioner’s perspective, as well as submissions onemerging national and international trends in publicpersonnel management. Articles in response topreviously published manuscripts are also welcome.

1. Public Personnel Management, the journal of theInternational Public Management Association for HumanResources (IPMA-HR), encourages contributions on allaspects of personnel management in the public sector.Manuscripts for publication should be sent to the editor.Authors are strongly advised to consult one or morerecent issues of the journal before submittingmanuscripts for publication. For questions or guidance indeveloping a manuscript, you may contact ElizabethKirkland, Editor, at [email protected].

2.Manuscripts are subject to a blind review process.Reviewers submit their recommendations as toacceptance, revision or rejection. In the cases of revisionor rejection, the unidentified reviewers comments will besent to the author.

3. Contributions should be accompanied by a statementthat if accepted for publication in Public PersonnelManagement, they will not be published elsewherewithout the agreement of the editor.

4. Articles are accepted for publication on theunderstanding that they are subject to editorial revisionand that the right of publication in any form or languageis reserved by the Association.

5. Manuscripts should be in English. Submit five copiesof the manuscript. If less than five are submitted, youwill be requested to send the additional copies.Manuscript copies must be double-spaced throughout,and submitted on 81⁄ 2” x 11” white paper, single sided.Pages must be numbered consecutively. Manuscriptsmust also be submitted on a diskette at the time ofsubmission in Microsoft Word text format. All diskettesmust be PC-readable. Authors should keep an original ofthe manuscript. Manuscripts may also be submitted viae-mail to Debbie Tankersely-Snook at [email protected]. E-mail submissions are acceptable on firstsubmission and must be sent as “attached” files with theattached file in Microsoft Word text format.

6. Place the title of the manuscript, the author’s name,address, telephone number, e-mail address (required) anda biographical sketch of no more than 50 words on aseparate cover page—please DO NOT send completeresumes for a biographical sketch. In the case ofcoauthors, respective addresses, telephone numbers, e-mail addresses, and biographical sketches should beclearly indicated, as well as the author who is to receivecorrespondence (primary author). Please notify theeditor of any change of address that occurs while a paperis under review, or in the process of publication.

7. The essential contents of the manuscript should besummarized in a single-paragraph abstract on a separatesheet following the cover page. The title of the articleshould appear again above the abstract, without theauthor’s name, as a means of identification. Manuscriptsare circulated for review without identifying the author.

8. The title of the article should appear again on the firsttext page of the manuscript, without the author’s name,as a means of identification.

9. All tables, photographs, maps, charts and diagramsshould be referred to as “Figures” or “Tables” and shouldbe presented either in position or at the end of themanuscript. All tables, charts and diagrams should alsoappear on the diskette. They should be numberedconsecutively (in arabic numerals) as they appear in thetext and should have informative titles. Their positions inthe manuscript should be indicated if presented at theend of the article.

10. For general style and submission questions, IPMA-HRfollows the guidelines included in the APA (AmericanPsychological Association) Publication Manual; or, youmay contact the editor.

11. Public Personnel Management does not acceptmanuscripts that use the author/date style of references.Authors should use the automatically coded endnotes(not footnotes) function in the MS Word program.References should be presented as endnotes and benumbered consecutively in superscript. All notes shouldbe presented in a separate, double-spaced listing at theend of the article in both the paper copy and disketteversions of the manuscript.

12. Because of the difficulties of scheduling andtransmission delays, proofs of articles accepted forpublication cannot be sent to authors.

13. Articles may be reproduced for internal,noncommercial use without authorization, provided thesource is identified. Requests for permission to publish,reproduce, or translate articles should be made to theeditor.

Send articles to:

Editorial OfficePublic Personnel Management1617 Duke StreetAlexandria, VA 22314E-mail: publications @ ipma-hr.org

Public Personnel ManagementSubscription Order

U.S. - Print and Online Rest of the World - Print and Online1 Year (4 issues) 1 Year (4 issues - includes air mail shipping)

�� $130.00 �� $175.00

Online Only 1 Year (4 issues)

�� $100.00

Subscriptions run for a period of one year (12 months). Issue dates for the year arespring (March), summer (June), fall (September) and winter (December). The subscrip-tion term may start at any time during the year. Requests for retroactive subscriptionstart dates will be honored when supplies of back issues allow. Claims are accepted forone year from publish date only.

Issue Start Date: �� January 1 �� April 1 �� July 1 �� October 1

Subscriber Name/Organization: ____________________________________________________

Address: _______________________________________________________________________

________________________________________________________________________________

City: _______________________________________ State/Province: ______________________

Zip/Postal Code: _______________________ Country: ________________________________

Telephone: (________)______________________ Fax: (________)_______________________

Email: _________________________________________________________________________

Payment: Total Order Amount $___________________

�� Check

�� Credit CardVISA/MC ________________________________ Exp Date______

Return To:IPMA-HR – Subscriptions, 1617 Duke Street, Alexandria, VA 22314tel: 703.549.7100; fax: 703.684.0948; email: [email protected]

�

Return To:

IPMA-HR - Subscriptions, 1617 Duke Street, Alexandria, VA 22314

tel: (703) 549-7100; fax: (703) 684-0948; e-mail: [email protected]

Subscriptions run for a period of one year (12 months). Issue dates for the year are spring

(March), summer (June), fall (September) and winter (December). The subscription term

may start at any time during the year. Requests for retroactive subscription start dates

will be honored when supplies of back issues allow. Claims for missing issues will only

be accepted for the issue immediately prior to the current one (i.e., if the winter issue is

out, claims will only be accepted for the fall issue prior to that winter issue). Customers,

including subscription agencies, claiming more than one missing issue for the same ad-

dress must provide their FedEx number for any amount of copies being shipped that ex-

ceeds one copy, and will be charged shipping and handling for copies that exceed a

singular copy.

Issue Start Date: � March 1 � June 1 � September 1 � December 1

Subscriber Name/Organization:____________________________________________________

Address: _______________________________________________________________________

________________________________________________________________________________

City: _______________________________________ State/Province: ______________________

Zip/Postal Code: ________________________ Country: ________________________________

Telephone: (________)_______________________ Fax: (________)________________________

E-mail: _________________________________________________________________________

Payment: Total Order Amount $___________________

� Check

� Credit Card

VISA/MC ________________________________ Exp Date______

U.S. - Print and Online Rest of the World - Print and Online

1 Year (4 issues) 1 Year (4 issues - includes air mail shipping)

� $130.00 � $175.00

Online Only

1 Year (4 issues)

� $100.00

Competency-BasedManagement—AnIntegrated Approach toHuman ResourceManagement in theCanadian Public Sector By Arieh Bonder, Carl-Denis Bouchard and Guy Bellemare, PhD

With a new appreciation for the value of their employees, many organizations aremoving rapidly to embrace Competency-Based Management, a relatively newapproach to human resource management. Competency-Based Managementinvolves the management of key HR activities such as staffing, learning andperformance management, around a competency profile for the work to be carriedout. This article describes how Service Canada, a key service delivery agencywithin the Canadian Federal Public Service, was able to bridge the two worlds ofjob analysis and competency modeling in order to successfully implement aworking competency framework in a large unionized organization.

IntroductionIncreasing productivity and efficiency in managing people has become a focal point,not only in private sector companies, but also in public organizations.1 Although theyare facing different challenges, public entities are required to quickly adapt to on-goingeconomic, political, technological and environmental changes. Furthermore, in thecontext of globalization, traditional human resources practices are being challengedjust as new approaches to human resources management are being implemented.

One of these approaches, Competency-Based Management (CBM), themanagement of key human resource processes around a competency profile for thework to be performed, has been underway in many organizations for the past 15 to 20years. By using a common language, by reflecting the values and mission of theorganization and by establishing clear expectations, CBM helps integrate humanresource activities into the business strategies of the organization. This paper exploreshow Competency-Based Management was applied in the Canadian federalgovernment, within Service Canada, a new service delivery agency.

Public Personnel Management Volume 40 No. 1 Spring 2011 1

Competency-Based Management: Focus on PeopleAs part of their new focus on people, today’s organizations have begun to place anincreased emphasis on human resource planning and what is now often referred to as“talent management.” These activities focus not only on examining the demographicsof the employee population, but also on an analysis of the competencies (theknowledge, skills and personal attributes) required of employees in order fororganizations to successfully achieve their mandates.

The identification and assessment of competencies, as well as the development ofcompetency-based management frameworks to support activities such as “gapanalysis,” recruitment, learning and other key human resource processes, all reflect anacknowledgement by organizations that their workforce is key to their success in themodern workplace. As Lawler has pointed out, competencies refer to the “skill setsthat are appropriate and unique to the organization and that will provide corecompetencies and competitive advantage.”2 Using competencies, organizations cannot only highlight the knowledge, abilities and personal qualities needed for success inkey jobs but can also identify the qualities needed for success across all jobs in theorganization.

Improving Service Delivery to Canadians Coast toCoast Service Canada is a new federal government organization, launched in September2005, aimed at providing a one-stop service to Canadians seeking access to a widevariety of government programs. Spread out across the country, with a staff of over22,000 employees, Service Canada is one of the largest federal governmentorganizations. Canadians can call a single telephone information line which is staffed byfully-trained agents who ensure people are connected to the services they need. AService Canada Web site connects them to online services around the clock and in 320Service Canada offices and 120 Outreach sites across Canada, frontline staff meetpersonally with Canadians to provide service.

As the “social face” of government, Service Canada delivers programs rangingfrom employment insurance assistance to old-age pension payments to programmesaimed at youth, students and people with disabilities. Each year Service Canadaprocesses more than five million applications for benefits and pays out over 65 billiondollars to Canadians.

Implementing Competency-Based ManagementWithin Service Canada: Origins of the Journey Service Canada’s competency work began in 1995 in its predecessor department,Human Resource Development Canada. It was initiated by a small group of HRAdvisors as an attempt to help integrate existing competency efforts that had sprungup across the department at the time. The aim was to develop a corporate CBM

Public Personnel Management Volume 40 No. 1 Spring 20112

framework for the organization. In the mid-90s, CBM was widely hailed across bothpublic sector and private sector organizations as a new and innovative approach to HRManagement. Numerous conferences on the subject were held in North America andEurope and countless consultants were engaged by both large and small organizationsto carry out “competency modeling” exercises. In the absence of any central orcorporate direction on the subject, many departments and agencies within theCanadian public service, including Service Canada, also began to undertake their ownindividual projects.

Learning from the Past The early framework or “architecture” for CBM in Service Canada was developedinternally, based upon best practices from the private and public sectors. From thestart, the intent was to develop a framework that provided for the integration of key HRprocesses around a competency profile. This can be seen in the logo (Figure 1)adopted early on in the process.

Service Canada’s approach to CBM was also fostered by a healthy respect for thelearnings of the past, specifically for the body of knowledge covered by the discipline ofjob analysis. This resulted in the development of a CBM framework based on theprinciple that “the job” was the fundamental unit in every organisation and that theingredients of job success were the traditional “KSAOs” (Knowledge, Skills, Abilitiesand Other Characteristics) of job analysis. This awareness led to the department’sdefinition of a competency, incorporating these elements, and to the decision on whata competency profile would look like.


Figure 1

TM

Building on this, a competency came to be defined as: “Any knowledge,skill/ability, or personal quality, demonstrated through behaviour that results in serviceexcellence.” Additionally, the structure of a competency profile in the organisation wasdesigned to reflect “core competencies” required of all employees (personalcharacteristics for the most part), “group competencies” required for certain job roles(primarily abilities and skills), and “task competencies” related to specific jobs(primarily knowledge). As leading industrial psychologist Kenneth Pearlman has said“good competency modelling has the ability to resonate with and communicate tomanagers in a way that is meaningful to them.” But, he added, “effective modelling”also applies “conventional, rigorous job analysis methods.” 3

Why the concern abut job analysis, particularly at a time when competenciesseem to be the primary focus in the workplace today? Job analysis is important becauseit provides an objective picture of the job, not the person performing the job, and assuch, provides fundamental information to support all subsequent and related HRactivities, such as recruitment, training, development, performance management andsuccession planning. Job analysis serves two critical functions with respect to theseprocesses. It helps ensure that decisions made with respect to HR processes are gooddecisions, i.e. fair and accurate (e.g., selection of the right person for the job,appropriate decisions about training, performance management, development, etc.)and it helps ensure the defensibility of decisions made (e.g. demonstration of the bonafide requirements used as the basis of selection). The identification of competenciesfor a job, based upon a rigorous job analysis, helps ensure the defensibility of humanresource tools used to assess those competencies, including selection tests andinterview tools.

Mindful of the need to ensure the job-relatedness of its competency work, ServiceCanada’s CBM framework used, as its basis, the existing job classification system inplace within the Canadian public service. Essentially, this involved a process ofcompetency profile development for key jobs in the organization, supported andinformed by the job descriptions for each of these positions.

By focusing its competency efforts at the level of the job (i.e. core, group and taskcompetencies) and building upon an established job analysis system, Service Canadawas able to build a solid and enduring foundation for its competency initiative. In thisway, too, Service Canada demonstrated how the two worlds of job analysis andcompetency modeling could be bridged (as opposed to favouring one or the other) toprovide a defensible, integrated and practical approach to human resourcesmanagement in a large government organization. In contrast, multiple competencyprojects across the Canadian Public Service, indeed in the private sector as well, havecome and gone, as Service Canada’s CBM framework took root and became the way inwhich HR business was conducted within the organization.


The Question of Scales An additional feature of the early work carried out to develop a CBM framework inService Canada was the decision to use a simple five-point universal Likert scale todifferentiate levels of competence (see Table 1)

Most competency efforts use scales to show how competencies can be possessedat different levels, but for the most part, organizations build scale descriptors withineach of the competencies they identify. This results in a great deal of fine detail andoften voluminous reading for the practitioners of the competency “tools.” However, asRankin has pointed out, “Competencies should be designed with the realization thattheir users are fallible human beings. The human mind is unable to handle largeamounts of detail. Expecting users to cope with several competencies, each of whichhas to be understood at many different levels of performance, is unlikely to succeed.”4

In contrast, Service Canada’s modular approach to scale usage proved to be a practicalapproach to differentiating competency levels, while still providing sufficient utility tousers who needed to make competency based assessment decisions for various HRprocesses.

Core Competencies for All Employees As indicated previously, one of the key decisions made in building a CBM frameworkwithin Service Canada was to identify a set of core competencies that would apply to allemployees. The decision to go with core competencies was taken early on, guided by


Table 1: Service Canada CBM Scale

0 Cannot Rate – Insufficient information to assess.

1 Introductory – Little or no knowledge/proficiency. Rarely demonstrates. Needs significant development.

2 Basic – Basic knowledge/proficiency. Sometimes demonstrates. May need development.

3 Proficient – Knowledgeable/proficient. Usually demonstrates. Little development required.

4 Very Proficient – In-depth knowledge/proficiency. Demonstrates most of the time. No development required.

5 Mastery – Expert knowledge/proficiency

Table 2: Client Focus - Definition and Behaviours

Client FocusAn underlying concern for helping internal and/or external clients and for being responsive totheir concerns

• Demonstrates respect and concern for every client

• Provides quality service

• Works with clients to achieve results

• Consistently delivers on commitments

the belief that one of the greatest values of CBM was that it provided a means to reflectand operationalize organizational values. For example, the core competency “ClientFocus” (see Table 2) reflected the service oriented and learning culture of ServiceCanada, a competency required by all employees.

Competencies for Every Job Beyond the core competencies, however, and as suggested earlier, it was determinedby the HR team designing Service Canada’s CBM framework, that in developingcompetency profiles, it was important to reflect competencies right down to the joblevel. In other words, neither core competencies, nor competencies for job roles orfamilies were sufficient in order to adequately manage an organization’s HR activities.What was needed was a “multiple-job approach” to competency development, one thatreflected the full spectrum of competencies needed to carry out any one job, fromabilities and skills, to personal qualities, to knowledge requirements. As Mansfield haspointed out “The multiple-job approach to developing competencies offers the dualadvantages of a common conceptual framework and customization for individual jobs.This is the only approach that facilitates comparison of competency models with eachother and comparison of employee profiles with multiple jobs.”5 It is this model thatService Canada eventually followed, integrating the range of competencies needed forall jobs within the organization within a national competency dictionary.

Building CBM in a Unionized Environment The Canadian federal public service has long been a unionized organization anddeveloping a new initiative within this environment, such as CBM, meant that unionconsultation would be key to the successful implementation of any such venture. Thebenefits of CBM, in ensuring the transparency of human resource processes, was clearto union representatives from the beginning. This was evidenced in a comment fromone representative who remarked, “What’s there to argue about?” following apresentation on CBM. Indeed, CBM is a human resource activity that is unique inproviding advantages, not only to an organization as a whole, but to every individualemployee as well. Union support for the CBM initiative was further re-enforced byinvolving union representatives, along with job incumbents and supervisors, in thedevelopment of the first competency profiles for the key front-line jobs within theorganization. Union consultation and involvement will continue to be a key componentof the development of CBM in Service Canada.

CBM in Service Canada Today Today, centered on its national competency dictionary, Service Canada’s CBMframework includes a definition of competencies, a standard format for a competencyprofile, the universal five-point rating scale and an Intranet-based “Web Suite.” Inaddition, Service Canada has developed more than 140 competency profiles (eachbased on a job description) covering more than 12,000 employees. Service Canada’s


CBM Web Suite is accessible to all employees and provides information on CBM as wellas tools such as the national competency dictionary, competency profiles, competencyself-assessment questionnaires for employees, assistance on developing learning plansand the “National Learning Inventory” which links all departmental learning anddevelopment activities to competencies. The site also provides online tools andinformation for managers to assist them in applying competencies to staffing and otherHR processes.

Day to day, competencies now appear on the selection notices used to staffpositions within Service Canada, employees use the same competencies to identifylearning needs on the job and these competencies are also used as the basis ofperformance management discussions between employees and their supervisors.Additionally, CBM also plays a key role to assist in the assessment and reclassification ofemployees in those positions. More than 10 years after launching this initiative,competencies are being used as part of the routine HR operations of Service Canadaand serve as the foundation for human resource management in the organization.

The Future While the foundation for CBM has been established, much remains to be done. Work iscontinuing to educate and train users, to develop profiles for new jobs, to update thedictionary and to enhance the HR tools on the website. Additionally, another challengefor the initiative, as indicated at the start of this article, was to have the senior managersof the organization endorse this new approach to HR management so that it trulybecame “the way we do HR business around here.” Although widely used across theorganization, CBM was applied on a voluntary basis by managers, so that thereremained pockets of the department where competencies were not consistently used.The main reason for this situation is that numerous organizational changes over theyears had prevented senior management from fully turning their attention to initiativessuch as CBM. With senior management support, competencies could finally serve asthe single foundation for HR management within the department.

A key step in this process has already occurred. Only recently, a presentation onCBM and its status within the organization was made to Service Canada’s seniormanagement’s team human resource committee. The result of the presentation was aclear consensus by the management team to move the endeavour forward. With theendorsement of senior management in place, the next step planned is acommunication strategy to ensure that managers and employees across theorganization are aware, both of the CBM framework and the tools and processesavailable to them, and of the need to ensure an integrated and consistent use of CBMacross Service Canada. An additional step will be the development of a governancestructure for CBM, outlining how key decisions related to the subject, such as changesto core competencies, will be made by the organization.


Conclusions and Implications While there is not, at this time, quantitative data available within Service Canada to saythat CBM is better or more efficient than traditional HR processes, it does seemsintuitively obvious that by linking all HR processes around a common language, anorganization should be able to achieve greater efficiencies across its HR processes. As aminimum, managers in Service Canada using CBM believe that they are significantlyincreasing the transparency of their HR transactions and, as a result, are creating aworkplace that fosters the respect of employees and makes their organization anemployer of choice.

What are the implications of Service Canada’s work for other organizations? Asindicated earlier, it is clear from this work that the two worlds of job analysis andcompetency modeling can not only co-exist, but even more so, can provide asynergistic benefit to the human resource management practices of an organization.

As indicated by Schippmann et al at the conclusion of a major review carried outby a committee of Industrial/Organizational Psychologists on competency modellingand job analysis, “what the future might hold is a blurring of borders as thecompetency modeling and job analysis approaches evolve over time. Thus the nextgeneration of approaches in each case may result in a blending of best practices suchthat there are more similarities than the differences that exist today.”6 Service Canada’swork on competency-based management serves as one example, at least, of thewisdom of their prediction.

Finally, it is worth mentioning that Service Canada’s work has also been taken farbeyond the boundaries of the organization with presentations on the CBM initiativebeing provided to other government departments, to HR practitioners in private sectorconferences and to public service officials from other governments, including the USA,Mexico, China, Poland, Russia, Malaysia and the Congo. With a rare working model ofcompetency-based management in place, Service Canada has much to offer to the restof the world in demonstrating the value of this new approach to human resourcemanagement.

Notes1 Slitter, J., Bouchard, C.D., & Bellemare, G. (2005). Managing Police Resources: A Competency-Based Approach to Staffing. Journal of Financial Crime, Cambridge University, 12, 327-33

2 Lawler, E.E. (1993). From Job-based to Competency-based Organisations. Journal ofOrganizational Behaviour, 15, 13-15.

3 Pearlman, K. (2002). Competency Modeling: Mirror into the 21st Century Workplace-or JustSmoke. Presentation at the 26th Annual International Public Management AssociationAssessment Council Conference on Personnel Assessment, New Orleans, July 2, 2002.

4 Rankin, N. (2004). The New Prescription for Performance: the Eleventh CompetencyBenchmarking Survey. Competency and Emotional Intelligence, 1-48

5 Mansfield, R. (1996). Building Competency Models: Approaches for HR Professionals. HumanResources Management, 35, 46-51.

6 Shippmann, J.S, Ash A, Battista M, Carr L, Eyde L.D, Hesketh B, Kehoe J, Pearlman K, Prien E.P,Sanchez J.S. (2000). The Practice of Competency Modeling. Personnel Psychology, 53, 703-740.


AuthorsArieh Bonder

Human Resources Branch, Service Canada140 Promenade du PortagePhase IV, Level 0, OB117Gatineau, Quebec, Canada, K1A 0J9(613) [email protected]

Carl-Denis Bouchard

Canada Public Service Agency90 Sparks Street7th FloorOttawa, Ontario, Canada, K1A OR3(613) 992-2870 [email protected]

Guy Bellemare, PhD

Université du Québec en OutaouaisPavillon TachéC-3712, Post Box 1250, Station HullGatineau, Québec, Canada, J8X 3X7(819) 595-3900, ext. [email protected]

Arieh Bonder is senior personnel psychologist with Service Canada, in the Canadian fed-eral public service, where he is responsible for developing assessment techniques andadvising managers on personnel selection issues in an organization of more than 22,000employees. Bonder is also the national coordinator of competency-based management forService Canada. Bonder is a member of the Canadian Psychological Association and is apast chair of the federal government’s Inter-Departmental Committee on Competency-Based Management.

Carl-Denis Bouchard is director, human resource management accountability andreporting, in the Canada Public Service Agency (CPSA). The CPSA is a central agency with-in the Canadian public service which serves as the center of excellence for the manage-ment of people within the federal government. Previous to this, Bouchard was director ofresourcing for Service Canada, where he was responsible for the implementation of newhuman resource legislation related to staffing in the public service and for the develop-ment of strategies and policies related to staffing for the department. He is also, currently,a PhD student in the Industrial Relations Program of the Université du Québec enOutaouais.


Dr. Guy Bellemare is professor of industrial relations in the Industrial RelationsDepartment of the Université du Québec en Outaouais, Gatineau, Quebec. His researchfocuses on social innovations in the workplace with a particular emphasis on work car-ried out in public and semi-public sector organizations, social-economic enterprises andprivate sector service organizations. He also carries out research on self-employed work-ers and supervisory practices and is also working on revitalizing theories related to thefield of industrial relations.


Does Public ServiceMotivation Affect theOccupation Choices ofPublic Employees?By Leonard Bright, PhD

This study explored the connection between public service motivation (PSM) andthe occupation choices of employees in public organizations. Specifically, itfocused on understanding whether high levels of PSM lead individuals into publicservice occupations verse non-public service occupations in public organizations.Using a random sample of respondents from three public organizations in thestates of Indiana, Kentucky, and Oregon, this study found that PSM was not apredictor of the occupation choices of public employees. Individuals with highlevels of PSM were no more likely to be present in public service work than non-public service work. However, this study found that education level and genderwere the best determinates of the occupation choices of public employees. Theimplications of this study are discussed.

IntroductionScholars and practitioners are interested in public service motivation (PSM). Thisinterest is influenced by a desire to uncover the benefits that PSM has for publicorganizations. Public service motivation has the potential of transforming the wayemployees are recruited, motivated, and retained in public organizations. For example,one potential benefit of PSM is its connection to the occupation choices of publicemployees. Scholars have hypothesized that an ethic to serve the community and helpothers eventually drives individuals into public service careers.1 Yet, there are variousoccupations that exist in public organizations. Are public employees with high levels ofPSM more attracted to one public sector occupation than another? While there isevidence that public service motives are related to the career inclinations ofindividuals,2 there is a need to empirically prove that public service motives areconnected to their career choices. Demonstrating this connection will influence thestrategies that are used to manage and motivate employees in public organizations. Itwill also add to the growing body of research on the affects that PSM has on thepreferences and choices of employees.

Consequently, this study will explore the degree to which high levels of PSM is anexplanation for why public employees choose public service occupations over non-


public service occupations in public organizations. This goal will be accomplished inseveral stages. This article will begin by exploring the current state of the research onPSM. Following this review, this article will bring the question and hypothesis of thisstudy into focus. Next, the methods used to collect the data and test the hypothesis willbe reviewed. Afterwards, the findings of this study will be discussed in detail. Thisarticle will end by exploring the implications this study has for research and practice inpublic administration.

Literature Review

Public Service MotivationPublic service motivation (PSM) is arguably one the most important ideas introducedinto the field of public administration within the last two decades. Scholars havecharacterized PSM in many different ways, such as a service ethic, calling, or altruisticaims that motivate individuals to serve the public interest, help others, and be useful tosociety. Some have argued that people with high levels of these characteristics are goodemployees in public organizations. As a result, many scholars set out to betterunderstand the impact that PSM has in public organizations with the hope ofdeveloping better ways of recruiting, motivating, and retaining public employees.Unfortunately, early research on PSM was unable to meet these goals. This was largelybecause the field had not developed an acceptable way of operationalizing PSM. Inplace of a more valid approach, many scholars defined PSM with various indirectproxies, such as employment sector, job preferences, altruistic intentions, organizationcommitment, and job satisfaction.3 However, while public service motives may berelated to these factors, PSM should be treated as a separate and distinct concept.

One of the most important breakthroughs in PSM was the pioneering work ofPerry and Wise.4 These scholars provided a conceptually sound definition of PSM, aswell as a tool that measures it apart from other related concepts. For example, theydefined PSM as an individual’s predisposition to respond to motives that are groundeduniquely in public organizations and institutions. According to Perry and Wise,5

individuals with high levels of PSM are attracted to public service work for variousreasons, such as self-interest, an ethical stance, or emotional attachments. They furtherhypothesized that PSM is related to the work preferences, performance, and careerchoices of public employees. Building on this framework, Perry6 went on to developthe only known scale that measures PSM by rating individuals’ attraction to publicpolicy making, the public interest, compassion, and self-sacrifice.

The Outcomes of Public Service MotivationHow have scholars used these theoretical and methodological advances to further ourknowledge of the benefits and consequences of PSM in public organizations?Specifically, does PSM have important affects on the work preferences, performance,and career choices of employees? Fortunately, existing research provides some answersto these questions. For instance, research has consistently shown that PSM is


connected to the work preferences of public employees. Scholars have found thatpublic employees with high levels of PSM want non-monetary opportunities less thanmonetary opportunities.7 Second, Naff and Crum8 found that individuals with highlevels of PSM were more satisfied and less likely to leave public organizations whencompared with their counterparts with lower levels of PSM. More recently, scholarshave even found a connection between PSM and employees’ perception of red tape.Public service motivation seemingly increases the tolerance that public managers havefor procedural rules.9

Unfortunately, the literature does not offer clear answers about the relationshipbetween PSM and the performance of public employees. In fact, even though Frankand Lewis10 found that most government employees rate their work effort highly, theliterature differs on the connection that PSM has to these ratings. On one hand, Naffand Crum11 found that public employees with high levels of PSM reported receivinghigher performance ratings than their counterparts. However, in a more recent study,Alonso and Lewis12 found contradictory results using preexisting data from two largedata samples of federal employees. These scholars found that while PSM was positivelyrelated to performance appraisals in one sample of employees, no relationship wasfound in their second sample. Even more, Bright13 most recently found that PSM hadno direct relationship to performance outside of employees’ perceived fit to publicorganizations. At the very least, these studies prove that the connections between PSMand the performance of public employees are more elusive than hypothesized by Perryand Wise.14

Regretfully, little attention has been given to the connection between PSM and theoccupation choices of public employees. Yet, Brewer, Selden, and Facer15 suggestedthat a connection does exist. These scholars examined whether the motives of PSMvaried in different combinations from one individual to another. They found fourdistinct clusters of public employees with high levels of PSM, which they labeledsamaritans, communitarians, patriots, and humanitarians. While each of thesevariations of PSM had distinct motivational tendencies and needs, they all shared astrong interest in serving the larger community. This study’s finding suggests that PSM,regardless of its particular variation, would attract individuals to occupations involvedin public service, and away from occupations further removed from public service.Thus, there is a need for additional research to prove this connection.

Study Questions and HypothesisThis study seeks to understand the relationship that PSM has to the occupation choicesof individuals. Are individuals with high levels of PSM more likely to aim for certainkinds of occupations in public organizations? Specifically, are individuals with highlevels of PSM more likely to pursue public service occupations than non-public serviceoccupations? The answers to these questions depend on our understanding of thecharacteristics of public service occupations. While it is difficult to define public servicework with great precision, it is even more difficult to develop a definition that will bringabout total agreement among scholars. One of the reasons is because manycharacterize all occupations that exist in government organizations as public service


occupations. While this may be the case in general, there is a need to define publicservice occupations much more narrowly for this study.

For the purpose of this study, public service occupations are defined by theirlocation in organizations, contact with citizens, and core job tasks. That is, publicservice occupations are located in the operating core of organizations, placesemployees in direct contact with citizens (usually daily contact as an essential part oftheir work), and contain core job tasks that emphasis public service to the largercommunity, as opposed to performing support and technocratic roles as described byMintzberg.16 Using these characteristics, we are better able to separate occupationssuch as firefighters, police officers, juvenile counselors, and medical doctors fromoccupations such as management analysis, database technicians, and planners, all ofwhich are present in public organizations.

The occupation distinctions are not used to suggest that non-public serviceoccupations do not contribute to the larger goals of public organizations. To thecontrary, all occupations in public organizations contribute to the goals of publicorganizations. What the author is suggesting is that different occupations contribute tothe larger goals of public organizations in different ways. The question this study willattempt to answer is whether high levels of PSM attract employees to certain kinds ofoccupations over others? Existing theory suggest that PSM will lead individuals towardthe characteristics of public service occupation as defined above. For example,according to Perry and Wise,17 individuals with high levels of public service motives areattracted to job tasks that “provide opportunities to address questions of social equity,to express loyalty to country, to advocate a valued special interest, or to pursue socialprograms.” Consequently, one can hypothesize that individuals with high levels of PSMwill choose public service occupations at a greater rate than non-public serviceoccupations.

Study Hypothesis: Individuals with high levels of PSM will be present in publicservice occupation at greater levels than non-public service occupations.

Methodology

SubjectsIn 2006, five hundred and eight six potential subjects were randomly selected from acity government, county jurisdiction, and public health agency in the States of Indiana,Kentucky, and Oregon. These organizations were chosen for the purpose of creating adiverse sample of participants who represented a board range of public sectoroccupations and localities. A survey was mailed to the respondents’ office addresseswith a self-addressed return envelope to the author of this study. Thirty-five percent ofthe surveys were returned of which one-hundred and seventeen surveys (N=117)were useable for the purposes of this study.18 The respondents represented a diversemix of public occupations, some of which included medical doctors, buildinginspectors, community health workers, registered nurses, police officers, management


analysts, caseworkers, secretaries, social workers, district attorneys, librarians,maintenance workers, detectives, animal control workers, and engineers.

Public Service MotivationPublic service motivation was measured using Perry’s19 24-item PSM scale. Therespondents rated their agreement with Perry’s 24-item survey from 1 (highly disagree)to 7 (highly agree), by the categories of public policymaking, public interest,compassion, and self-sacrifice. The participants’ responses were summed into eachcategory. The categories were the observed indicators of PSM. The fit indices suggestthat this scale has a moderate level of validity (NFI = .637, CFI= .765, RMSEA=.069).

Occupation ChoicesThe occupation choices of the respondents were collected in several stages. First, therespondents were asked to list their job titles. Second, the job titles were examined todetermine their location in organizations, contact with citizens, and core job tasksusing job classification information. Job titles that were located in the operating core,placed employees in direct contact with citizens, and contained core job tasks thatemphasized public service were labeled “public service occupations.” All other jobtitles were labeled “non-public service occupations.” Third, the data was purged ofgeneric job titles and supervisors.20 As shown in Table 1, this strategy resulted in sixty-two (n= 62) respondents in non-public service occupations and fifty-five respondents(n=55) in public service occupations.

Control VariablesThis study used four control variables: age, race, gender, and education level. The ageof the participants was collected from the following open-ended survey question: Whatyear were you born? The year of birth was subsequently transformed by subtracting theyear of birth from the current year of this study. The race of the participants wascollected with the following multiple-choice question: How would you describe yourracial or ethnic group? The response categories included Black/African-American,Hispanic/Latino, White/Caucasian, Asian/Pacific Islander, Native-American, Eskimo orAleut, and None of the above. White/Caucasian participants were classified as non-minorities, while all other participants were classified as minorities. Minorityparticipants were coded as 0, while nonminority participants were coded as 1.

The gender of the participants was collected from the following multiple-choicesurvey question: What is your gender? The response category included male or female.Females were coded as 1 and males coded as 0. The education level of the participantswas collected from the following multiple-choice survey question: What is the highestlevel of education you have completed? The response category included: Did NotComplete High School, High School Diploma/GED, Some College, Associates Degree/Technique Certificate, Bachelors Degree, and Masters Degree or Higher. Theseresponses were coded from 1 (Did Not Complete High School) to 6 (Masters Degree orHigher).


Analysis ProcessThe data was analyzed in several stages. First, the data was reviewed to ensure that thedata was in good order. Curran, West, and Finch21 suggested that skewness rangesshould be less than two and kurtosis ranges less than seven. As shown in Table 2, all thevariables used in this study fell between these suggested ranges. Also, Table 3 displaysthe bivariate correlations among the study variables. Secondly, as shown in Figure 1, aconceptual confirmatory structural equation model was developed to test thehypothesis of this study using AMOS. This model tested PSM predicting the occupation


Table 1: Occupation Choices By Job Title

Non-public Service Occupations Public Service Occupations

Frequency Frequency

Accounting Assistant 4 Aging Services Coordinator 1

Accounts Payable 1 Animal Control Officer 1

Administrative Specialist 26 Appraiser 5

Billing Clerk 1 Case Manager 1

Billing Specialist 1 Correction Technician 1

Building Engineer 1 Counselor 2

Buyer 1 Police Detective 3

Cartographer 1 Deputy Police Officer 8

Computer Operator 1 Dietitian 1

Courier 1 District Attorney 3

Engineering Assistant 1 Environmental Resource Specialist 1

Engineering Project Manager 1 Health Specialist 1

Engineering Technician 1 Health Worker 1

Graphic Designer 1 Housing Service Specialist 2

Laboratory Technician 1 Building Inspector 3

Management Analyst 4 Juvenile Counselor 3

Management Specialist 1 Librarian 3

Medical Assistant 3 Mental Health Coordinator 4

Medical Records Clerk 1 Nurse 4

Medical Secretary 1 Nutrition Technician 1

Network Analyst 2 Occupancy Specialists 1

Planner 2 Patient Advocate 1

Property Manager 1 Program Educator 1

Receptionist 3 Right of Way Agent 1

Risk Services Manager 1 Social Worker 1

Total 62 Substance Abuse Worker 1

Total 55


Table 2: Description Of Study Variables

Variable Name Variable Label N Mean Skewness Kurtosis

Control Variables

Age AGE 114 42.73 .401 -.580

Gender GENDER 117 .68 -.758 -1.450

Minority Status RACE 117 .84 -1.855 1.464

Education Level EDU 117 4.20 -.093 -1.081

Public Service Motivation PSM

Public Policy Making POL 114 11.12 .336 .404

Public Interest INT 113 20.42 .180 -.580

Compassion COM 111 39.77 -.003 -.412

Self-Sacrifice SAC 112 39.88 -.100 -.627

Occupation Choices OCCUPATION CHOICE 117 .47 .121 -2.020

Table 3: Bivariate Correlations (Listwise N=181)

1 2 3 4 5 6

1. Age 1

2. Gender -.106 1

3. Minority Status **.204 *-.164 1

4. Education Level .038 *-.154 .089 1

5. PSM .065 .060 -.050 **.257 1

6. Job Type -.079 *-.149 .088 **.343 .088 1

** Correlation is significant at the 0.01 level (2-tailed).* Correlation is significant at the 0.05 level (2-tailed).

Figure 1: Conceptual Model

POL

INT

COM

SAC

e1

e2

e3

e4

OCCUPATION CHOICE e5PSM

AGE EDU GENDER RACE

choices of the respondents with the control variable of age, education level, gender,and minority status taken into account. Fit indices indicate that this model has a goodfit to the data (IFI=.946, CFI = .933, NFI= .722, RMSEA= .039). The regressioncoefficients are standardized maximum likelihood estimates. Statistical significance wasset at .05, two-tailed.

Study Findings

Description of the RespondentsAs shown in Table 4, a GLM multivariate model compared the mean responses for age,education level, gender, between the two occupation groups. The findings revealedthat the two groups of respondents were similar in age and ethnicity. For example, bothgroups of respondents mainly identified themselves a Caucasian and approximately


Table 4: Description of the Sample

Non-public Service Public ServicesOccupations Occupation

Frequency Percentage Frequency Percentage

Age M=43 M=42

Education Level M= 3.79* M=4.65*

1. Did Not Complete High School 1 1.6 0 0

2. High School Diploma/ GED 7 11.3 2 3.6

3. Some College 23 37.1 10 18.2

4. Associate Degree 13 21.0 9 16.4

5. Bachelors Degree 9 14.5 18 32.7

6. Masters Degree or Higher 9 14.5 16 29.1

Gender M=.82* M=.51*

0. Male 11 17.7 27 49.1

1. Female 51 82.3 28 50.9

Race/Ethnicity

1. African American/ Black 8 12.9 3 5.5

2. Hispanic/ Latino 2 3.2 1 1.8

3. White/Caucasian 50 80.6 48 87.3

4. Asian/ Pacific 2 3.2 2 3.6

5. Native American/ Eskimo/ Aleut 0 0 1 1.8

Note: Bold*= Significantly Different at .05M= Mean

forty-three years of age. When the ethnicity of the respondents was closely examined, itwas clear there were comparable levels of Blacks, Hispanics and Asian participants inboth occupation groups. However, two notable differences were the education leveland gender of the two occupation groups. For example, the respondents in publicservice occupations were significantly more educated than the respondents in non-public service occupations. Most of the respondents in the public service occupationearned at least a bachelors degree, while most of the respondents in non-public serviceoccupations earned only an associate degree. In addition, even though most of therespondents in both groups were female, there were significantly fewer malerespondents in non-public service occupations than in public service occupations.Forty-nine percent of the respondents in public service occupations were male, whileonly seventeen percent of the respondents in non-public service occupations weremale.

Hypothesis TestingAs well as comparing the similarities and differences between the respondents in non-public service and public service occupations, this study also explored the connectionthat PSM had to their occupation choices. It was hypothesized that a connection wouldexist between PSM and the occupation choices of the respondents. That is, individualswith high levels of PSM were expected to be present in public service occupations atgreater levels than in non-public service occupations. As shown in Figure 2 and Table 5,this hypothesis was strongly rejected by the data. Public service motivation was notrelated to the occupation choices of the respondents. Respondents with high levels ofPSM were equally likely to choose public service occupations and non-public serviceoccupations, particular when their gender and education levels were considered.Education level and gender emerged as the best determinates of the occupationchoices of the respondents. The respondents who were highly educated and malewere significantly more likely to work in public service occupations than therespondents who were less educated and female.


Figure 2: Statistical Model

POL

INT

COM

SAC

e1

e2

e3

e4

OCCUPATION CHOICE e5PSM

AGE EDU GENDER RACE

.04

-.04 .27 -.28 .06

Discussions of the FindingsThe findings of this study suggest that PSM does not determine the occupation choicesof public employees within public organizations. The respondents with high levels ofPSM were no more likely to work in public service occupations than non-public serviceoccupations, as defined in this study. What could possibility be the reasons why PSMdid not affect the occupation choices of the respondents? Gabris and Simo22 contendthat other more basic considerations drive the occupation choices of individuals, suchas job availability and material needs. This explanation is consistent with this study’sfindings for education level. The respondents in public service occupations weresignificantly more educated than the respondents in non-public service occupations.This could be a signal that many individuals with high levels of PSM lack theeducational qualifications needed for entry into public service occupations. Inaddition, education attainment may also cultivate an interest for occupations that arepublic service focused. This explanation is not unreasonable since many public affairsprograms actively seek to socialize students toward an ethic for public service. Publicorganizations may also play a role in equalizing public service motives across alloccupational groups through socialization practices.23

Similarly, the findings of this study also suggest that gender matters. The malerespondents in this study were significantly more likely to work in public serviceoccupations than non-public service occupations. This provides some support toexisting research that suggests that male employees are interested in different kinds ofpublic service work than female employees.24 However, these findings do not suggestthat female respondents are not interested in public service work. To the contrary, aspreviously discussed, female respondents outnumbered their male counterparts in bothoccupation groups. What the findings suggest is that female respondents were not asdiscriminating in their occupation choices as were their male counterparts. Females wereequally represented in both public service and non-public service occupations, whilemale employees were less represented in non-public service occupations.


Table 5: Standardized Maximum Likelihood Regression Weights byModel Paths

Model Paths B Estimate P

POL <--- PSM .503 1.000

INT <--- PSM .784 1.357 ***

COM <--- PSM .311 .825 .013

SAC <--- PSM .607 2.011 ***

JOBTYPE <--- RACE .061 .081 .473

JOBTYPE <--- GENDER -.283 -.295 ***

JOBTYPE <--- AGE -.037 -.002 .665

JOBTYPE <--- EDU .272 .102 .001

JOBTYPE <--- PSM .040 .011 .695

*** Significant at .000 Level

Conclusion of the StudyThis study sought to add to existing literature by exploring the extent to which PSM isan explanation for the occupation choices of non-managerial employees in publicorganizations, as implied by Perry and Wise.25 The findings suggest that PSM does notinfluence the occupation choices of public employees once they enter publicorganizations. Educational attainment and gender were significantly better predictorsof employees’ public service and non-public service occupational choices. However,more research is necessary to confirm this finding and to provide a clearer explanation,particularly in light of the weaknesses of this study. For example, one weakness of thisstudy is that it focused on non-managers. Focusing on non-managers may havereduced the chance of finding significant differences among the occupational choicesof public employees with high levels of PSM. This is important given existing researchthat has shown that managers have significantly greater levels of PSM than their non-managerial counterparts.

A second weakness of this study is that it did not take into account the practicesor environment of public organizations that either inhibit or socialize public serviceethics into employees. It is possible that the lack of a significant relationship foundbetween PSM and the occupation choices of the respondents was due to the fact mostnon-managers have comparability high or low levels of PSM because of the practices oftheir organization. Organizations may equalize the levels of PSM in their employeesovertime. This is supported by existing research that found a negative relationshipbetween PSM and years of experience.26 The longer employees work in publicorganization, the lower their levels of PSM.

A third weakness of this study is that it focused on the occupation choices ofemployees, rather than on their occupation attraction. It was assumed that publicemployees will choose occupations that are attractive. However, attraction and choicesare not perfectly related in organizations. It is possible for individuals to be attracted tooccupations, but do not choose to work in those occupations for a variety of reasons. Afuture study that investigates the degree to which PSM increases employees attractionto public service careers would be a better test of Perry and Wise’s27 hypothesis.

A final weakness of this study was the makeup of the respondent. Therespondents were drawn mainly from public organizations. Yet, should we assume thatpublic service occupations are confined to public organizations? There is evidence thatsuggests that public service occupations or interests in public service work are notconfined to public organizations.28 While it may be case that individuals are more likelyto find public service careers in public organizations,29 government reform andpartnership efforts have also spread public service work into nonprofit and businessorganizations.30 Therefore, adding respondents from nonprofit and businessorganizations could shine a stronger light on the connection between PSM and theoccupation choices of employees.

Moreover, this study concluded that public employees with high levels of PSM arepresent in both public service and non-public service occupations at relatively equallevels. While there are other potential reasons for this finding, it is also possible that


public employees with high levels of PSM are simply attracted to all occupations thatexist in public organizations. These employees may recognize the connection theiroccupations have to the public interest, regardless of where these occupations are inpublic organizations. This suggests that employees with high levels of PSM are mostattracted to organizations with public service missions.31 If this is the case, it wouldadvantageous for public managers to clearly communicate the connection that alloccupations have to the central public service goals of their organization. This strategywill be motivating to all employees, and especially those with strong motives to servethe community.

Notes1 Perry, J. (1996). Measuring public service motivation: An assessment of construct reliability andvalidity. Journal of Public Administration Research and Theory, 1, 5-22.

2 Brewer, G. A., Selden, S. C., & Facer, L. F. (2000). Individual conceptions of public servicemotivation. Public Administration Review, 60, 254-264.

3 Crewson, P. E. (1997). Public service motivation: Building empirical evidence of incidence andeffect. Journal of Public Administration Research and Theory, 4, 499-518, Gabris, G. T., & Simo,G. (1995). Public sector motivation as an independent variable affecting career decisions. PublicPersonnel Management, 24, 33-51.

4 Perry, J., & Wise, L. (1990). The motivational basis of public service. Public AdministrationReview, 50, 367-373.

5 Ibid.

6 Perry, J. (1996). Measuring public service motivation: An assessment of construct reliability andvalidity. Journal of Public Administration Research and Theory, 1, 5-22

7 Bright, L. (2007). Why do public employees desire intrinsic non-monetary workplaceopportunities? Public Personnel Management, 38, 15-37; Bright, L. (2005). Public employeeswith high levels of public service motivation: who are they, where are they, and what do theywant? Review of Public Personnel Administration, 25, 138-154; Brewer, G. A., Selden, S. C., &Facer, L. F. (2000). Individual conceptions of public service motivation. Public AdministrationReview, 60, 254-264.

8 Naff, K. C., & Crum, J. (1999). Working for America: Does public service motivation make adifference? Review of Public Personnel Administration, 19, 5-15.

9 Scott, P.G. & Pandey, S.K (2005). Red tape and public service motivation: Findings from anational survey of managers in state health and human services agencies. Review of PublicPersonnel Administration, 25, 155-180.

10 Frank, S., & Lewis, G. B., (2004). Government employees: Working or hardly working?American Review of Public Administration, 34, 36-51.

11 Naff, K. C., & Crum, J. (1999). Working for America: Does public service motivation make adifference? Review of Public Personnel Administration, 19, 5-15.

12 Alonso, P., & Lewis, G. B. (2001). Public service motivation and job performance: evidence fromthe federal sector. The American Review of Public Administration, 31, 363-380.

13 Bright L. (2007) Does Person-Organization Fit Mediate the Relationship Between Public ServiceMotivation and the Job Performance of Public Employees? Review of Public PersonnelAdministration, 27, 361-379.



15 Brewer, G. A., Selden, S. C., & Facer, L. F. (2000). Individual conceptions of public servicemotivation. Public Administration Review, 60, 254-264.

16 Mintzberg, H. (1993). Structure in fives: Designing effective organizations. Prentice Hall:Englewood Cliffs.


18 Two-hundred and five surveys were ultimately collected in 2006. However, 88 surveys wereexcluded from this study because they did not contain occupational information, containedgeneric job titles, or were obtained from managers.

19 Perry, J. (1996). Measuring public service motivation: An assessment of construct reliability andvalidity. Journal of Public Administration Research and Theory, 1, 5-22.

20 See Note 18.

21 Curran, P. J., West, S. G, & Finch, J. F. (1996). The robustness of test statistics to nonnormalityand specification error in confirmatory factor analysis. Psychological Methods, 1, 16-29.

22 Gabris, G. T., & Simo, G. (1995). Public sector motivation as an independent variable affectingcareer decisions. Public Personnel Management, 24, 33-51.

23 Moynihan, D.P, & Pandey, S.K. (2007). The role of organizations in fostering public servicemotivation. Public Administration Review, 67, 40-53.

24 Bright, L. (2005). Public employees with high levels of public service motivation: who are they,where are they, and what do they want? Review of Public Personnel Administration, 25, 138-154; Dehart-Davis, L., Marlowe, J., & Sanjay, P. K. (2007) Gender dimensions of public servicemotivation. Public Administration Review, 66, 873-887.


26 Moynihan, D.P, & Pandey, S.K. (2007). The role of organizations in fostering public servicemotivation. Public Administration Review, 67, 40-53.


28 Light, P. (2002). The content of character: The state of the nonprofit workforce. The NonprofitQuarterly, 9, 6-16; Light, P. (1999). The new public service. Brooking Institution Press:Washington, D.C.

29 Perry, J., & Porter, L.W. (1982). Factors affecting the context for motivation in publicorganizations. Academy of Management Review, 7, 89-98.

30 Light, P. (2002). The content of character: The state of the nonprofit workforce. The NonprofitQuarterly, 9, 6-16.

31 Scott, P.G. & Pandey, S.K (2005). Red tape and public service motivation: Findings from anational survey of managers in state health and human services agencies. Review of PublicPersonnel Administration, 25, 155-180; Weiss, J.A., & Piderit, S.K (1999). The value of missionstatements in public agencies: Journal of Public Administration Research and Theory, 9, 193-223.


AuthorLeonard Bright, PhD

University of LouisvilleSchool of Urban and Public Affairs426 West Bloom StreetLouisville, Kentucky 40208(502) [email protected]

Dr. Leonard Bright is an assistant professor at the University of Louisville. He receivedhis PhD from Portland State University in 2003. He specializes in management and leader-ship topics in public organizations. He is widely published on these topics in journalssuch as the American Review of Public Administration, Journal of Public Affairs Education,Public Personnel Management, and the Review of Public Personnel Administration.


Diversity andRepresentation in theU.S. FederalGovernment: Analysis ofthe Trends of FederalEmployment By Sungjoo Choi, PhD

This article aims to describe the trends in the current federal employment withregard to diversity and representation. To do so, it examines the diversity andrepresentation of federal employee populations by agency and by pay or gradelevel. While the federal government has been successful in diversifying itsemployee populations by employing women and minorities, their employment stillappears to be concentrated on the lower- and middle-level positions, resulting inlarge underrepresentation at the higher managerial and executive levels. Thisstudy, by analyzing the 2004 demographic data of federal employees by agencyand by grade level, evaluates the horizontal and vertical integration of women andminorities in the federal government. The findings show that while race or gendergroups in the federal government are equally represented across federal agencies,women and minorities are still underrepresented at the higher echelons. Thisarticle provides insight into improving the representation of women and minoritiesin the federal government.

Over the past several decades, increased diversity in the workforces of organ-izations has been highlighted as one of the most noticeable work-relatedtrends. Civil rights legislation and Affirmative Action have changed the demo-

graphic composition of the American workforce, leading to unprecedentedly highdiversity or heterogeneity within organizations. Especially in regard to public employ-ment, considerable efforts through Equal Employment Opportunity (EEO) and Affir-mative Action programs have been devoted to achieve social justice and bureaucraticor administrative accountability by providing the equal access to public jobs for peo-ple from diverse social groups and integrating them into the public workforce.1

Since the Civil Rights Act of 1964, the U.S. federal government has played aleading role in diversifying employee populations by significantly increasing theemployment of both women and minorities in federal agencies. The federal civilservices started programs to achieve equal employment opportunity objectives in the


1940s and have invested considerable resources and energy to correct problems ofunderrepresentation of women and minorities in federal workforces.2 As a result,public organizations tend to have more diverse employee populations than those inthe private sector, suggesting they have more successfully employed people withdifferent backgrounds, although often not promoting them to higher-level positions.3

Some relevant studies have shown that U.S. workers, particularly in the public sector,are becoming older and more diversified with respect to race and gender.4

A large literature has been devoted to assessing the representation of minoritiesand women in the public sector.5 However, scholars have concentrated only limitedattention toward developing analytic tools that measure the representativeness of thepublic workforces, or have focused on aggregate analysis of the federal government oronly the limited grade levels across all federal agencies. While some studies haveassessed diversity at the agency level6, they are already out of date and may beirrelevant to describe the current employee populations of federal agencies.

This article reports on the variation of the federal employment at the agency levelusing the archival data of the demographic profile of federal employees published bythe Office of Personnel Management (OPM) in 2004 and 2006. Scholars have beenconcerned about the neglect of diversity and representation issues regarding“politically appointed policy leaders — individuals with the authority, resources, andmeans to influence agency culture and decisions”7— in that a top position in ahierarchy and its significant influence over decision making would be a key element inrepresentative bureaucracy.8 In light of this, the present study examines the variation ofemployment across grade levels, including the higher-level positions and Senior Paylevels, focusing primarily on the race/ ethnicity and gender of employees. This researchwill contribute to our understanding of the diversity and representation trends of thecontemporary federal workforces.

The present study measures the variation of the federal employment in twoperspectives: diversity and representation. In the first section, theoretical backgroundsof these two perspectives will be discussed. In the second section, the data will bediscussed and the measures of diversity and representation that will be used inanalyzing the federal workforces will be introduced. In the third section, the findingswill be reported concerning the variation of employment by agency and by rank: thehorizontal diversity and the vertical representation. The fourth section will concentrateon assessing the representation of the top level of the federal bureaucracy, or theSenior Executive Services (SES). Finally, the results will be evaluated, and then adiscussion of the implications that the results provide will follow.

Diversity and Representation

Conceptual Differences of Diversity and Representation Two primary streams of the recent research on public organizational diversity arerepresentative bureaucracy and research on impacts of diversity. In the field of publicadministration, the issues concerning representation or representative bureaucracy9


have been more frequently discussed than the effects of diversity in the workforce.However, these two concepts have been used frequently without being clearlydistinguished.

Blau’s (1977) distinction of differentiation may provide some intuitive ideas thathelp us understand the conceptual differences of diversity and representation. Blau(1977) defines the two generic forms of differentiation, under which the variety of itsspecific forms can be subsumed: heterogeneity and inequity. Heterogeneity, orhorizontal differentiation, refers to the distribution of a population among groups interms of a nominal parameter.10 Inequality, or vertical differentiation, refers to thestatus distribution in terms of a graduated parameter.11 Blau (1977) conceptualizedinequality as “the average difference in status between any two pairs relative to theaverage status” and attempted to measure inequality by using the algebraicspecification of the Gini index.12 The operational criterion of the degree ofheterogeneity in a population is that two randomly chosen persons do not belong tothe same group. The larger the number of groups and the more evenly the populationis divided among them, the greater is the heterogeneity. On the other hand, inequalityis hard to define since much inequality in one sense may imply little inequality inanother. Paradoxically, much concentration of parameters such as power or wealthimplies widespread equality. Thus, the criterion of the degree of inequality is that thegreater the average status distance between all pairs of persons relative to their averagestatus, the greater the inequality.13

Nevertheless, the significant conceptual overlaps of representation and diversitycould be recognized. The present study assumes that the concept of diversity focuseson heterogeneity of group members in terms of parameters or attributes of people inthe society, while representation is more concerned about equality, that is, the equalrepresentation of each group mirroring the composition of the population of thesociety.

Representative Bureaucracy The rationale of representative bureaucracy is that when a public bureaucracy matchesthe general population on salient attributes of diversity, such as race/ ethnicity, orgender14, the bureaucracy will be more responsible for and more responsive to thepublic interest by better ensuring that all politically significant interests and values arerepresented in the formulation and implementation of policies and programs.15 If civilservants hold attitudes similar to those of the people they represent, their decisionswill resemble the decisions that those it represents would have made undercomparable circumstances.16

The representative bureaucracy study has moved from a normative study to focuson theoretical development in an empirical study. The issues of the empirical researchof representative bureaucracy include 1) the extent to which passive representationhas been adopted in governmental organizations and 2) the linkage between passiverepresentation and active representation. More recently, some scholars have attemptedto examine the impact of representation and diversity on organizational performance.As governmental organizations have been strongly encouraged to adopt representative


bureaucracy, the effective management of increased diversity in organizations hasbecome a critical concern in public administration. Relevant studies contended thatrepresentative bureaucracy would contribute to the efficiency and effectiveness ofpublic administration as well as to democratic accountability.17 The attempts to ensurethat the composition of bureaucracy reflects that of the population have increased thediversity within bureaucracies. Scholars including Foldy (2004) and Thomas and Ely(1996) suggested that diverse groups can perform better when they believe culturaldiversity can provide a good source of new ideas and work experiences. Pitts (2005), inhis study of racial integration and organizational performance of the Texas schooldistricts, found that the representation of teachers have a significant positive impact onthe performance of minority students in school, suggesting that ethnic representationcan positively influence performance outcomes.18

Diversity Effects While in the field of business management numerous studies have already beenconducted on the effects of diversity on an organization and employees, little scholarlyresearch in public administration has sought to understand the workforce diversity ofpublic organizations. The findings of previous research on diversity in groups suggestthat diversity offers a great opportunity for organizations as well as an enormouschallenge.19 Some studies argue that more heterogeneous workgroups tend toconsider a greater range of perspectives and to produce more high-quality solutionsthan do homogeneous groups.20 Others contend that heterogeneous groups are lesslikely to be integrated in work processes21, and the levels of dissatisfaction andturnover of heterogeneous groups are likely to be higher than those of homogeneousgroups.22 However, the results do not seem consistent about the direct effects ofdiversity on organizational outcomes, suggesting that there exists a more complexrelationship between diversity and organizational effectiveness than often realized.23

For example, recent research proposed that there exist mediating or moderatingeffects of team processes and moderating effects of contextual factors, such asorganizational culture, demographic characteristics of group members andsupervisors, and multidimensional diversity. Moreover, given that public organizationsoperate in distinctive political environments featuring higher levels of politicalinfluences from external sources caused by more complex relationships amongmultiple stakeholders or political actors, research on diversity in public organizationsmust be a great challenge to diversity researchers and theorists.

Data and Measurement

DataIn this study, I focus primarily on two demographic attributes to examine the variationof the diversity and representation of federal agencies: race/ ethnicity and gender.Concerning the question “what factors among social and demographic characteristicssignificantly influence an individual’s attitude and value?,” the literature has reached a


consensus that race and ethnicity are possibly the most important demographiccharacteristics for comparing bureaucratic and public representation in the UnitedStates.24 Scholars have also argued that gender is an important demographic variable tobe considered in the American bureaucratic setting.25 In fact, women and racialminorities in the United States faced great discrimination well into the twentiethcentury in their attempt to secure employment with the federal service.26 Muchscholarly effort has been concentrated on exploring the extent to which women andminorities who had previously been discriminated against and underrepresented havebeen integrated into bureaucracies.

To measure the diversity and representation of federal employment, I use the datadrawn from the Central Personnel Data File (CPDF) and the Demographic Profile of theFederal Workforce published by the Office of Personnel Management (OPM). Thedata27 come from the 291 federal agencies that include the 15 executive branchdepartments28 and independent agencies, representing one time period and collectedin 2004 and 2006.

Diversity Measures Diversity theorists and researchers have sought to refine measures of the diversity ofworkforces. Diversity researchers have employed various measures of diversity29, buthave not seemed to have reached an agreement. Recently, three kinds of diversitymeasures have been used most frequently: the coefficient of variation, the Blau indexof heterogeneity, and the entropy index of diversity. Diversity researchers use differenttypes of measures depending on the characteristics of the variables. Blau’s index ofheterogeneity and the entropy index are appropriate for categorical variables such asgender, race/ ethnicity, and functional and educational background. The coefficient ofvariation is used for continuous variables such as age and tenure. While recent researchhas frequently employed Blau’s index of heterogeneity and the coefficient ofvariation30, some researchers suggest that despite the disadvantage of not being easilyinterpretable, the entropy index of diversity or Shannon index (uncertainty index) isthe most appropriate measure of diversity in mathematical terms.31 The present studyemploys the entropy index of diversity to measure race and gender diversity. Theentropy index is computed with the following formula:

E = ∑ Qr logn (1/Qr) (r = 1, …, n), where Qr denotes the proportion of thepopulation of group r

The entropy index of diversity ranges from a minimum value of 0, which means that thepopulation contains a single group, and to a maximum value of 1, which means all ngroups are evenly represented in the population. The indices of each agency’sattributes of diversity – race and gender – are calculated through the entropy index ofdiversity. The present study classifies federal employees into five racial/ ethnic groups –American Indian or Alaskan Natives, Asian or Pacific Islanders, Blacks, Hispanics, andWhites – based on the CPDF’s categorization. The gender of the federal employeesincludes both male and female.


Representation MeasuresThe diversity indices are limited to measuring the horizontal representation, that is, theincidence of diverse groups in an organization. They fail to deal with representation inorganizations that are hierarchical in the distribution of status and influence amongtheir memberships because it in effect counts individual members as equallyrepresentative of their groups, regardless of their positions in the hierarchy.32 Ruccucciand Saidel (1997) argued that the commonly used measures of representation do notcapture the representativeness of an entire hierarchy of the bureaucratic system,especially neglecting the representation of the top levels possessing resources, theauthority, and influence on policy making. While in terms of total numbers of positionswomen and minorities appear to be well represented in public organizations, they arestill underrepresented in the upper levels such as the managerial and senior executivelevels and overrepresented in the lower echelons of bureaucracies.33 Researchersfound that in various levels of governments employment shares of women andracial/ethnic minorities are disproportionately higher at lower-level jobs.34

In this study, the vertical representation of groups in bureaucratic hierarchies willbe measured computing the penetration ratios of the groups. The penetration ratio,which is the most commonly used measure of representation, assesses the extent towhich a group is represented in an organization by computing its overall proportionalincidence.35 However, many researchers have indicated that the measure fails to reflecton representation in organizations that are hierarchical in the distribution of status andinfluence among their memberships. In this study, I measure the verticalrepresentation by computing penetration ratios by rank. A value of the penetrationratio of a group greater than a value of 1 suggests that the group is overrepresented inan organization. A value of the ratio less than a value of 1 indicates underrepresentationof a group in the organization.

Diversity and Representation in Federal Agencies

Horizontal Representation: Diversity in Federal Agencies Since the 1964 Civil Rights Act and 1965 Voting Rights Act were passed, scholars andpolicy makers have labored over how to define and achieve the fair representation ofsocially and politically significant groups within both elected and unelected institutionsof government.36 The federal government has been on the leading edge of diversity intheir workforces. Table 1 presents the trends of federal civilian employment by race andgender between 1994 and 2004. During this period, the proportion of Hispanicemployment in the federal government has increased by about 50 percent. Theproportion of aggregate minority employment has also increased by about 10 percent.

Diversity by agency. Table 2 reports the diversity measures of the 15 executivebranch departments and selected independent agencies with regard to race andgender. In terms of race diversity, among executive branch departments, theDepartment of Health and Human Services represents the highest level of diversitywith an entropy of 0.77 and the third highest level of minority employment (44.2%). On


the other hand, the Department of Transportation shows the lowest racial diversitywith an entropy of 0.49. It also reports the second lowest proportion of the minorityemployment (21.8%), following the Department of Agriculture with the lowestproportion (21.6 %) of minorities in its employee population. While the Department ofHousing and Urban Development is ranked as the highest in terms of the proportion ofthe minority employment (49.6%), its diversity measure is found to be lower than thatof the Department of Health and Human Services, suggesting that the employment ofthe department is concentrated on the minority employment to some extent.

Not surprisingly, the Equal Employment Opportunity Commission is found to bemost highly diversified in terms of race with an entropy of 0.7 among selectedindependent agencies and agencies under the Executive Office of President. It hiresminorities by the 61.4 percent of its total employment, corresponding to its mission ofcommitment to the diversity and representation of public workforces. The Office ofManagement and Budget reports the lowest level of race diversity (0.47) and theminority employment accounts for only the 23.8 percent of the agency’s workforce.

According to Table 2, the executive branch departments are highly successful inbalancing the employment of women and men. The Department of Labor achievesperfect gender diversity with an entropy of 1, which suggests the department employsthe same proportions of women and men in its workforce. The Department ofTransportation is ranked as the lowest with an entropy of 0.84 and female employeescomprise the 26.9 percent of the employee population. The Equal EmploymentOpportunity Commission and the Social Security Administration report higherproportions of female employees than male employees in their employee population,67.1 percent and 70.3 percent, respectively. The diversity indices of these agencies arerelatively lower than those of other agencies, suggesting that women areoverrepresented in these agencies.

Diversity by sub-agency. Table 3 illustrates top- or bottom-ranked sub-agencies interms of race and gender diversity. The most diversified sub-agency with regard to raceis the Office of Senior Coordinator for Pacific/ Hawaii (0.84) under the Department ofHousing and Urban Development. The proportion of minorities in this agency


Table 1: Employment Trends of the Federal Government (ExecutiveBranch Non-Postal)

1994 1996 1998 2000 2002 2004

Total 2,043,449 1,890,406 1,804,591 1,755,689 1,813,047 1,851,349

Female (%) 43.9 44.0 44.4 45.0 44.6 44.4

Minority (%) 28.4 29.1 29.6 30.2 30.7 31.4

African-American (%) 16.7 16.6 16.7 17.0 16.9 16.9

Hispanic (%) 5.7 6.1 6.4 6.6 6.9 7.3

Asian/Pacific Islanders (%) 4.1 4.3 4.5 4.5 4.8 5.0

American Indian/Alaska Natives (%) 2.0 2.0 2.1 2.2 2.2 2.1

SOURCE: The 2004 Demographic File of Federal Employees, U.S. Office of PersonnelManagement

comprises 56.4 percent: Blacks (24.2%), Hispanics (14.3%), Asian or Pacific Islanders(16.4%), American Indian or Alaskan Natives (1.5%), and Whites (43.6%). As can beseen, the memberships of this agency are more evenly distributed among these fiveracial groups than the other agencies. The least diversified agency is found to be theBureau of the Public Debt (0.21) under the Department of Treasury. The proportion ofminorities in this agency comprises only 7.4 percent and Whites occupy the 92.6percent of the employee population of this agency.

Interestingly, fourteen sub-level agencies and independent agencies achievedperfect heterogeneity in terms of gender, suggesting that the sub-level federal agencies


Table 2: Race and Gender Diversity by Agency (Executive BranchDepartments and Selected Independent Agencies)

Entropy Index Agency Race Gender Minority% Female %

Dept. of Agriculture .4862 .9851 21.6 42.8

Dept. of Commerce .5533 .9960 28.8 46.3

Dept. of Defense .5536 .9475 27.2 36.6

Dept. of the Army .5474 .9503 27.4 36.9

Dept. of the Navy .5619 .8833 28.1 30.2

Dept. of the Air Force .5056 .9086 23.4 32.4

Dept. of Labor .608 1.000 35.5 50.4

Dept. of Justice .5634 .9734 30.0 40.4

Dept. of Energy .503 .9535 22.8 37.4

Dept. of Education .6242 .9587 45.0 61.9

Dept. of Health and Human Services .7657 .9665 44.2 60.7

Dept. of Homeland Security .6468 .9229 36.3 33.8

Dept. of Housing and Urban Development .6734 .9696 49.6 60.2

Dept. of Interior .552 .9714 26.9 40.1

Dept. of State .52 .9996 26.0 48.8

Dept. of Transportation .4854 .8398 21.8 26.9

Dept. of the Treasury .6201 .9541 36.8 62.5

Dept. of Veteran Affairs .6491 .9799 38.2 58.3

Office of Management and Budget .4723 .9969 23.8 53.3

Equal Employment Opportunity Commission .7041 .9139 61.4 67.1

Merit Systems Protection Board .5611 .9871 34.1 56.7

National Aeronautics and Space Administration (NASA) .5039 .9322 23.3 34.8

Office of Personnel Management (OPM) .5455 .9540 36.3 62.6

Social Security Administration .6904 .8775 44.6 70.3

General Services Administration .6061 .9998 36.9 49.1

Small Business Administration .6497 .9976 39.4 52.9


Table 3: Top and Bottom Five Federal Agencies in Race and GenderDiversity

RACE DIVERSITYDiversity Index

Top Five Agencies (Sub-Agency Level) (Entropy) Minority (%)

HU-Office of Senior Coordinator for Pacific/Hawaii .8352 56.4

IN-Office of the Secretary of the interior .7421 44.4

HSAB-Bureau of Citizenship and Immigrant Services .7398 44.2

CM-Patent and Trademark Office .7324 55.0

TRAD-U.S. MINT .7182 46.1

Diversity Index Bottom Five Agencies (Sub-Agency Level) (Entropy) Minority (%)

HSBE-Federal Law Enforcement Training Center .3439 13.6

IN-Geological Survey .3195 11.8

IN-Indian Affairs .2937 88.0

DLMS-Mine Safety and Health Administration .2421 8.9

TRAB-Bureau of the Public Debt .2076 7.4

GENDER DIVERSITYDiversity Index

Top Fourteen Agencies (Sub-Agency Level) (Entropy) Female (%)

AG-Office of Budget and Program Analysis 1.0000 50.0

AGHS-Homeland Security Staff 1.0000 50.0

VABA-Immed. Office of Asst. Sec. for Human Res. and Adm. 1.0000 50.0

VADC-Deputy Assistant Secretary for Budget 1.0000 50.0

VAJB-Dep. Asst. Sec. Intergov. and Internat. Affairs 1.0000 50.0

VAJC-Dep. Asst. Sec. for Public Affairs 1.0000 50.0

GB-Overseas Private Investment Corporation 1.0000 50.0

SS-Selective Service System 1.0000 50.0

HE-Food and Drug Administration 1.0000 50.1

IN-Mineral Management Service 1.0000 49.9

FT-Federal Trade Commission 1.0000 50.1

TR-Departmental Offices 1.0000 49.8

EP-Environmental Protection Agency 1.0000 50.2

AM-Agency for International Development 1.0000 50.3

Diversity Index Bottom Five Agencies (Sub-Agency Level) (Entropy) Female (%)

VAPA-National Cemetery Administration .7454 21.2

GW-International Boundary and Water CMSN: US and MEX .7433 21.1

EDEQ-Office of Safe and Drug-Free Schools .6769 82.1

EDET-Office of English Language Acquisition .6409 83.7

DLWB-Women’s Bureau .4138 91.7

have been successful in obtaining the balance of the employment of men and women.Not surprisingly, the least diversified agency is found to be the Women’s Bureau (0.41),which has a higher level of female employment (91.7%) than of male employment.

Representation of women and minorities by agency. The entropy measuresfocus primarily on the even distribution of the five racial groups in the agency, nottaking into account the proportional incidence of the groups mirroring the wholeemployee population. To assess the proportional representation of women andminorities, I compute penetration ratios for the five racial groups and for women andmen in the executive branch departments. Figure 1 shows the proportionalrepresentation of the racial groups in the executive branch departments. Thepenetration ratios for Whites indicate that their proportional incidence is consistentwith that in the relevant population in all of the executive departments. Asian or PacificIslanders are also relatively well represented in most of these agencies with slightoverrepresentation in the Department of Commerce and the Department of Healthand Human Services. Blacks are overrepresented in the Department of Education(2.12) and the Department of Housing and Urban Development (2.2). Hispanics areoverrepresented in the Department of Homeland Security (2.28). On the other hand,the penetration ratios for American Indian or Alaskan Natives have greater variationthan those for other groups. Especially, they are highly overrepresented in theDepartment of Health and Human Services (7.91) and the Department of the Interior(7.14), but in most of the other agencies, are underrepresented.


Figure 1: Representation of Race/Ethnicity by Agency Figure 1. Representation of Race/Ethnicity by Agency

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

9.0

Pene

trat

ion

Rat

io (p

i)

Executive Branch Department

BlacksHispanicsAsian or Pacific IslanderAmerican Indian or Alaskan NativeWhites

As discussed earlier, most of the agencies are very successful in employingwomen and the proportion of female employees corresponds to that of maleemployees in the majority of the executive departments. Figure 2 reveals the variationof the balance between women and men in these agencies. While underrepresented inthe Department of Transportation (0.61), women are largely overrepresented in theDepartment of Treasury (1.41), the Department of Education (1.39), and theDepartment of Housing and Urban Development (1.36).

Vertical Representation: Representation by Rank As previously pointed out, the entropy index and the penetration ratio do not revealthe vertical representation of groups in hierarchical bureaucracies. Hence, to evaluateboth the vertical and horizontal diversity and representation, it is necessary toinvestigate the distribution of the positions occupied by women and minorities overthe various levels or positions in the hierarchy. In effect, Figure 3 shows that the ranksof positions are negatively related to the levels of race diversity and minorityemployment. Both of the race diversity and the proportion of the minorityemployment are consistently decreasing as the ranks of positions increase. In a similarline, Figure 4 exhibits the levels of gender diversity and the proportion of femaleemployees in federal agencies by rank. The variation of female employment is similar tothat of the minority employment in that women are concentrated in the lower-level


Figure 2: Representation of Gender by AgencyFigure 2. Representation of Gender by Agency

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6Pe

netr

atio

n R

atio

(pi)


Female Male


Figure 3: Race/Ethnicity Diversity by Pay Plan and Grade Figure 3. Race/Ethnicity Diversity by Pay Plan and Grade

0.46 0.45 0.44

0.43 0.44 0.41

0.38

0.42

0.33

0.29 0.30

0.27

0.23 0.20

0.17

0.14

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

0.50

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

Perc

ent (

%)

Ent

ropy

Inde

x

General Schedule (GS) and Senior Pay Levels

Diversity Index Minority (%)

Figure 4: Gender Diversity by Pay Plan and Grade Figure 4. Gender Diversity by Pay Plan and Grade

62.7 59.5

63.3 65.0 65.4 68.5

61.8 64.6

52.9 50.0

46.2 42.5

35.3 33.0

28.1 25.8

0

10

20

30

40

50

60

70

80

0.0

0.2

0.4

0.6

0.8

1.0

1.2Pe

rcen

t (%

)

Ent

ropy

Inde

x


Diversity Index Female (%)

positions, underrepresented at the higher and Senior Pay levels. The balances betweenthe women and men employment are accomplished at the levels between the GS-9 andGS-11.

Representation by race. The proportional representation of racial groups iscomputed by rank – 15General Schedule (GS) and Senior Pay levels. As for therepresentation of racial minority, the penetration ratios by pay plan and grade of threegroups – Blacks, Hispanics, and American Indians or Alaskan Natives – are higher thanunity at the lower- and middle-levels, suggesting that these groups are overrepresentedat these levels. American Indian or Alaskan Natives show the larger variation ofpenetration ratios over the ranks than the other groups. This group reaches thehighest level of penetration at the GS-4 (2.14) and the lowest at the GS-15 (0.38) andSenior Pay levels (0.38). Blacks and Hispanics demonstrate similar patterns ofpenetration over the ranks. Hispanic employees reach the highest penetration ratio atthe GS-2 (1.56) and Blacks, at the GS-8 (1.58). The penetration ratios of these threegroups decline with fluctuation and become lower than unity at the GS-10 and higher-level grades, resulting in the underrepresentation of these groups at the higher-levelpositions.

Contrary to the trends of Blacks, Hispanics, and American Indian or AlaskanNatives, the penetration ratios of Whites steadily increase as grade level goes up fromthe lowest echelons to the highest and Senior Pay levels, showing the smaller variation


Figure 5: Representation of Race/Ethnicity by Pay Plan and Grade Figure 5. Representation of Race/Ethnicity by Pay Plan and Grade

Pene

trat

ion

Rat

io (p

i)


BlacksHispanicsAsian or Pacific IslanderAmerican Indian or Alaskan NativeWhites

across grade level than the other groups. They are underrepresented at the positionsunder the GS-10 and overrepresented at the upper levels. Not surprisingly, Whites arelargely overrepresented at the Senior Pay levels and the other four groups areunderrepresented. At the Senior Pay levels, Whites accomplish the highest penetrationratio (1.25), while each of the other groups reports the lowest level of penetration. Therepresentation of Asian or Pacific Islanders by rank is more complicated to interpretthan that of the other groups. The group is slightly underrepresented at the majority ofthe levels under the GS-11, largely overrepresented at the GS-1 and GS-3 (1.82 and1.63, respectively). With the GS-12 as a turning point, this group achieves equalrepresentation at the higher-levels of positions and is overrepresented at the level ofthe GS-15 (1.26).

Representation by gender. Although federal employment in terms of genderdiversity demonstrates that the agencies achieved a reasonable level of genderdiversity, women still do not appear to be equally represented in all of the verticalpositions in the hierarchy of the federal government. Figure 6 explores therepresentation of female and male employees in federal agencies by rank. As can beseen, the level-specific penetration ratios of female and male employment have aninverse “mirror” relationship. This shows that a large gap between women’s and men’spenetration ratios exists over the ranks of federal employment, especially at the lowestand highest levels. Apparently, women are underrepresented in the upper positionshigher than over the GS-11 and overrepresented in the middle- and lower-levelpositions. This suggests that although, in terms of the number of positions, women areequally represented in federal agencies, they are concentrated in the levels under theGS-12, not promoted to the higher or Senior Pay levels.


Figure 6: Representation of Gender by Pay Plan and Grade Figure 6. Representation of Gender by Pay Plan and Grade

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Pene

trat

ion

Rat

io (p

i)


Female Male

Senior Executive Service The Senior Executive Service (SES), since it was created by the 1978 Civil Service RightsAct, has exercised considerable influence over governmental policy-making processes.Given that much literature argued that the representation of the SES contributes tobenefiting politically important groups in meaningful ways, the demographiccomposition of the SES has received great attention from scholars and political leaders.Traditionally, the positions of the SES have been dominated by White males, while theshare of minorities and women has increased consistently during the past severaldecades. Even so, minorities and women are still noticeably underrepresented at thetop levels of the federal bureaucracy. The data are drawn from the demographic file offederal employment in 2006.

Diversity and representation by race/ethnicity. As expected, Whites areconsistently overrepresented at the SES levels in the executive branch departments.The average penetration ratios of the groups are 0.81 for American Indians or AlaskanNatives, 0.45 for Asians or Pacific Islanders, 0.57 for Blacks, 0.49 for Hispanics, and 1.20for Whites. The Department of Housing and Urban Development achieved therelatively reasonable level of the representation of minority, compared with the otherdepartments. Most strikingly, the American Indian or Alaskan Native group shows theextraordinary variation in its proportional representation in the departments. This


Figure 7: Representation of Race/Ethnicity at the SES Level byAgency

Figure 7. Representation of Race/Ethnicity at the SES Level by Agency

0

1

2

3

4

5

6

7

Pene

trat

ion

Rat

io (p

i)


American Indian or Alaskan NativeAsian or Pacific IslanderBlacksHispanicsWhites

group is highly overrepresented in the Department of Health and Human Services andthe Department of the Interior, while it is commonly underrepresented in the othersor even not present in several departments, including the Department of Education,the Department of Labor, and the Department of State. The penetration ratios ofAmerican Indian or Alaskan Natives in the Department of the Interior and theDepartment of Health and Human Services are 5.97 and 2.55, respectively. Blacks areproperly represented in the Department of Education accomplishing a penetrationratio of 0.96, while overrepresented in the Department of Housing and UrbanDevelopment with a penetration ratio of 1.52. Hispanics are also well represented inthe Department of Housing and Urban Development with a penetration ratio of 0.95.

Diversity and representation by gender. Women are also commonlyunderrepresented at the SES levels. Especially, in the departments that administertraditionally male-dominated jobs such as the Department of Energy, the Departmentof Defense, the Department of Homeland Security, the Department of Justice, and theDepartment of Veterans Affairs, the gaps between the representation of women and ofmen is larger than in the other departments. The largest gap in proportional incidencebetween women and men is experienced by the Department of Defense (a penetrationratio of 0.47 for women and 1.43 for men). On the other hand, women and men arealmost equally represented in the Department of Health and Human Services (apenetration ratio of 0.97 for women and 1.02 for men). The Department of Education


Figure 8: Representation of Gender at the SES Level by Agency Figure 8. Representation of Gender at the SES Level by Agency

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Pene

trat

ion

Rat

io (p

i)


Female Male

and the Department of Housing and Urban Development also report relatively smalldifferences in penetration ratios between women and men.

Discussion and ConclusionWhile the federal government agencies achieved higher levels of diversity in theiremployee populations than did organizations in the private sector by employingwomen and minorities with different backgrounds, they do not appear to promotethem to higher positions. Still, women and minorities are underrepresented at theupper-level positions in hierarchical bureaucracies. In that the vertical ranks of publicbureaucracies are necessarily highly correlated to the authority and influence onimportant policy issues and decision-making processes37, the unequal representationof women and minorities at the upper-policy levels can possibly result in the unfairtreatment of people who have the memberships of these groups.

To assess not only the quantity but also the quality of diversity in federal agencies,this study measured the levels of diversity and representation of the federal workforcesin terms of race/ ethnicity and gender in two ways. By computing the heterogeneityindex (entropy index), I assessed how evenly the positions of federal agencies aredistributed to diverse racial or gender groups. Further, by using the proportionalincidence ratio (penetration ratio), I evaluated how properly these groups arerepresented across the vertical levels in the federal government.

The results reveal that federal agencies, especially the executive branchdepartments and large independent agencies (over 500 employees), are successful inaccomplishing the well-diversified employee populations with regard to race andgender although many small independent agencies are still poorly diversified. Most ofthe minority groups – Blacks, Hispanics, Asian or Pacific Islanders, and American Indianor Alaskan Natives – show reasonably good ratios of presence in federal agencies.

However, the findings concerning the representation of women and minoritiesacross grade level tell us a different story. The groups of minority employees – Blacks,Hispanics, Asian or Pacific Islanders, and American Indian or Alaskan Natives – showthe higher probability of presence at the lower-level positions, largelyunderrepresented at the higher-level positions and Senior Pay levels. The employmentof Whites shows the opposite trends to those of the minority groups. In addition,Whites tend to be distributed fairly uniformly not only across the organization but alsothroughout grade level, compared with the other groups. In a similar vein, therepresentation of women is also skewed toward the lower-level positions. Women arestill largely underrepresented at the higher-level ranks as well as the Senior Pay levels.

Some relevant literature suggests that the underrepresentation of women andminorities at the higher-level positions may be related to the social and economicbackgrounds that they might have, for example, educational backgrounds or trainingopportunities and selection processes.38 In addition, women may be disadvantaged bythe traditional stereotypic role differences between men and women. Future studieswill investigate what determines the successful integration of women and minorities inthe federal government. Such research will provide greater insight into how to improvethe diversity and representation in the federal government.


Notes1 Brewer, G. A. (2002). Public workforce integration: A field essay. Presented at the AnnualMeeting of the Southern Political Science Association in Savannah, Georgia; Kellough, E. J.(1990). Integration in the public workplace: Determinants of minority and female employmentin federal agencies. Public Administration Review, 50 (5), 557-566; Krislov, S. (1967). The Negroin Federal Employment: The Quest for Equal Opportunity. Minneapolis: University ofMinnesota Press; Krislov, S. & Rosenbloom, D. H. (1981). Representative Bureaucracy and theAmerican Political System. New York, NY: Praeger Publishers; Saltzstein, G. H. (1979).Representative bureaucracy and bureaucratic responsibility: Problems and prospects.Administration and Society, 10 (1): 464-475; Thompson, F. J. (1976). Minority groups in publicbureaucracies: Are passive and active representation linked? Administration and Society, 8 (3):201-226.

2 Kellough, E. J. (1990). Integration in the public workplace: Determinants of minority and femaleemployment in federal agencies. Public Administration Review, 50 (5): 557-566.

3 Cornwell, C. & Kellough, J. E. (1994). Women and minorities in federal agencies: Examiningnew evidence from panel data. Public Administration Review, 54 (3): 265-270; Foldy, E. G.(2004). Learning from diversity: A theoretical exploration. Public Administration Review, 64(5): 529-538; Riccucci, N. M. (2002). Managing Diversity in Public Sector Workforces. Boulder,Colorado: Westview Press.

4 Bond, J. T., Galinsky, E. & Swanberg, J. E. (1998). The 1997 National Study of ChangingWorkforce. New York: Families and Work Institute; Johnston, W. B. & Packer, A. H. (2000).Workforce 2000 – Work and Workers for the 21st Century. Washington, DC: U.S. Department ofLabor; Pitts, D. W. (2005). Diversity, representation, and performance: Evidence about race andethnicity in public organizations. Journal of Public Administration Research and Theory, 15(4): 615-631.

5 Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence: Bureaucraticrepresentation in the American states. Journal of Public Administration Research and Theory,11 (3): 379-402; Kellough, E. J. (1989). The 1978 civil service reform and federal equalemployment opportunity. The American Review of Public Administration, 19 (4): 313-324.

6 Kellough, E. J. (1990). Integration in the public workplace: Determinants of minority and femaleemployment in federal agencies. Public Administration Review, 50 (5): 557-566; Kranz, H.(1976). The Participatory Bureaucracy: Women and Minorities in a More RepresentativePublic Service. Lexington, MA: Lexington Books; Grabosky, P. N. & Rosenbloom, D. H. (1975).Racial and ethnic integration in the federal service. Social Science Quarterly, 56: 865-871;Nachmias, D. & Rosenbloom, D. H. (1973). Measuring bureaucratic representation andintegration. Public Administration Review, 33 (6): 590-597.

7 Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence: Bureaucraticrepresentation in the American states. Journal of Public Administration Research and Theory,11 (3): 379-402; Riccucci, N. M. & Saidel, J. R. (1997). The representativeness of state-levelbureaucratic leaders: A missing piece of the representative bureaucracy puzzle. PublicAdministration Review, 57 (5): 423-430.

8 Meier, K. J. (1993). Latinos and representative bureaucracy: Testing the Thompson andHenderson hypotheses. Journal of Public Administration Research and Theory, 3 (4): 357-375;Sigelman, L. & Karnig, A. K. (1976). Black representation in American state: Comparison ofbureaucracies and legislatures. American Politics Quarterly, 4: 237-246.

9 The underlying rationale of representative bureaucracy is to provide equal employmentopportunities for women and minorities who were historically discriminated against and tocomprise the population of employees mirroring the whole population of society.


10 Nominal parameters include sex, race, religion, ethnic affiliation, clan, occupation, place ofwork, place of residence, industry, marital status, political affiliation, national origin, andlanguage (Blau 1977).

11 Graduated parameters include education, income, wealth, prestige, power, socioeconomicorigin, age, administrative authority, and intelligence (Blau 1977).

12 Blau (1977) argued that inequality in a society should be defined as the average difference instatus rather than absolute status differences. Although the Gini index represents mean statusdistance relative to mean status, it is not apparent in the formula calculating it (Alker andRussett 1966). Blau (1977) and van de Geer created an alternative formula for the Gini index.

2∑sifi (pbi - pai)/ 2∑sipi

where si is mean status in a category, pi is the fraction of the population in that category, andPbi and Pai are the fractions of the population whose status is below and above that category,respectively. Thus, Pi + Pbi + Pai = 1.00 for every category. The sum is taken over all categories.The numerator is mean status distance, or absolute inequality, and the denominator is twicemean status. The ratio of the two is relative inequality (Blau 1997, 57-58).

13 Inequality must be distinguished from status diversity. Status diversity refers to the graduated-parameter equivalent of heterogeneity (Blau 1997, 10). While it is easy to distinguish perfectequality from a state of inequality, given two different, unequal distributions of some socialreward, it is not easy to decide which distribution is the more equal (Allison 1978). A socialstructure is delineated by its parameters, which are attributes of people in the society thatinfluence their role relations and thereby differentiate their social positions (Blau 1977). Twokinds of variables can be derived from these parameters. One variable refers to the attributes ofindividuals such as occupation, religion, income, and wealth. The other variable refers to newvariables the distributions of these attributes produce such as the social structuralcharacteristics in terms of the occupational distribution and the income distribution in society.The structural analysis focuses on these variables characterizing structural conditions and ontheir influences on social associations (Blau 1977). The structural analysis of various forms ofdifferentiation, interrelations, the conditions leading to them and changes in them, and theirimplications for social relations (Blau 1977) can substantially contribute to understanding ofhuman conduct and relations.

14 Meier, K. J. & Nigro, L. G. (1976). Representative bureaucracy and policy preferences: A study inthe attitudes of federal executives. Public Administration Review, 36 (4): 458-469; Mosher, F. C.(1968). Democracy and the Public Service. New York: Oxford University Press; Selden, S. C.(1997). The Promise of Representative Bureaucracy: Diversity and Responsiveness in aGovernment Agency. Armonk, NY: M.E. Sharpe.

15 Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence: Bureaucraticrepresentation in the American states. Journal of Public Administration Research and Theory,11 (3): 379-402; Meier, K. J. (1993). Latinos and representative bureaucracy: Testing theThompson and Henderson hypotheses. Journal of Public Administration Research andTheory, 3 (4): 357-375.

16 Selden, S. C. (1997). The Promise of Representative Bureaucracy: Diversity andResponsiveness in a Government Agency. Armonk, NY: M.E. Sharpe.

17 Ely, R. J. & Thomas, D. A. (2001). Cultural diversity at work: The effects of diversity perspectiveson work group processes and outcomes. Administrative Science Quarterly, 46: 229-273; Foldy,E. G. (2004). Learning from diversity: A theoretical exploration. Public Administration Review,64 (5): 529-538; Pitts, D. W. (2005). Diversity, representation, and performance: Evidence aboutrace and ethnicity in public organizations. Journal of Public Administration Research andTheory, 15 (4): 615-631; Thomas, D. A. & Ely, R. J. (1996). Making differences matter: A newparadigm for managing diversity. Harvard Business Review, 74 (5): 79-90.


18 Pitts, D. W. (2005). Diversity, representation, and performance: Evidence about race andethnicity in public organizations. Journal of Public Administration Research and Theory, 15(4): 615-631.

19 Milliken, F. J. & Martins, L. L. (1996). Searching for common threads: Understanding themultiple effects of diversity in organizational groups. Academy of Management Review, 21 (2):402-434; Shaw, M. (1981). Group Dynamics: The Psychology of Group Behavior. New York, NJ:McGraw-Hill; Tsui, A. S., Egan, T. D. & Xin, K. R. (1995). Diversity in organizations: Lessons fromdemography research. In Chemers, M., Okampo, S. & Constanzo, M. (Eds.), Diversity inOrganizations: New Perspectives from a Changing Workplace. Thousand Oaks, CA: SagePublications.

20 Cox, T. J., Lobel, S. A. & McLeod, P. L. (1991). Effects of ethnic and group cultural differences oncooperative and competitive behavior on a group task. Academy of Management Journal, 34(4): 827-847; Hoffman, R. & Maier, N. (1961). Quality and acceptance of problem solutions bymembers of homogeneous and heterogeneous groups. Journal of Abnormal and SocialPsychology, 62: 401-407; McLeod, P. L. & Lobel, S. (1992). The effects of ethnic diversity on ideageneration in small groups. Paper presented at the Annual Meeting of the Academy ofManagement, Las Vegas, Nevada; Ruhe. J. A. (1978). Effect of leader sex and leader behavior ongroup problem-solving. Proceedings of the American Institute for Decision Sciences, 123-127;Watson, W., Kumar, K. & Michaelsen, L. K. (1993). Cultural diversity’s impact on interactionprocess and performance: Comparing homogeneous and diverse task groups. Academy ofManagement Journal, 36 (3): 590-602.

21 O’Reilly, C. A., Caldwell, D. F. & Barnett, W. P. (1989). Work group demography, socialintegration, and turnover. Administrative Science Quarterly, 34: 21-37.

22 Jackson, S. E., Joshi, A. & Erhardt, N. L. (2003). Recent research on team and organizationaldiversity: SWOT Analysis and Implications. Journal of Management, 29 (6): 801-830; Wagner, G.W., Pfeffer, J. & O’Reilly, C. A. (1984). Organizational demography and turnover in top-management groups. Administrative Science Quarterly, 29: 74-92.

23 Williams, K. Y. & O’Reilly III, C. A. (1998). Demography and diversity in organizations: A reviewof 40 years of research. Research in Organizational Behavior, 20: 77-140; Wise, L. R. &Tschirhart, M. (2000). Examining empirical evidence on diversity effects: How useful is diversityresearch for public-sector managers? Public Administration Review, 60 (5): 386-394.

24 E.g., Cayer, J. N. & Sigelman, L. (1980). Minorities and women in state and local government:1973-1975. Public Administration Review, 40 (5): 443-450; Dye, T. & Renick, J. (1981). Politicalpower and city jobs: Determinants of minority employment. Social Science Quarterly, 62 (3):475-486; Herbert, A. W. (1974). The minority administrator: Problems, prospects, andchallenges. Public Administration Review, 34 (6): 556-563; Krislov, S. (1974). RepresentativeBureaucracy. Englewood Cliffs, NJ: Prentice-Hall, Inc.; Meier, K. J. (1975). Representativebureaucracy: An empirical analysis. American Political Science Review, 69: 526-542; Meier, K. J.(1993). Latinos and representative bureaucracy: Testing the Thompson and Hendersonhypotheses. Journal of Public Administration Research and Theory, 3 (4): 357-375;Rosenbloom, D. H. & Featherstonhaugh, J. G. (1977). Passive and active representation in thefederal service: A comparison of blacks and whites. Social Science Quarterly, 57 (1): 873-882;Rosenbloom, D. H. & Kinnard, D. (1977). Bureaucratic representation and bureaucraticbehavior: An exploratory analysis. Midwest Review of Public Administration, 11 (1): 35-42;Smith, R. (1980). Representative bureaucracy: A research note on demographic representationin state bureaucracies. Review of Public Personnel Administration, 1 (3): 1-14; Thompson, F. J.(1978). Civil servants and the deprived: Socio-political and occupational explanations ofattitudes toward minority. American Journal of Political Science, 22 (2): 325-347.

25 E.g., Daley, D. (1984). Political and occupational barriers to the implementation of affirmativeaction: Administrative, executive, and legislative attitudes toward representative bureaucracy.Review of Public Personnel Administration, 4: 4-15; Davis, C. & West, J. (1985). Implementingpublic programs: Equal opportunity, affirmative action, and administrative policy options.


Review of Public Personnel Administration, 4 (2): 16-30; Dometrius, N. & Sigelman, L. (1984).Assessing program toward affirmative action goals in state and local governments: A newbenchmark. Public Administration Review, 44 (3): 241-246; Dye, T. & Renick, J. (1981). Politicalpower and city jobs: Determinants of minority employment. Social Science Quarterly, 62 (3):475-486; Hale, M. M. & Kelly, R. M. (1989). Gender, bureaucracies, and public sector careers. InGender, Bureaucracy, and Democracy, edited by Hale, Mary M. & Kelly, Rita Mae, New York,NY: Greenwood Press; Krislov, S. (1974). Representative Bureaucracy. Englewood Cliffs, NJ:Prentice-Hall, Inc.; Meier, K. J. (1975). Representative bureaucracy: An empirical analysis.American Political Science Review, 69: 526-542; Meier, K. J. (1993). Latinos and representativebureaucracy: Testing the Thompson and Henderson hypotheses. Journal of PublicAdministration Research and Theory, 3 (4): 357-375; Selden, S. C. (1997). The Promise ofRepresentative Bureaucracy: Diversity and Responsiveness in a Government Agency. Armonk,NY: M.E. Sharpe.

26 Aron, C. S. (1987). Ladies and Gentleman of the Civil Service. New York, NY: Oxford UniversityPress; Rosenbloom, D. H. (1977). Federal Equal Employment Opportunity. New York, NY:Praeger Publishers; Shafritz, J. M., Riccucci, N. M., Rosenbloom, D. H. & Hyde, A. C. (1992).Personnel Management in Government: Politics and Processes. 4th ed., New York, NY: MarcelDekker.

27 The sample includes 291 federal agencies including sub-agencies of 15 executive departments –Dept. of Agriculture, Dept. of Commerce, Dept. of Defense, Dept. of Education, Dept. ofEnergy, Dept. of Health and Human Services, Dept. of Homeland Security, Dept. of Housingand Urban Development, Dept. of the Interior, Dept. of Justice, Dept. of Labor, Dept. of State,Dept. of Transportation, Dept. of the Treasury, and Dept. of Veterans Affairs and small size(under 100 employees), medium size (100-999 employees) and large size (over 1000employees) independent agencies.

28 The executive branch includes all agencies except the U.S. Postal Service, Postal RateCommission, Central Intelligence Agency, National Security Agency, Tennessee Valley Authority,White House Office, Office of the Vice President, Board of Governors of the Federal ReserveSystem, National Imagery and Mapping Agency, and Defense Intelligence Agency.

29 The absolute diversity index (AbD) and the relative diversity index (RelD) (e.g., Tam and BassettJr. 2004), Blau index of variability (e.g., Ely 2004; Harrison, Price, and Bell 1998; Jackson andJoshi 2004; Mohammed and Angell 2004; Pitts 2005; Smith et al. 1994; Wiersema and Bantel1992), the coefficient of variation (e.g., Ely 2004; Harrison, Price, and Bell 1998; Jackson andJoshi 2004; Jehn and Bezrukova 2004; Wiersema and Bantel 1992), the entropy index ofdiversity (e.g., Ancona and Caldwell 1992; Jehn and Bezrukova 2004; Jehn, Northcraft, andNeale 1999; Krishnan and Park 2004; Pelled, Eisenhardt, and Xin 1999; Teachman 1980;Wiersema and Bantel 1992), Gini coefficient (e.g., Meier 1975), Hirshman-Herfindahle index,Lieberson index (e.g., Guajardo 1999), McIntosh D Evenness index, or Nachmias-RosenbloomMeasure of Variation (MV) index, the relational demography score (e.g., Tsui, Egan, and O’ReillyIII 1992, 2001), and the Teachman’s index (e.g., Keller 2001; Pelled, Eisenhardt, and Xin 1999;Teachman 1980).

30 e.g., Bantel, K. A. & Jackson, S. E. (1989). Top management and innovations in banking: Doesthe composition of top team make a difference? Strategic Management Journal, 10: 107-124;Jackson, S. E., Brett, J. F., Sessa, V. I., Julin, J. A. & Peyronnin, K. (1991). Some differences makea difference: Individual dissimilarity and group heterogeneity as correlates of recruitment,promotions, and turnover. Journal of Applied Psychology, 75 (5): 675-689.

31 Miller, V. P. & Quigley, J. M. (1990). Segregation by racial and demographic group: Evidence fromthe San Francisco bay area. Urban Studies, 27:3-21; Pielou, E. C. (1977). Mathematical Ecology.New York: John Wiley and Sons; Reardon, S. F. (1998). Methods of measuring diversity andsegregation in multi-group populations: with examples using racial school enrollment data.Unpublished draft, supported by the Spencer Foundation Small Grants Program; Theil, H.,


1972, Statistical Decomposition Analysis. Amsterdam: North-Holland Publishing Company;Zoloth, B. S. (1976). Alternative measures of school segregation. Land Economics, 52: 278-298.

32 Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence: Bureaucraticrepresentation in the American states. Journal of Public Administration Research and Theory,11 (3): 379-402.

33 Cornwell, C. & Kellough, J. E. (1994). Women and minorities in federal agencies: Examiningnew evidence from panel data. Public Administration Review, 54 (3): 265-270; Foldy, E. G.(2004). Learning from diversity: A theoretical exploration. Public Administration Review, 64(5): 529-538; Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence:Bureaucratic representation in the American states. Journal of Public Administration Researchand Theory, 11 (3): 379-402; Riccucci, N. M. (2002). Managing Diversity in Public SectorWorkforces. Boulder, Colorado: Westview Press.

34 E.g., Cornwell, C. & Kellough, J. E. (1994). Women and minorities in federal agencies:Examining new evidence from panel data. Public Administration Review, 54 (3): 265-270;Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence: Bureaucraticrepresentation in the American states. Journal of Public Administration Research and Theory,11 (3): 379-402; Sigelman, L. (1976). The curious case of women in state and local government.Social Science Quarterly, 57 (1): 591-604.

35 Sigelman, L. & Karnig, A. K. (1976). Black representation in American state: Comparison ofbureaucracies and legislatures. American Politics Quarterly, 4: 237-246.

36 Greene, V., Selden, S. C. & Brewer, G. (2000). Measuring power and presence: Bureaucraticrepresentation in the American states. Journal of Public Administration Research and Theory,11 (3): 379-402.

37 Baldwin, J. N. (1996). Female promotions in the male-dominant organizations: The case of theUnited States military. Journal of Politics, 58: 1184-1197; Greene, V., Selden, S. C. & Brewer, G.(2000). Measuring power and presence: Bureaucratic representation in the American states.Journal of Public Administration Research and Theory, 11 (3): 379-402.

38 Baldwin, J. N. (1996). Female promotions in the male-dominant organizations: The case of theUnited States military. Journal of Politics, 58: 1184-1197.

AuthorSungjoo Choi, PhD

Assistant Professor Kennesaw State University Department of Political Science and International Affairs1000 Chastain Rd. Kennesaw, GA 30144Phone: (770) [email protected]

Dr. Sungjoo Choi is an assistant professor of public administration in the department ofpolitical science and international affairs at Kennesaw State University. Her research inter-ests include diversity management, representative bureaucracy, and gender pay gap.


Sexual HarassmentTraining: Person versusMachineBy Melissa K. Preusser, Lynn K. Bartels, PhD, and Cynthia R. Nordstrom, PhD

Computer-based training is rapidly being implemented in organizations. Manyadvantages of this approach have been suggested (e.g., self-paced instruction,uniform content, and long-run cost efficiency) along with several potentialdisadvantages (e.g., noncompletion rates, computer availability). Despite thepurported advantages of computer-based training, little research has examined theeffect of computer-based training on learning. This field study systematicallycompared computer-based sexual harassment training with traditional instructor-led training. Overall, the results suggested that having a training course on sexualharassment, regardless of method of training, increased learning among theuniversity employees who participated in these sessions. Reactions to both typesof training were positive.

The use of technology in the workplace has increased drastically over the lasttwo decades. This is especially true in the area of training. According to a reportby the American Society of Training and Development (ASTD), organizations

delivered about 10.5 percent of their training through technology in 2001 up from 8.8percent in 2000.1 The use of technology in training has many advantages includingdecreased costs, self-paced instruction, and uniform content.2 While the usefulness ofcomputer-based training is readily apparent when one is training computer softwareprograms such as Excel or Windows, it is less obvious whether this type of training iseffective for soft-skills training (such as interpersonal skills or sexual harassment). Thecurrent research is designed to address this issue examining whether computer-basedsexual harassment training is effective in promoting trainee satisfaction and learningcompared to the more traditional instructor-led training.

Focusing on how to most effectively deliver sexual harassment training content isimportant due to the many organizational and personal consequences associated withsexual harassment. Sexual harassment has been linked with increased job stress, loss ofwork commitment, decreased satisfaction, loss of customers, absenteeism, turnover,transfer requests, damage to the company’s image, etc.3 Monetarily, organizations alsofeel the effects of sexual harassment when a case is litigated. There are approximately60 cases of sexual harassment filed per workday around the country and the averagepayout per case is around $38,500.4 Since the 1991 amendments to Title VII, whichallow for jury trials in sexual harassment cases, the awards have been as high as $34million for damages as was the case for Mitsubishi Motor Manufacturing.5


Given the ramifications associated with sexual harassment, organizations havebeen scrambling to ensure that employees are fully briefed as to what constitutessexual harassment and what actions they should take as either victims of the behavioror as supervisors who learn of potentially illegal conduct.6 Often the first placeemployers turn is the Equal Employment Opportunity Commission Guidelines (1980)where the legal definition of sexual harassment is presented:

Unwelcome sexual advances, requests for sexual favors, and other verbal orphysical conduct of a sexual nature constitute sexual harassment when:submission to or rejection of this conduct explicitly or implicitly affects anindividual’s employment, unreasonably interferes with an individual’s workperformance or creates an intimidating, hostile or offensive work environment(p.203).7

Unfortunately, the definition is fairly vague allowing for different interpretationsacross individuals.8 What may be perceived as innocent flirting by one employee maybe construed as unwelcome harassment by another.9 Consistent with this, humanresource professionals report that the majority of employees are uncertain as to whatbehaviors legally constitute sexual harassment.10 In an effort to bridge this knowledgegap, most organizations now have explicit written policies pertaining to sexualharassment.11 But policies alone are not sufficient to curb sexual harassment. Instead,policies complemented by sexual harassment training are cited as a more effectiveapproach to decreasing sexual harassment and the organization’s legal liability.12

Several states have specifically put forth legislation that either requires or stronglyencourages employers to provide harassment prevention training including California,Connecticut, Colorado, Illinois, Maine, Massachusetts, Tennessee, Texas, Utah, andVermont.13 Training may include some or all of the following components: definitionsand types of sexual harassment; company policies, and video or role-play exercises.

If prevailing laws are not enough to encourage employers to provide sexualharassment training, the potential return on investment for providing such trainingmay be. Employers have reaped financial benefits associated with sexual harassmenttraining even after taking into account training program costs.14 For example, ahealthcare company after implementing sexual harassment prevention training foundthat the number of complaints fell 36% within one year and the number of claims thatwent into litigation dropped by 41% saving them an estimated $800,000 in legal costs.The company also found that employees in their exit interviews were far less likely topoint to sexual harassment issues as their reason for quitting. Employer insurancecarriers also see the potential return on investment. Employment Practices LiabilityInsurance, for example, often provides a premium discount to employers who providesexual harassment prevention training.15

Although most organizations now see the need and utility of providing sexualharassment training, one issue that has not been given much consideration is the bestmode for disseminating this information. Specifically, how should the training bedelivered and what effect does delivery mode have on attitudes, learning and results?


Two types of delivery will be discussed further, computer-based training and instructor-led training.

Computer-Based TrainingAs noted earlier, computer-based training is on the rise. Despite this, there is a relativedearth of research examining the effectiveness of this mode of delivering trainingcontent. There have been studies which point to some of the advantages associatedwith computer-based training. For instance, a study of the effectiveness of computerskills training (Microsoft Word and Excel) identifies that both the end-of training andone-month follow-up performance evaluations were significantly better for employeeswho took the computer-based training, but this training was on computer software.16

Another study pointed out that computer-based training lessens the cognitive loadassociated with listening, watching, and incorporating training materials and allowstrainees to work at their own pace.17 Other advantages related to using computer-basedtraining include self-paced instruction, just-in-time learning, uniform content, andlong-run cost efficiency.18 In addition, organizations may see up to a forty percentdecrease in training time and costs using e-learning.19 Yet, there are also disadvantagesassociated with computer-based training that need to be considered.

The disadvantages of computer-based training include older employee resistance,employees’ fear of new learning environments, poor marketing of the new type oftraining to employees, training time restrictions, noncompletion rates, seating, and/orcomputer availability.20 Another potential drawback is the potentially high cost ofdeveloping the system.21 The current problems facing organizations with the “transferof training” to the job may be compounded in the future with computer-based trainingsince employees will have to be motivated enough to complete the entire trainingcourse by themselves and ensure they are learning the material.22 Employees need self-motivation and self-efficacy to benefit from this type of training when the training is notmade mandatory. Given some of these problems, organizations may opt to delivertraining in the more traditional manner – with an instructor conducting a live trainingsession.

Instructor-led TrainingTraditionally, companies have relied on instructor-led training using PowerPoint, stand-up lecture, and hands-on experience to train employees on various competencies andskills. Because of its increased familiarity, employees may prefer instructor-basedcompared to computer-based training.23 Interestingly, even though, in general, lectureused in training has negative connotations (e.g., boring, passive learning, etc.); a recentmeta-analysis has shown that it is actually an effective training method in theworkplace. Specifically, instructor-led training (including lecture, audiovisual materials,and discussion) produced significant learning results for cognitive, interpersonal, andpsychomotor skills whereas computer-based training was non-significant for all ofthese learning outcomes.24


Just as is the case with computer-based training, instructor-led training has severaladvantages associated with it. For instance, live training allows employees to leave theworkplace to learn new skills or competencies.25 Additionally, instructor-led trainingprovides employees with social pressures not only to attend training but to learn thematerial. The training also is easily adapted to the different learning styles of trainees,and there is increased availability of trained professionals to respond to trainees’questions.26 The disadvantages of instructor-led training can include limited seatingand time availability, learning at the pace of others, and inconsistency in materialcovered.27 There also may be additional costs associated with live training.

Training EvaluationRegardless of which type of training method is utilized, its effectiveness has to beevaluated. The evaluation process also can provide answers as to which type of trainingmethod is more effective. Unfortunately, in general, evaluation of training is often notperformed and this is particularly true in the case of sexual harassment training.28 In2001, there were only nine studies that had thoroughly evaluated a sexual harassmenttraining program.29 Many employers harbor the mistaken belief that any kind of sexualharassment training (regardless of length or quality of the program) will be enough tooffer legal protections.30 In truth, however, there have been several court cases inwhich companies “cut corners” and either offered very minimal training (10 minutevideo with handouts) or did not ensure that employees understood theirresponsibilities when it came to sexual harassment and lost a subsequent court case.31

In other words, it is not enough for employers to “check the box” when it comes tosexual harassment training. They need to be able to show that the training is bothsubstantive and effective.32

For employers interested in mounting an evaluation of training, it is oftenKirkpatrick’s (1959) model that they turn to for guidance.33 Kirkpatrick establishedlevels of training criteria to guide trainers in examining the effectiveness of trainingwhich include: reactions, learning, behavior, and results.34

Reactions are the trainees’ feelings about the training. In terms of sexualharassment training, the reaction evaluation would simply identify if trainees thoughtthat the training/trainer was effective and if they liked the training. Learning assesseshow well trainees learned facts or information. To evaluate learning in sexualharassment training, an evaluation could have trainees define different types of sexualharassment, provide multiple choice questions regarding victims’ rights/responsi-bilities, or ask true/false questions on proper reporting procedures in sexualharassment cases. The behavioral dimension evaluates the extent to which trainingcontent is used on the job. It is more difficult to evaluate and generally must be doneby observation.35 For example, observing that employees are sharing fewer sexuallycharged jokes may be one sign of behavioral change. The final criterion, results,assesses whether desired organizational outcomes are achieved. Results from sexualharassment training could be measured by identifying a decrease in the number ofsexual harassment lawsuits or a decrease in turnover due to sexually problematicworkplace behavior.


Although Kirkpatrick’s model remains perhaps the most popular schematic fortraining evaluation, others have expanded on the model. The learning criterion can beexpanded by focusing on three types of learning outcomes: cognitive, skill-based, andaffective.36 Sexual harassment training hopes to achieve all of these learning outcomes.Following training, it would be ideal if trainees gained knowledge about sexualharassment (cognitive), could apply the knowledge automatically to new scenarios(skill-based) and modified their attitudes about sexual harassment in a more positivedirection (affective).

The current exploratory study sought to answer a number of questions. First,given the prevalence of sexual harassment training and the increased use oftechnology-based training, can soft skills such as sexual harassment be effectivelytrained using a computer-based mode of delivery? Second, which type of delivery(computer-based vs. instructor-led) leads to better outcomes? The followinghypotheses were developed based upon the fairly limited research in this arena.

In terms of trainees’ reactions, research has suggested that employees continueto prefer instructor-led training.37 Accordingly, trainee reactions to the instructor-ledsexual harassment training should be more positive than reactions to computer-basedtraining.

Hypothesis One: Trainee reactions following the training will be more positiveconcerning the instructor-led training compared to the computer-based training.

With regard to the cognitive (knowledge-based) component of the learningdimension, when comparing training methods, A recent meta-analysis noted that forcognitive skills, a combination of lecture, audiovisual, and discussion was an effectivemethod to increase learning. Computer-assisted instruction was not a significantlymore effective method for cognitive skills.38

Hypothesis Two: The cognitive learning outcomes for the instructor-led trainingwill be greater than the cognitive outcomes for the computer-based training.

In terms of the skill-based component of the learning dimension, a meta-analysisfound that for interpersonal skills or tasks (skill-based), lecture, audiovisual, anddiscussion are the most effective methods. Computer-based training showed nosignificant effect in interpersonal skills or tasks.39

Hypothesis Three: The skill-based learning outcomes will indicate that instructor-led participants are better able to apply skills than those in the computer-basedtraining.40

With the last component of the learning dimension (affect/attitudes), previousresearch identified that lecture and video are effective methods in changing attitudesfollowing training while computer-based training, in general, is not a method of choicefor changing trainee attitudes.


Hypothesis Four: The instructor-led training participants will have more positivechanges in attitude (affect) than those in the computer-based training.

Method

ParticipantsThe sample consisted of 70 employees, 34 males and 36 females, from a medium-sized,midwestern public university. The most frequent age-range of participants in the studywas 34-41 years and over 50% of participants had 16 or more years of work experience.When reporting the highest level of education completed, 6% completed high school,34% noted completing some college, 10% completed an Associate’s Degree, 24%obtained a Bachelor’s Degree, 21% achieved a Masters Degree, 1% had a DoctorateDegree and 3% noted “Other”. Eighty percent of the participants had previouslycompleted a computer training course or class. The participants were invited tovoluntarily participate through the Human Resource Department.

Computer-based Training Program

The computer-based training was acquired through Costal Training TechnologiesCorporation and is titled: “Sexual Harassment: You Make the Call”. The CD-ROMversion of this training can be purchased through Coastal Training TechnologiesCorporation.41 This computer-based sexual harassment training session takesapproximately 1.5-2 hours to complete. The program consists of five sections,introduction to the course, sexual harassment defined, points of view, handling sexualharassment, and a comprehensive test. Scenarios are given in each section and therealso are approximately two quiz questions asked in each sub-section. The participant isable to take notes by selecting the notes tab. The computer-based session does notallow for participant questions. The computer-based sessions were completed byparticipants at their desks, individually, in their offices. Participants in the computer-based training did not complete the comprehensive test at the end of the computerprogram. Each participant was timed to determine the average time it took to completeeach session.

Instructor-led Training Program

The instructor-based training session was created to train the same informationpresented by the computer version. The live version of this training utilized the videoclips from the CD-ROM version of the training. The researcher created the PowerPointpresentation and facilitator notes directly from the CD-ROM version of the training. Amember of the University Legal Counsel (blind to the hypotheses) facilitated the livetraining. This session also took approximately 1.5-2 hours and covered the same topicsas listed above. This session allowed participants to ask questions of the facilitator, takenotes if needed, and interact with the other participants.


Design and ProcedureUpon arriving to the training session, all participants were instructed to complete aninformed consent sheet, a demographic questionnaire and a pre-test. The pre-testassessed participant current knowledge, skills and attitudes about sexual harassment.Following the pre-test, the participants began the training session. The participantswere randomly assigned to attend either the instructor-led or the computer-basedtraining session. Following the training session, all participants completed a post-testwith equivalent information to the pre-test. The post-test was used to measurereactions to the training method and also changes in sexual harassment knowledge,skills and attitudes. After the post-test, trainees were debriefed about the study andfurther questions were answered.

Measures

Demographic questionnaire. The demographic questionnaire consisted of sevenquestions and asked employees about gender, work experience, age, education, ethnicbackground, work experience, and familiarity with sexual harassment training. Thesequestions were utilized to describe group characteristics and also to see if there weregroup differences based on any of the variables.

Training type preference. Trainees filled out four questions designed to determine theirpreference for computer or live training, before and after the training. A sample item is,“I prefer training to be instructor-led.” Responses were made on Likert scales rangingfrom 1 (Strongly Disagree) to 5 (Strongly Agree). This was used to assess which type oftraining is preferred overall before training, and if trainees preference for instructor vs.computer-based training had changed. Coefficient alphas for all scales can be found inTable 1.

Learning. The learning test consisted of 30 questions and was designed to measuretrainee learning based on the three types of learning: cognitive, skill-based, andaffective described by Kraiger et al. (1999).42 These questions were derived directlyfrom the content of the training courses and the comprehensive test following thecomputer-based training session. Some of the questions from the comprehensive teston the CD-ROM were slightly modified and not all the questions from the CD-ROM testwere used because there were several versions of the test on the disk. There were 26multiple choice items that were cognitively based and dealt primarily with definitionalissues (e.g., “Repeated sexually oriented, unwelcome behavior in the workplace islegally referred to as: (a) hostile environment sexual harassment, (b) quid pro quosexual harassment, (c) aggressive harassment, (d) aggravated assault). Four multiplechoice items were skill-based, presenting trainees with a scenario and asking them toapply skills learned in addressing the question (e.g., “Jane was touched on theshoulder by Ray when she was alone in her office, and by Tom at an office meeting.Why might she perceive Ray’s action as harassing but not Tom’s? (a) Ray intended toharass Jane. (b) Jane is an unfair person by nature, (c) Jane’s dress was provocative, so


she excused Tom’s actions, (d) Jane probably perceives herself as less vulnerable in agroup setting). Five Likert-type items dealt with participants' affective reactions to theissue of sexual harassment (e.g., People who are sexually harassed usually invite it.)Response scales ranged from 1 (Strongly Disagree) to 5 (Strongly Agree). Participantsfrom both types of training took the same pretest and posttest administered in a paperand pencil format.

Reactions. The reaction questionnaire was used to get a general reaction from thetrainees about the training session. There were ten questions designed to assess iftrainees liked the course, the content, and most importantly the method of trainingdelivery. Seven of the questions were Likert type items where responses ranged from 1(Strongly Disagree) to 5 (Strongly Agree). A sample item is, “Overall, I was satisfied with


Table 1: Means, Standard Deviations, and Intercorrelations amongMajor Resea rch Variables

Variable M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. Groupa 1.5 .50 -

2. Overall Pretestb 17.8 3.1 .12 .61

3. Cognitive (pre)c 15.7 2.6 .10 .95** .58

4. Skill (pre)d 2.1 1.0 .11 .59** .31** .25

5. Preference (pre) (Live)e 2.8 .85 .14 .12 .13 .03 .57

6. Preference (pre) (Computer)e 2.6 .60 -.04 -.10 -.14 .08 -.41** .52

7. Affect (pre)f 3.7 .68 .18 .26* .31** .01 .10 -.05 .52

8. Overall Posttestb 21.7 2.2 -.06 .55** .51** .35** .04 -.11 .15 .56

9. Cognitive (post)c 18.4 1.9 -.02 .56** .57** .23 .01 -.14 .24* .94** .53

10. Skill (post)d 3.3 .75 -.13 .20 .07 .43** .07 .02 -.16 .56** .25* .11

11. Preference (post) (Live)e 2.7 .80 .19 .22 .22 .12 .78** -.38** .14 .21 .24* .01 .60

12. Preference (post)e (Computer) 2.6 .67 .00 -.04 -.08 .09 -.21 .61** -.04 -.04 -.07 .07 -.18 .66

13. Affect (post)f 3.4 .87 .19 .22 .21 .12 -.09 .04 .68** -.05 .00 -.16 -.08 -.03 .57

14. Reaction g 4.3 .43 -.13 -.00 .05 -.14 -.12 .04 .09 -.06 -.06 -.16 -.12 -.15 .06 .86

Note. N = 70. Reliability coefficients are located in the diagonal in boldface type.

* p < .05 (two-tailed). ** p < .001 (two-tailed).a Group distinguishes computer-based training participants from instructor-led training partici-pants. The instructor-led training was coded as 1 and the computer-based training was codedas 2. b The pretest and posttest measures contained 25 items. c The pre and post cognitivemeasures consist of 21 items (also in the overall pre and post test); this is labeled as a scoreout of 21. d The pre and post skill measures consist of 4 items. e The pre and post preferencemeasures each contained 2 questions. These items range from 1 (Strongly Disagree) to 5(Strongly Agree). f The pre and post measures of affect each originally contained 3 items.These items range from 1 (Strongly Disagree) to 5 (Strongly Agree). In the post affect measure,question 7 was eliminated to increase the reliability of the measure. g The reaction measureconsisted of 7 items ranging from1 (Strongly Disagree) to 5 (Strongly Agree).

this training course”. Three items were open-ended, (e.g., What did you like mostabout this session?). This measure was only administered after the training session toget immediate reactions to the specific training.

Results

Descriptive Statistics

A correlation matrix including selected research variables can be found in Table 1.Correlations indicate the degree of linear relationship between variables. Reliabilityestimates which show how much internal consistency is present in a measure areprovided along the diagonal. Using the Bonferroni approach to control for Type I erroracross the 91 correlations; a p value of .001 (.05 / 95 = .001) or less was required forsignificance. Correlations significant at the .05 level also are noted in the correlationmatrix. Correlation coefficients also were computed for the demographic informationcollected. The demographic variables, including gender, were not significantlycorrelated with any of the major research variables and are not included in Table 1. Theindependent variable, computer or instructor-based, was not significantly correlatedwith any of the pretest or posttest information. However, the pretest and the posttestscores for the learning measure overall were significantly correlated, r(68) =.55, p <.001. This correlation indicates that higher pretest scores were associated with higherposttest scores. The pre and post scores on the skill-based learning measure weresignificantly correlated, r(68) = .43, p < .001, as were the cognitive learning measures,r(68) =.57, p < .001, which indicates that the scores on these two pretest and posttestmeasures were related. Regardless of training method, scores on all learning measuresincreased. Means and standard deviations for pre and post test measures can be foundin Table 2. The correlations for the pre and post preference for both computer-basedtraining, r(68) =.61, p < .001, and instructor-led training, r(68) = .78, p < .001, weresignificant indicating that pre-training and post-training preferences for computer-based on instructor-led training tended to be consistent over time. However, trainingtype preference was not correlated with reactions and was therefore not used as acovariate or control variable in the analysis. The pre and post scores on the affectmeasure also were significantly correlated at r(68) = .68, p < .001. Due to theseintercorrelations, pretest scores were used as covariates to control for preexsistingdifferences in all subsequent analyses.

Reaction Data

The reaction data were analyzed using a one-way analysis of variance (ANOVA) whichindicates whether there are differences across groups. The independent variable, typeof training, has two levels: instructor-led and computer-based training. The dependentvariable or outcomes was the reaction test scores. Contrary to hypothesis 1, the ANOVAwas not significant, F(1, 68) = 1.143, p = .29 suggesting that participant reactions weresimilarly positive regardless of the mode of training delivery.


Learning Data

To analyze the learning data, three one-way analyses of covariance (ANCOVA) wereperformed using type of training as the independent variable, learning test (cognitive,skill-based, and affective outcomes) scores as the dependent variable, and pretestinformation as the covariate. ANCOVA tests for differences between groups controllingfor pretest differences. The analysis for the homogeneity-of-slopes assumptionindicated that learning test post scores for the cognitive and skill-based and therespective pretest scores did not differ significantly as a function of type of training,F(1,66) = .577, p = .45, partial η² =.01 (cognitive) and F(1,66) = .723, p = .398, partialη² = .01 (skill-based). Neither ANCOVA was significant, F(1, 67) = 5.15, p = .475(cognitive) and F(1,67) = 2.93, p = .092 (skill-based). Because neither the cognitivenor the skill-based outcomes were significantly higher for the live training controllingfor pretest scores, Hypotheses 2 and 3 were not supported. There were no significantdifferences between live training and computer-based training groups on cognitive orskill-based learning.

The homogeneity-of-slopes assumption was analyzed for the affective outcomedata and indicated that the post affective outcomes and the type of training did differsignificantly as a function of the pre affective outcomes. Because of this violation, anANCOVA was not utilized for this hypothesis. The average affect was computed for eachparticipant and a one-way analysis of variance was conducted. The independentvariable was training type and the dependent variable was the average affect towardssexual harassment issues. The ANOVA also was not significant, F(1,68) = 2.566, p =.114. The instructor-led training participants did not have a significantly more positivechange in affect than those in the computer-based session which was inconsistent withHypothesis 4.

DiscussionThe purpose of this research was to examine whether computer-based vs. instructor-led sexual harassment training was superior in generating positive trainee reactionsand learning (cognitive, skill-based, and affective). Overall, the results suggested thathaving a training course on sexual harassment, regardless of method of training,increased learning among the university employees who participated in these sessions.

Reactions Overall, the reactions to both types of sexual harassment training were positive. Mostparticipants indicated reaction scores of “agree” and “strongly agree”, indicatingsatisfaction with the training regardless of training method. The hypothesis thatinstructor-led participant reactions would be more positive was not supported. Theopen-ended questions for the instructor-led group indicated that the facilitator waswell received and considered an “authority” on sexual harassment issues and that thetraining was very clear and organized. The computer-based group thought that thetraining was easy to understand and follow and the content was very informative. Bothgroups noted the usefulness of the video clips used in the sessions.


Learning and Trainee PreferencesIn both conditions, participant learning increased (See Table 2). However, mode oftraining delivery did not differentially impact cognitive/skill-based or affective learning.This result would likely be viewed positively by organizations looking to lower costs forsexual harassment training because purchasing a computer-based training programand administering it throughout your organization is generally less costly thanpresenting instructor-led training sessions.

In addition to considering program cost when determining which type of trainingto provide, organizations might also consider employee preferences. In the currentstudy, the mean for a preference for instructor-led training was 3.8 before and aftertraining. The mean for computer-based training preference was 2.8 before training and3.0 after training. These data suggest a very slight preference for instructor-led trainingand are consistent with previous research.43 Affective outcomes did not seem tochange much following training.

ImplicationsWhen choosing which type of sexual harassment training to implement, a non-trivialissue is the location of the targeted employees. If the to-be-trained employees arelocated all over the country or the globe, computer-based training may be the bestalternative.44 However, if the goal of training is to create rapport between differentbranches or companies in a firm, the extra cost for the travel and facilitator might bewarranted. Another consideration should be the urgency of the training. If the trainingneeds to be completed quickly, computer-based training may be the answer. Computer-


Table 2: Pre and Post Training Means and Standard Deviations

Measure n Group Pre Mean (SD) Post Mean (SD)

Cognitive Live Training 15.4 (2.9) 18.4 (1.7)35

Computer-Based 15.9 (2.4) 18.4 (2.1)35

Skills Live Training 2.0 (1.2) 3.4 (.69)35

Computer-Based 2.6 (.89) 3.2 (.80)35

Affective Live Training 2.4 (.74) 2.4 (.65)35

Computer-Based 2.1 (.59) 2.2 (.68)35

Preference Live Training 3.8 (1.0) 3.8 (.92)35

Computer-Based 2.8 (.75) 3.0 (.83)35

based training can be loaded to an internet or intranet site relatively quickly and thetraining can generally be bought from a vendor or off-the-shelf.

In order to make effective use of computer-based training, some researchers haveadvocated using blended training where computer-based training is used to deliver thebasic information or content to be covered in training and then following-up withinstructor-led training.45 This should speed up the instructor-led training as well asallow more time for trainee questions, application of the new information, and/orgroup work. Others have suggested adaptive guidance as a way to improve computer-based training by incorporating missing elements from the instructor-led training.Essentially, adaptive guidance is a form of electronic feedback designed to help traineesmake more effective learning decisions. It helps trainees interpret their pastperformance as well as suggest a course of action for future study and practice toimprove skills.46

LimitationsThere are limitations to consider when interpreting the results of this study. Despiteattempts to standardize training content, there were minor differences in the two typesof training. For example, because of its live nature, the instructor-led training likely hadmore variation. For instance, it is likely that the trainer worded things slightlydifferently across sessions. Additionally, the instructor-led sessions included moreparticipant interaction. These issues are not unique to this study; inconsistent materialis a common disadvantage of instructor-led training. Pre and post data however seemto suggest that trainees had similar experiences in training yielding similar scores.Realistically though, it is nearly impossible to create computer-based and instructor-ledtraining sessions that are exactly the same. There are inherent differences in the twotypes of training which cannot be avoided.

The second limitation is the lack of evaluation of the behavior and results levels ofthe Kirkpatrick (1959) model.47 Although this study certainly is a step in the rightdirection – offering much needed field data evaluating sexual harassment training ingeneral as well as the specific type of training program, we did not have theopportunity to examine the last two dimensions of Kirkpatrick’s model. Unfortunately,even if we had had the opportunity to measure these variables, it is unlikely that wewould have found significant effects given the relatively small sample of employees.That is, the likelihood of anyone being charged with sexual harassment is fairly slim.Future research with a larger organization should try to evaluate all the levels in orderto get the greatest benefit from the evaluation process.

A final limitation relates to the generalizability of the results. This study utilized asample of university employees in the midwest. All participants voluntarily participated.If the training had been mandatory, participants may not have been as motivated tolearn the material as those who volunteered. University employees may differ fromthose in other public or private organizations. Before generalizing these results toother work groups, it is suggested to compare demographic information (age, workexperience levels, etc.) as well as the type of sexual harassment training being utilized.


ConclusionOur data do not suggest replacing all instructor-led training with computer-basedtraining but they do indicate that computer-based training may be a viable alternativeto instructor-led training in some instances. Future research is needed to determinethe types of training that are conducive to computer-based training and what elementsto look for when shopping for an effective off-the-shelf training session. Organizationswould be wise to continue to utilize instructor-led training but also experiment withcomputer-based training to ease employees into using a different method for learning.Overall, trainees learned about sexual harassment and responded very positively to thetraining.

Notes1 Thompson, C., Koon, E., Woodwell, W.H., & Beauvais, J. (2002). Training for the next economy:An ASTD state of the industry report on trends in employer-provided training in the UnitedStates, Report by American Society of Training and Development, Alexandria, VA.

2 Gordon, J., & Hequet, M., (1997). Live and in person, Training, 34, 24-31.

3 Faley, R.H., Knapp, D.E., Kustis, & Dubois, C.L. (1999). Estimating the organizational costs ofsexual harassment: The case of the U.S. army, Journal of Business and Psychology, 13, 461-484;Glomb, Munson, Hulin, Bergman, & Drasgow, (1999). Structural equation models of sexualharassment: Longitudinal explorations and cross-sectional generalizations, Journal of AppliedPsychology, 84, 14-28; Mueller, C.W., DeCoster, S., & Estes, S.B., (2001). Sexual harassment inthe workplace, Work and Occupations, 28, 411-446; Peirce, E., Smolinski, C.A., & Rosen, B.(1998). Why sexual harassment complaints have fallen on deaf ears, Academy of ManagementExecutive, 12,, 41-54.

4 Bland, T.S., & Stalcup, S.S. (2001). Managing Harassment, Human Resource Management, 40,51-61.

5 Gibson, S.G., & Roberts-Fox, H., (1998). Supreme court ruling on sexual harassment disputes,"The Industrial-Organizational Psychologist, 36; Peirce, Smolinski, & Rosen, 1998.

6 Faley, Knapp, Kustis, & Dubois. (1999).

7 Equal employment opportunity guidelines on sexual harassment. (1980). 45 Fed. Reg. 25025.

8 Blakely, G.L., Blakely, E.H. & Moorman, R.H. (1998).The effects of training on perceptions ofsexual harassment allegation. Journal of Applied Social Psychology, 28, 71-83.

9 Keyton, J., & Rhodes, S.C., (1999). Organizational sexual harassment: Translating research intoapplication, Journal of Applied Communication Research, 27, 158-173.

10 Laabs, J.J. (1995). Sexual harassment, Personnel Journal, 74, 36-46.

11 Society for Human Resource Management, (1999). SHRM Sexual Harassment Survey,Alexandria, VA: SHRM.

12 Beauvais, K., (1986). Workshops to combat sexual harassment: A case study of changingattitudes," Signs, 12, 130-145; Glomb, Munson, Hulin, Bergman, & Drasgow, (1999); Icenogle,M.L., Eagle, B.W., Ahamad, S., & Hanks, L.A. (2002). Assessing perceptions of sexual harassmentbehaviors in a manufacturing environment, Journal of Business and Psychology, 16, 601-616;Laabs, J.J. (1995); Licata, B.J., & Popovich, P.M. (1987). Preventing sexual harassment: Aproactive approach, Training and Development Journal, 41, 34-38; Peirce, Smolinski, & Rosen(1998); Meyer, A. (1992). Getting to the heart of sexual harassment, HR Magazine, 37, 82-84.

13 Johnson, M.W., (2004). Harassment and discrimination prevention training: What the lawrequires, Labor Law Journal, 55, 119-129.


14 Johnson (2004).

15 Johnson (2004).

16 Desai, M.S., Richards, T., & Eddy, J.P. (2000). A field experiment: Instructor-based training vs.computer-based training," Journal of Instructional Psychology, 27, 239-243.

17 Leidner, D.E., & Jarvenpaa, S.L. (1993).The information age confronts education: Case studieson electronic classrooms, Information Systems Research, 4, 24-54.

18 Gordon, J., & Hequet, M. (1997).

19 Allen, M. (2003).The lessons of e-learning," Optimize, 41, 23.

20 Brown, K.G., (2001). Using computers to deliver training: Which employees learn and why,?"Personnel Psychology, 54, 271-296.; Gale, S.F. (2003). Making e-learning more than “pixie dust,"Workforce Management, 11, 58-62; Gordon, & Hequet, 1997.

21 Brown (2001); Gale (2003); Gordon & Hequet (1997).

22 Brown (2001).

23 Desai, Richards, & Eddy (2000).

24 Arthur, Bennett, Edens, & Bell (2003). Effectiveness of training in organizations: A meta-analysisof design and evaluation features, Journal of Applied Psychology, 88, 234-245.


26 Desai, M.S., & Richards, T., (1999). End-user training: A meta model, Journal of InstructionalPsychology, 26 , 74-85; Gordon, & Hequet, (1997).

27 Desai, & Richards, (1999).

28 Fizgerald, L.F., & Shullmann, S.L., (1993). Sexual harassment: A research analysis and agenda forthe 1990’s, Journal of Vocational Behavior, 42, 5-27; Moyer, R.S., & Nath, A., (1998). Someeffects of brief training interventions on perceptions of sexual harassment, Journal of AppliedSocial Psychology, 28, 333-356.

29 Bingham, S.H., & Scherer, L.L., (2001). The unexpected effects of a sexual harassmenteducational program. Journal of Applied Behavioral Science, 37, 125-153.

30 Johnson (2004).

31 Johnson (2004).

32 Johnson (2004).

33 Kirkpatrick, D.L., (1959). Techniques for evaluating training programs. Journal of the AmericanSociety of Training and Development, 13, 21-26.

34 Alliger, G.M.. & Janak, E.A. (1989). Kirkpatrick’s levels of training criteria: Thirty years later,"Personnel Psychology, 42, 331-342.

35 Alliger & Janak (1989).

36 Kraiger, Ford & Salas (1993).


38 Arthur, Bennett, Edens, & Bell (2003).

39 Arthur, Bennett, Edens, & Bell (2003).

40 Noe, R., (2005). EmployeeTraining and Development, 3rd edition, Boston: McGraw Hill.

41 Costal Training Technologies Corporation (http://www.coastal.com/products.html)


42 Kraiger, Ford, & Salas (1993).


44 Bell, S., & Kozlowski, S.W.J., (2002). Adaptive guidance: Enhancing self-regulation, knowledge,and performance in technology-based training, Personnel Psychology, 55, 267-306.

45 Desai, Richards & Eddy (2000); Gale (2003).

46 Bell & Kozlowski (2002).

47 Kirkpatrick (1959).

AuthorsMelissa K. Preusser

Training and Organization Development Specialist120 South Central, Suite 1000Clayton, MO 63105314-721-8000 x [email protected]

Lynn K. Bartels, PhD

Southern Illinois University EdwardsvilleDepartment of PsychologyCampus Box 1121Edwardsville, IL 62026-1121(618) [email protected]

Cynthia R. Nordstrom, PhD

Southern Illinois University EdwardsvilleDepartment of PsychologyCampus Box 1121Edwardsville, IL 62026-1121(618) [email protected]

Melissa K. Preusser is the training and organizational development specialist for MillsProperties, Inc., a St. Louis-based property management company. She specializes indeveloping and facilitating management development training solutions. She earned hermaster’s degree in industrial/organizational psychology from Southern Illinois UniversityEdwardsville.

Dr. Lynn K. Bartels is an associate professor of industrial/organizational psychology andassociate department chair at Southern Illinois University Edwardsville. She teachescourses in employee selection and development and does research on assessment cen-ters, interviews and feedback.


Dr. Cynthia Nordstrom is an associate professor in industrial/organizational psycholo-gy at Southern Illinois University Edwardsville. She teaches courses in organizationaldevelopment, leadership and motivation and organizational psychology. Her researchinterests include electronic performance monitoring, training and employee selectioninterviews.


The Impact ofManagement WorkHabits on Public SectorPerformance: A Study ofLocal GovernmentManagers By Jonathan P. West, PhD, and Evan M. Berman, PhD

This study finds that hands-on strategies of working with individual managers areeffective in identifying, addressing and overcoming bad management work habits.Up to now, very little has been known about the work habits of public sectormanagers and their impact on perceptions of organizational performance. Basedon a national survey of senior managers in city governments with populations over50,000, this study finds that while productive work habits are widely present, badmanagement work habits are also significantly present. The latter most commonlyinvolve managers being overly passive, judgmental, defensive, intimidating,closed-minded or tardy in their performance. Bad (or, unproductive) managementwork habits significantly 1) offset performance gains from good (or, productive)management work habits, and 2) are a highly significant barrier to cultivating goodwork habits. Raising awareness of managers’ bad habits is an essential step inaddressing them. Managers’ work habits are significantly associated withperceptions of organizational productivity, and addressing bad management workhabits is therefore a significant pathway to higher performance.

An understudied subject in public administration is the nature and extent ofmanagerial work habits. Many, if not most managers, display productive, pro-fessional work habits, such as being committed to working efficiently, acting

with civility towards others and showing proper commitment to ethics. But it is anec-dotally observed that some managers have unproductive work habits, too; they mightfrequently be tardy, arrogant or closed-minded.1 Indeed, many organizations have atleast a few managers with such exceedingly bad work habits.2 Also, most managers,as all people, have behaviors that can benefit from improvement. The consequencesof bad management work habits can be significant, such as causing problems in deal-ing with stakeholders, becoming obstacles in adopting and implementing modernperformance management work practices, and otherwise affecting the leadership oforganizations.3


The study examines the following questions: What is the nature and extent ofwork habits among managers? How do management work habits, both good and bad,affect perceptions of organizational performance? What is the impact of bad workhabits on the presence of modern, professional work habits? What strategies do seniormanagers use to address the unproductive work habits of subordinate managers andsupervisors, and how effective are these strategies? This study addresses thesequestions through the perceptions of senior managers about the work habits ofmanagers and supervisors that report to them. It is based on an extensive mail surveyamong chief administrative officers in cities with populations over 50,000, as well as in-depth interviews among a sample of respondents. The combination of quantitative andqualitative methods enriches understanding of the study topic, management workhabits, and of efforts to deal with them.

This research contributes to the literature and our knowledge in several ways.First, while classic authors like Mary Parker Follett, Max Weber and Frederick WinslowTaylor all note the importance of productive supervisory practices, systematic analysisof such practices in contemporary government is lacking. For example, while studies inthe 1970s and 1980s, such as Stinson and Mueller4 discuss behaviors associated withmanagerial professionalism (e.g. reading professional journals, attending conferences),these behaviors are insufficiently linked to this study’s specific concern about bad workhabits and organizational performance. Second, whereas studies of modernperformance often discuss specific management strategies,5 including implementationstrategies6 and various conditions affecting adoption and diffusion,7 this article studiescertain behaviors of the protagonists themselves, managers, and their impact onperformance, thus adding to our knowledge of factors affecting performance ofmodern public organizations. Third, this study contributes to recent discussions ofinformal approaches for managing workplace relations, such as through psychologicalcontracts.8

This study has limitations. First, our study necessarily relies on subjective,perceived assessments of managers’ orientations and behaviors, as no objective dataexist of managers’ orientations and behaviors such as based on administrative records.While research practices such as triangulation are used to further the validity of theseperceptions, it is clear that all subjective data are subject to some imperfections; thepossibility of some respondent bias and faulty observation by respondents cannot beruled out. Second, while senior managers are in a good position to assess the workhabits of those managers and supervisors that report to them, other stakeholders suchas their employees may have different perceptions about these subordinate managers’work habits. Third, while we take a broad look at management work habits, no studycan be exhaustive of all possible work habits. This study acknowledges these limitationsand imperfections of the data and the study scope.

FrameworkIn this study, the phrase “work habits” is defined as customary ways individuals behavewhen engaged in work. Managerial work habits involve customary behaviors, such as


typically showing concern for others, setting high standards or demonstratinginitiative. Customary behaviors occur typically, though not necessarily each time. Whilethis definition of ‘work habit’ follows common usage, work habits are in fact betterdescribed as ‘habitudes’ than as habits. Habitudes are repetitive, typical or usualpatterns, whereas habits are defined as deeply ingrained behaviors or practices that areoften done without much conscious thought. Work habits include the latter, as well asbehaviors that involve a good deal of conscious thought, such as being committed to astrong sense of ethics, and the reasons for that. Despite this, the phrase ‘work habit’rather than ‘work habitude’ has become the prevailing, common usage, and it is thusused here.9

Habits serve important functional purposes; they are relatively efficient ways ofdealing with work and other aspects of life. By devoting relatively little attention tomost habits, people are able to save time, conserve cognitive resources and focus moreof their efforts on other matters or situations, such as those that require someimmediate or unfamiliar response.10 The lack of full consciousness about habits isusually justified in so far as they lead to repeatedly satisfactory outcomes, requiringonly modest awareness of situational applications. Work habits typically also involveaccompanying mindsets, which are characteristic outlooks and ways of interpretingsituations. Many authors focus on the mindset (or, attitude)-behavior relationship;11

mindsets affect characteristic behaviors ex ante (“when this happens, thenalways/never do that…”), as well as ex post facto (“I justify my actions like this…”).People vary in their consciousness of accompanying mindsets, and managers aresometimes unconscious of them (“I just usually do things this way; I can’t tell youwhy….”). This may occur when they have forgotten initial thoughts associated withwell-established and long-existing work habits. Consistent with the above definition ofwork habits, and emphasizing the empirical orientation of this manuscript, this articlefocuses on the behavioral manifestations of managerial work habits, whileacknowledging accompanying, cognitive mindsets.12

The terms “good” and “bad” that are used in connection with work habits arecommonly applied with regard to the balance of consequences or outcomes for work.The literature on performance management is quite clear about which work habits arebelieved to further modern performance activities, and those which are believed to bedetrimental. Being regularly or typically aggressive or intimidating with others isgenerally regarded as a (very) bad work habit, as it hinders the ability to get along andthus achieve work goals. While interpersonal forcefulness can occasionally beproductive, it is certainly not productive on a regular basis.13 Being pro-active, settinghigh standards, being flexible, finding ways of getting along well with others, beingcommitted to a strong sense of ethics and having good communication skills aregenerally regarded as furthering high performance.14 Being passive, indecisive, closed-minded, sloppy, tardy or intimidating are among habits frequently associated with lowperformance.15 Indeed, such bad habits can become the Achilles’ heels oforganizations.16 While the above lists are not exhaustive, they are frequently mentionedin numerous articles concerning both public and private organizations.17


Good and bad habits are typically not mutually exclusive; the absence of a badhabit does not imply the presence of a good habit. For example, the absence ofcustomarily being passive (a bad habit) does not imply the presence of a good habitsuch as being pro-active. Good habits and bad habits stand apart, and may affect eachother. Generally, the relationship between good work habits and bad work habits isassumed to be a negative one. For example, being customarily proactive (a good habit)makes it less likely that someone is routinely tardy (a bad habit), as being proactiveusually means that one is concerned about things that lie ahead, including work that isdue, and barriers that might affect it. A general mindset about professionalmanagement practices may favor good habits and actively work against bad habits,broadly. A very important implication, sometimes overlooked by practitioners andscholars alike, is that getting managers to adopt good work habits may require them toarrest or reduce pertinent bad work habits, too. Knowledge and positive intent toadopt good work habits may not be enough for implementing management reforms.For example, seminars and training activities touting the virtues and promotingdevelopment of good management habits are likely to be only partially successful atbest in the absence of also addressing management bad habits.18

In this study, good managerial work habits are assumed to further organizationalperformance, and bad work managerial habits are assumed to reduce organizationalperformance. As the saying goes, one rotten apple spoils the barrel. It takes only a fewmanagers with a few unproductive work habits such as being overly passive to reducetheir department’s performance by lowering expectations, making it harder for topmangers to implement jurisdiction-wide reforms, and creating a variety of problemsthat require top management attention. Anecdotally, some managers are described as‘killers’ of management reform.19 By contrast, studies of high performance suggest thatmany managers in such settings have a broad range of productive orientations andbehaviors. They may vary in these, but they need to have enough of them (e.g., beingpro-active, having good communication skills, taking a stand on ethical issues) to dealeffectively with the various technical, human, networking and leadership aspects of themanagerial job. From this also follows another important implication, namely, thatstandards for evaluating good and bad work habits differ. Organizations need toevaluate good management habits by their abundance, and bad habits by their nearabsence.

There is ample basis in the literature to suspect that reducing or eliminating badwork habits is often a difficult task. First, many bad habits involve beneficial outcomesof a personal nature; for example, tardiness as a habit allows people to reduce stress,and intimidation allows people to control others. Eliminating bad habits implies findingother ways of dealing with these human challenges of stress and control. People mayneed help in devising new adaptive strategies and dealing with unfavorable underlyingconditions.20 Second, a key feature of all habits is that they do not require muchthought. People may be largely unaware or in denial about their bad habits, or at leastchallenge the need for any course that diminishes the personal benefits that theyderive from them.21 Indeed, a management process of changing bad work habits iswell-established in the psychological literature and builds upon the work of Lewin,22


often involving four specific steps, such as, uncovering, redefining, learning, andrelearning to produce change in work habits;23 the first step involves awareness orrealization made by a manager that their bad work habits are problematic. Berman24

uses the phrase “rule of three” to suggest that people often need to hear informationseveral times before they take it seriously.

The above factors help explain why relinquishing bad habits is often a difficultprocess that typically requires persistence, practice and, hence, considerable time.25

Processes of changing work habits are also furthered by organizational policies. Forexample, organizations can include manifest behaviors as criteria in performanceappraisals of managers and promotion decisions. They can reward managers forproductive orientations and behaviors, and punish others for unproductive ones.Organizations can also institute training programs that stress good work habits andidentify and correct those that are bad or unproductive. Formalizing mentor programsand supporting on-the-job coaching can help managers to identify habit-based workpractices that are positive or problematic, leading to tailored feedback, realistic goalsetting, and appropriate action steps for improved performance. The literature onleadership, productivity, organizational change, learning theory, and staff developmentsuggests these strategies have been used successfully.26

MethodsDuring 2006, a mail survey of city managers and chief administrative officers (CAOs)was undertaken in all 653 U.S. cities with populations over 50,000,27 which wasfollowed-up by in-depth, telephone interviews with selected respondents in Spring andSummer 2006. The extensive mail survey, which has over 275 items, involved a pilotsurvey and three rounds of mailings that resulted in a final sample of 212 completedresponses. As discussed below, we examined for possible problems of sample bias andnon-response, but find these to be very minor and of little material import. Theresponse rate of 32.5% is consistent with that reported elsewhere.28 Regarding thesample, 55.5% of surveys were completed by the addressee (City Manager or ChiefAdministrative Officer, as appropriate), and, of the remainder, about half werecompleted by an assistant city manager. Almost all of the remaining were completed byrespondents with such titles as city clerk, director of human resources, director ofadministrative services, or chief of staff. The respondent sample is referred to as “seniormanagers,” reflecting this diversity of positions. On average, respondents indicatedthat they have worked 22.5 years in government, and have spent 11.7 years within theirpresent jurisdiction. 89.7% of respondents stated that they were very familiar orfamiliar with the performance of managers in their jurisdiction. This sample consistedof senior, informed leaders.

Among the respondents, 53.8% report that their highest degree was in publicadministration, the remainder reported business administration, political science, law,engineering, urban planning and other fields. Of the respondents, 70.1% ofrespondents have an MA degree, and 69.3% are male. 19.6% are younger than 45 yearsold, 37.8% are between the ages of 45 and 54, and 42.6% are older than 54. The sample


is broadly representative of the population of cities.29 In the sample, 69.7% ofrespondents are from cities with populations between 50,000 and 99,999, 19.0% of therespondents are from cities between 100,000 and 249,999, and the remainder, 10.9%are from cities with populations of 250,000 and more. Of the population of cities, 62.2%had populations between 50,000 and 99,999, 27.4% had populations between 100,000and 249,999, and 10.4% had populations of 250,000 and more. In the sample, 75.4% ofthese respondents are in cities that have a city-manager form of government,compared to 62.5% of cities in the population. In the subsequent analysis, we examinethe impact of such minor differences in city size and form of government, but concludethat these do not materially affect study findings.

Measurement validity is furthered in the following ways. First, the survey assessesbehaviors and communications of subordinate managers and supervisors that directlyreport to respondents. Such respondents are likely to be well informed regarding thebehaviors and communications of these direct subordinates as a function ofrespondents’ responsibilities.30 Second, the survey items address empiricallyobservable phenomena such as behaviors and communications, rather than mindsetsof subordinates which cannot be directly or even accurately observed. This surveyavoids, for the most part, using the term ‘habit’ which, for some, may also havejudgmental connotations. However, the term “work habit” is used in the context ofaddressing work orientations and behaviors where doing so in the survey produceslittle ambiguity (see Table 3) and where it is a more efficient way of expressing thisphrase.31 The research (and, hence survey questions) direct respondents to assesshabits within the body of managers and supervisors that report to them, rather thanindividual managers.

The possibility of sample bias is examined by comparing the responses ofaddressees (city managers and CAOs) with those of other respondents. While a fewsuch differences exist, they are relatively minor and do not significantly affect ourresults reported here. We also compared the responses of those working more versusfewer years in their present jurisdictions, and further assessed whether relevantpersonal orientations might be associated with different perceptions of habits, but findno such relationships. For example, respondents who express a strong work ethic forthemselves (e.g., “I like a busy schedule” or “I feel guilty when I take a break fromwork”) are not associated with perceptions of bad habits among managers reporting tothem (both p > .05). We also conducted more than a dozen in-depth interviews(approximately one hour each) among managers in responding jurisdictions, includingthose who indicate very high or low prevalence of bad habits. This helped to validatemail survey responses. Finally, to examine non-response bias, we completed a phonesurvey among a random sample of non-respondents of the mail survey (n=41).Comparing these responses to those of respondents of the mail survey, we find nostatistically significant difference between respondents and non-respondents to themail survey.32 We also examined differences between early and late responders, but findno significant differences between these groups of respondents.


Results

Management Work Habits Tables 1 and 2 shows perceptions of, respectively, good and bad management workhabits. Table 1 shows the prevalence of productive (or, good) work habits. Amongrespondents, 88.6% agree or strongly agree that managers and supervisors who reportdirectly to them regularly demonstrate civility, courtesy and decency.33 Respondentsalso agree or strongly agree that they take a stand where issues of ethics are at stake(86.9%), safeguard confidential information (86.9%), show concern for the personalwell being of team members (85.7%), refuse to comply with unethical requests(82.9%), consider the rights of others in making decisions (80.0%), and set highstandards for themselves (77.0%). Clearly, most senior managers note many goodmanagement work habits in their subordinate managers and supervisors.34 Perhaps it isnot so surprising that many respondents have favorable assessments, as these habitsare de facto requirements of many managers’ jobs. About half of the respondents note


Table 1: Managers’ Good Work Habits

“Managers and supervisors who report directly to me regularly…” Agree or Association withStrongly Agree Bad Habits

A. Work Productivity work efficiently in getting the job done 76.6 -.381**set high standards for themselves 77.0 -.290**demonstrate proper motivation at work 71.7 -.213**show initiative and are pro-active in decision-making 68.5 -.348**are flexible and quick to adapt 55.6 -.303**

Index 68.0 -.392**

B. Work Collaboration & People Relations demonstrate civility, courtesy and decency 88.6 -.211**show concern for the personal well being of team members 85.7 -.239**demonstrate a caring attitude toward others 79.8 -.231**contribute their “fair share” on collaborative tasks 74.3 -.298**work effectively with other managers and supervisors 74.7 -.371**seek the input of those affected by decisions 69.1 -.266**listen to others’ views before making any judgments 55.4 -.174**

Index 83.0 -.345**

C. Ethics take a stand where issues of ethics are at stake 86.9 -.178**safeguard confidential information 86.9 -.295**refuse to comply with unethical requests 82.9 -.196**consider the rights of others in making decisions 80.0 -.267**avoid conflicts of interest 79.5 -.236**

Index 87.4 -.294**

Cronbach alpha scores of index variables are 0.80 (work productivity), 0.86 (work collaboration &people relations), 0.74 (ethics). The aggregate index has alpha = 0.80

Agree or Strongly Agree for index variables defined as scores between 1.0 and 2

that subordinate managers are flexible and quick to adapt (55.6%) and listen to othersviews before making judgments (55.4%).

Table 2 shows perceptions of bad work habits among subordinate managers andsupervisors. Table 2 shows that, among bad work habits, most common is being overlyjudgmental of the shortcomings of others (39.1% of respondents who strongly agree,agree or somewhat agree that at least one manager reporting to the respondent hasthis work habit), overly aggressive and intimidating to others (33.3%), being passiveand unwilling to take needed actions (33.3%), and being defensive and unable to take


Table 2: Managers’ Bad Work Habits

Among managers and supervisors who report directly to me, there are one or more who Somewhat Somewhatregularly are….” Agree1 Agree Disagree Disagree2 DK

A. Work Passivitypassive and unwilling to take needed action 10.9 22.4 13.2 49.4 4.0unable to complete work on time 6.9 14.9 19.0 56.3 2.9repeatedly late for work or meetings 6.4 11.0 13.9 65.3 3.5indecisive and unwilling to take a stand 5.2 13.8 20.1 58.6 2.3

B. Collaboration overly judgmental of the shortcoming of others 19.0 20.1 16.1 40.3 4.6overly aggressive and intimidating to others 14.4 19.0 17.8 46.5 2.3willing to blame their failures on others 8.0 9.8 21.8 55.7 4.6willing to claim credit for the accomplishments of others 9.2 12.1 16.1 60.3 2.3

distrustful of the motives of other team members 4.7 18.0 12.2 61.0 4.1a source of unnecessary conflict with team members 5.2 9.2 16.2 68.2 1.2

unwilling to compromise with others 2.3 7.5 25.4 62.4 2.3

C. Ideas and Informationdefensive and unable to accept criticism 9.8 17.8 21.3 48.3 2.9closed to new ways of doing things 3.4 15.5 19.5 59.8 1.7prone to make hasty decisions without due deliberation 3.4 10.3 26.4 56.9 2.9

unwilling or unable to listen attentively 2.9 9.2 19.5 67.8 0.6

D. Self-management perform their job well some days but poorly on other days 4.6 10.9 23.0 56.3 5.2

unable to keep promises 2.9 6.9 16.1 69.5 4.6unable to control his/her temper 2.3 8.6 14.9 70.3 2.9knowingly make unreasonable demands 1.7 4.6 14.9 75.3 3.5does not recognize the impact of their feelings on their actions 1.1 3.9 33.7 55.1 6.2

1) Includes Strongly Agree and Agree 2) Includes Somewhat Disagree, Disagree and Strongly Disagree

Cronbach alpha scores of index variables are 0.79 (work passivity), 0.87 (work collaboration &people relations), 0.79 (ideas and information) and 0.65 (self-management). The aggregateindex has alpha = 0.90

criticism (27.6%). Other bad work habits include not being able to complete work ontime (21.8%), being distrustful of the motives of team members (22.7%) and claimingcredit for the accomplishments of others (21.3%). The items shown in Table 2 aregrouped into four categories: work habits that reflect work passivity, concern problemsof self-management, as well as those that impede collaboration, and hamperacceptance of new ideas and decision-making. Somewhat fewer respondents reportproblems of self-management such as inconsistent performance on different days(15.5%) and managers being unable to control their temper (10.9%).35

Table 1 also shows associations with bad work habits, discussed in the aboveTable. It is important to note that the good habits in Table 1 are worded quitedifferently from the bad work habits in Table 2. Then, Table 1 shows that each of thegood habits is statistically negatively associated with the index of bad work habits (thealpha value of this index is 0.90, shown in Table 2). Beyond this, further analysis alsoshows that the following bad work habits are most strongly, negatively associated withthe index of good work habits: managers do not recognize the impact of their feelingson their actions (tau-c=-.408, p <. 01), managers are closed to new ways of doingthings (tau-c=.-454, p <.01), and managers are willing to blame their failures on others(tau-c=.-416, p<.01).36 The strong negative association between good work habits andbad work habits suggests that they may crowd each other out; the presence of badwork habits can be a barrier to developing good work habits, and the presence of goodwork habits presupposes the absence or very limited existence of bad work habits. Thematter of causality is taken up in the next section.

Also, while the prevalence of any specific bad management work habits in Table 2is not very high (especially among those who agree or strongly agree), further analysisshows that many managers are likely to experience one or more of these. Figure 1shows that, among respondents, 15.7% agree or strongly agree that three or more badhabits are present (such jurisdictions are labeled “pervasive”), 29.7% of respondentsagree or strongly agree that one or two bad work habits are present, or ‘somewhat’agree that five or more bad work habits are present (labeled “present”), and 17.4% ofrespondents somewhat agree that two, three or four bad habits are present (labeled“somewhat present”). While the cut-off points of these categories are somewhatdebatable and subject to what-if analysis, it should be noted that about two-thirds ofrespondents at least somewhat agree that two or more bad habits are present.37 Whilecities vary greatly in the bad work habits of managers, about 45% of respondents reportthat one or more bad management work habits are present. The same items thatwere mentioned earlier as being most prevalent are also most prevalent among citieslabeled as “pervasive” or “present” regarding bad managerial work habits.

Finally, the presence of work habits, good or bad, does not vary significantly bycity size, form of government or region (respectively, F= 2.84, 3.76, 3.86, all p > .05).38

However, using the above classification, we find that cities with populations between100,000 and 249,999 have a somewhat lower prevalence of bad management habits,according to respondents. For example, in such cities 54.8% are classified as not havingbad management work habits, compared with 34.7% of those with populations lessthan 100,000, and 21.1% of those with populations over 250,000. Perceptions of bad


management work habits are not associated with respondents’ age, gender or highestattained degree (respectively, F = 0.40, 2.94, and 1.21, all p > .05), nor are theyassociated with whether the survey was completed by the addressee or someone else(t= -0.25, p > .05). Perceptions of good habits also are not associated with whether thesurvey was completed by the addressee, or the gender, age or the highest attaineddegree of respondents.

Work Habit Management Strategies and Outcomes Table 3 shows the strategies that senior managers use in dealing with their subordinatemanagers and supervisors. Managers commonly point out changes that need to bemade by their subordinate managers (89.2%), and often challenge managers to changetheir work habits (76.4%). About two-thirds of managers also call attention to negativework habits (65.9%), gather evidence to support the needed change (60.8%), openlydiscuss with managers the importance of good work habits (60.8%), set explicit goalsfor new work habits managers need to achieve (56.3%), include work habits as acriterion in performance appraisal (61.9%) and build good work habits into theorganizational culture (59.7%). Slightly more than half of respondents also stress thefollowing: indicating that not changing some habits is unacceptable (53.4%),celebrating the success of changed work habits (53.4%), and sharing stories of successand difficulties in changing work habits (55.7%). Table 3 shows that a broad range of


Figure 1: Bad Work Habits Among Managers

Present29.7%

Pervasive15.7%

Not Present37.2%

SomewhatPresent17.4%

appropriate change-oriented strategies are used. On average, cities use 14.5 of these 31strategies. The high frequency of these tactics is consistent with the above finding thatabout two-thirds of cities experience bad managerial work habits to some extent.


Table 3: Work Habit Management

“In your experience with your current organization, which of the following actions have you undertaken in relation to another manager or supervisor reporting to you?” Yes

A. Awareness – problem identification pointed out what change is needed 89.2 %challenged the manager to change work habits 76.4called attention to negative effects of bad work habits 65.9gathered evidence to support that change was needed in work habits 60.8 stressed that not changing some habits is unacceptable 53.4 escalated my admonishments to alter work habits 37.9encouraged acknowledgement that current work habits were “bad” 35.2

Index (mean) 59.8

B. Awareness – solution identification openly discussed with the manager the importance of good work habits 60.8shared stories of success and difficulties in changing work habits 55.7 provided information about managerial “good” work habits 49.4created a motivating vision of what new work habits will produce 48.9 insured that the manager had knowledge and skills to change work habits 47.7brainstormed as a management team to identify desirable work habits 46.0told a manager to look towards a promising future with good work habits 34.7 showed how change in work habits advances shared values 25.0

Index (mean) 46.0

C. Management set explicit goals for new work habits managers need to achieve 56.3celebrated the success of changed work habits 53.4encouraged the manager to take the first step in changing work habits 53.4 redesigned work in ways that prompt changed work habits 40.3 provided incentives for continuing with positive work habits 35.2provided evidence that changed work habits were successful 35.2provided a coach to work one-on-one to support new work habits 27.8proceeded step-by-step in the direction of improved work habits 20.5separated the manager from influences that reinforced bad habits 14.8

Index (mean) 37.4

D. Policy included work habits as a criterion in performance appraisal 61.9built good work habits into the organizational culture 59.7 included work habits as a criterion in promotion decisions 45.5instituted training programs that stress good work habits 42.0used management development programs to cultivate good work habits 40.9established organizational standards to institutionalize good work habits 40.3rewarded managers for good habits and punished for bad habits 37.5

Index (mean) 46.8

Cronbach alpha scores of index variables are 0.81 (problem identification), 0.77 (solution identification), 0.74 (management work, 0.76 (policy). The aggregate index has alpha = 0.82

Table 3 also groups these strategies into four categories that correspond with thesubstantive activities noted in the framework; bringing awareness to the existence ofbad work habits (A), providing information about new behaviors and preferredoutcomes (B), working with subordinate managers and supervisors to realize the newwork habits (C), and the use of various organizational strategies and policies in supportof these efforts (D). The alpha measures for these index variables are shown in Table 3.Further analysis of these four index variables shows that only bringing awareness to theexistence of bad work habits (A), is directly associated with the reduction in badmanagerial habits (tau-c = -.185, p <.01); however, the other three index variables(identified as B, C,. and D) are associated with problem identification (respectively, tau-c = .590, .413 and .291, all p <. 01) and, hence, indirectly associated with the reductionof bad management work habits. This finding suggests that making subordinatemanagers aware of the existence of their bad work habits is key to the process ofeffecting change. Change processes that only exhort managers to adopt new workhabits, even when accompanied by reinforcing policies, are not associated withchanges in bad work habits.

A city manager from a California city suggests that a key to effecting change isincreasing awareness by directly confronting subordinate managers and supervisors oftheir bad work habits:

I think the first thing is feedback, and as soon as possible. Let the manager knowright away, and it has to be consistent with follow-ups. For example, we had asupervisor of a water treatment plant that was a 24 hour operation. He keptfalling asleep on the job. We are responsible for providing our citizens withclean water, and there could be severe consequences if the supervisor wasasleep during a critical juncture of the job; we could become legally liable forcontaminated water. The supervisor was reprimanded three times.... Yes, Ifrequently use feedback but it has to be immediate. If not, the entire situationbecomes difficult.

But this manager also admits, “Confronting people is difficult. I learned thatignoring the situation is a big mistake. You have to be upfront and direct with people.”A second respondent used counseling with mixed success, sometimes ending inseparation from the public service, “In my experience one-on-one counseling anddiscussion regarding bad habits is how I dealt with managers’ bad habits. Managerstend not to be informed of their bad habits; this is a mechanism to inform them. Insome instances it helped and in others it did not help. In the instances it did not help, itwas already time to recommend a release of duty from his/her job function.” Regardingthe latter, a third manager also stated, “…counseling and discipline…I keep it asinformal as possible, but if there is no change in bad behavior then I start progressivediscipline.” A fourth seasoned administrator from the Southeast also uses this strategy,but with a twist:


“When the habit involves a manager who directly reports to me I intervenedirectly. If does not take long before word gets out that I came for an irregularvisit, it is known by the person with the bad habit and by those with whom he/sheworks that it was time for a talk. This gets their attention. Sometimes such amature adult conversation results in a contrite manager; sometimes it doesn’t.”

A fifth senior manager from a Western city also used feedback stressing the needfor positive reinforcement of good habits: “Complimenting on good habits isimportant: ‘You handled that well,’ and ‘I am proud of you using your good judgment.’”He cautions, however, that when dealing with bad habits “In my opinion yelling andbelittling never works. I don’t think I promote an office with yelling and screaming….This is definitely a bad strategy and should never be used. I try to discourage all mymanagers from using this strategy.”

Table 4 examines perceptions of organizational performance. Many respondentshave favorable assessments of their city’s performance. For example, 75.2% ofrespondents agree or strongly agree that they have a strong customer orientation, and64.3% of them agree or strongly agree that they use up-to-date information technologyapplications. In addition, slightly more than half, 58.6% agree or strongly agree thattheir organizational productivity is high.

The relationships among work habits, habit management strategies andperformance are best examined through a structural equation model. Though acomplex figure, the key study variables are shown in bold, and the key studyrelationships are shown by bold arrows. These relationships are discussed in theframework and as research questions: (i) the impact of good and bad managementwork habits on each other; (ii) the impact of good management work habits and badwork management habits on organizational performance, and (iii) the impact ofmanagement strategies to reduce bad management work habits. Figure 2 also showsfive (5) other variables that are included as control variables, whose purpose is to allowfor a more stringent test of the above key relationships. Methodologically, including arange of control variables also increase the model’s degrees of freedom, thereby


Table 4: Outcomes

Somewhat“In our city…” Agree1 Agree Disagree2 DK

Performance we have a strong customer orientation 75.2 17.6 5.3 1.9collaboration with community leaders is good 75.2 19.5 3.9 1.4organizational productivity is high 58.6 28.6 9.1 3.8we use up-to-date information technology applications 64.3 20.0 15.2 0.5we frequently develop new, innovative programs 58.1 28.6 11.4 1.9citizen trust in local government is high 47.6 29.5 18.1 4.8

Cronbach alpha of this index variable is 0.85.

1) Includes Strongly Agree and Agree 2) Includes Somewhat Disagree, Disagree and Strongly Disagree

improving estimation and providing a more rigorous test of model fit, albeit at theexpense of some model parsimony. In any event, Figure 2 is one of several models thatcan be shown, all of which produce similar substantive conclusions regarding the keyrelationships.

Regarding control variables, shown as thin-lined boxes in Figure 2, the literaturesuggests that cultures of entitlement negatively affect organizational performance byreducing enthusiasm and support for high performance.39 “Entitlement” is measuredby the survey item “In our city, people just act busy rather than doing meaningfulwork.” The adoption of codes of conduct is associated with management practices thatincrease professionalism and performance, such as more promptly responding tostakeholders and reducing unethical conduct.40 “Code of Conduct” is measured by “Inour city, we have a code of conduct.” Some studies suggest that smaller cities are lesslikely to engage in some of activities mentioned in Table 4 because they may lack theresources or management talent for doing so.41 City size is measured by populationsize. Some studies also suggest that jurisdictions with a city manager form ofgovernment are more professionalized and, hence, more concerned with aspects oforganizational performance mentioned in Table 4. We also test whether assessments ofbad management work habits vary between city managers and other seniorrespondents.42


Figure 2: Bad Work Habits Among Managers

Note: the standardized regression coefficients are shown along the paths.

FIGURE 2: IMPACTS OF MANAGING MANAGER’S WORK HABITS

= key study variable

= control variable

Note: the standardized regression coefficients are shown along the paths.

Good ManagementWork Habits Org. Performance e1

Bad ManagementWork Habits

Management, Policy &Solution Awareness

e2

e4 e3

p = .271 (df = 29)

Entitlement

e5

Code ofConduct-.25

.40

ProblemAwareness

e6

-.06

-.18

.60-.31

-.01 -.49

City Size

Resp.

CM

-.18

-.16

.23

.22

-.18

.07

.31

-.21

Chi-square = 33.167

Before discussing the substantive results of Figure 2, we note that this modelsatisfies the usual goodness-of-fit test assumptions for such causal models. Thevariance-covariance matrix is consistent with that of the data (Chi-square = 33.17, p >.05), the RMSEA is .03 (under the norm of .05), the Goodness of Fit Index is .956, theAdjusted Goodness of Fit Index is .917, which exceeds the threshold of .90, and themaximum Modification Index is 2.2 . The Tucker-Lewis Index is 0.93 and theComparative Fit Index is 0.967, indicating good incremental fit, as they are close to thenorm of 1.0. The Parsimony-Adjusted Normed Fit Index of .494 compares favorablywith competing models. Figure 2 shows the effect sizes (standardized coefficients) ofthe relationships. The dotted arrows show relationships that are not significant at thecustomary 5% or 1% levels, but they are included for reasons of theoretical interestmentioned above.

We now discuss each of the results of Figure 2 with reference to each of the abovekey relationships. Regarding relationship (i), the impact of perceived good and badwork habits on each other, the results show that bad management work habits have asignificant negative impact on good management work habits, but good managementwork habits are not statistically associated with reducing bad management work habits.Figure 2 also shows indirect effects of perceived bad management work habits onperceived good management work habits; the former are significantly associated withperceptions of increased entitlement orientations among managers and employees,which in turn are negatively associated with good management work habits. The resultis that the total standardized effect (shown in Table 5) of bad work habits on good workhabits is considerable, -.536 (bad habits → good habits), and that the total standardizedeffect of good work habits on bad work habits is quite small, -.052 (good habits → badhabits). Alternative model specifications do vary a bit in the magnitude of the latter, butit is never greater than about 35% of the impact of bad management work habits ongood management work habits, and this direct impact is never statistically significant.43

An important practical implication of this finding is that managers cannot rely on thepresence of good management work habits in order to reduce the presence of badmanagement work habits; additional efforts should be sought after, such as thosediscussed in connection with relationship (iii), below.

Regarding relationship (ii), the effects of management work habits onorganizational performance, Figure 2 shows that while good management work habitsdirectly affect organizational performance, much of the impact of bad managementwork habits occurs indirectly. While the direct standardized effects on organizationalperformance are respectively, .399 (good work habits → performance) and -.056 (badwork habits → performance), the total direct effects (which include direct and indirecteffects) are respectively .381 and -.344. These results show that bad habits aresignificantly, negatively associated with organizational performance, and theimpact of bad management work habits nearly offsets performance gains from goodmanagement work habits; a practical implication is that managers will want to take thepresence of bad management work habits seriously. These results are consistent withthe simple bivariate analysis, too: the Pearson correlation coefficient of good


management work habits with organizational performance is r = .357 (p < .01), and ofbad management work habits with organizational performance, r = -.250 (p < .01).

A variety of interview comments support that the impacts of bad managementwork habits on organizational performance are often indirect. For example, a citymanager from a city on the Midwest observes: “Attitude works its way down the ladder,”and a California manager states, “Persistent bad habits can be problematic becausemanagers set the tone in the administration. It is likely to spread to employees.”Another stated: “Often bad habits revolve around comings and goings. This causes


Table 5: Standardized Total And Direct Effects Of RelationshipsShown In Figure 2

STANDARDIZED TOTAL EFFECTSMgt.

Policy &City Code Of Respon- Problem Bad Solution Good Org.Size Conduct CM dent Aware- Work Aware- Work Entitle- Perfor-

ness habits ness Habits ment mance

Problem Awareness -.008 -.024 -.025 -.015 -.030 .096 .584 -.147 .059 -.107

Bad Work habits .002 .007 .008 -.151 -.302 -.033 -.182 .052 -.020 .033

Mgt., Policy &Solution Aware. -.013 -.039 -.042 -.025 -.050 .159 -.030 -.244 .097 -.178

Good Work Habits -.001 -.004 -.004 .084 .167 -.536 .101 -.029 -.203 -.018

Entitlement .001 .002 .002 -.047 -.093 .300 -.056 .016 -.006 .010

Org. Performance .072 .218 .231 .054 .107 -.344 .065 .381 -.332 -.012

STANDARDIZED DIRECT EFFECTSMgt.

Policy &City Code Of Respon- Problem Bad Solution Good Org.Size Conduct CM dent Aware- Work Aware- Work Entitle- Perfor-

ness habits ness Habits ment mance

Problem Awareness .000 .000 .000 .000 .000 .000 .602 .000 .000 .000

Bad Workhabits .000 .000 .000 -.156 -.312 .000 .000 .006 .000 .000

Mgt., Policy &Solution Aware. .000 .000 .000 .000 .000 .000 .000 -.179 .000 -,184

Good Work Habits .000 .000 .000 .000 .000 -.488 .000 .000 -.214 .000

Entitlement .000 .000 .000 .000 .000 .310 .000 .000 .000 .000

Org. Performance .073 .220 .234 .000 .000 -.056 .000 .399 -.253 .000

Note: Total standardized effects = Direct standardized effects + Indirect standardized effects.Hence, the indirect standardized effects are readily inferred from the above tables.

inequities because those who don’t arrive late or leave early resent the behavior ofpeople who take advantage of lax enforcement. It can become a morale issue in yourunit. There is a consequence to the organization.” A senior manager in a Western cityechoes this: “Managers will be imitated by their employees. If a manager continuouslycomes into work late, employees will imitate their manager and come into work later.This will impact productivity in the office.”44 These comments show indirect effects onperformance.

What is the impact of effort to manage bad management work habits (relationshipiii, above)? Figure 2 confirms the results discussed in connection with Table 3, above,namely that only efforts to increase awareness about bad management work habitsare directly, statistically associated with reducing perceived, bad management workhabits. Efforts that exhort managers to adopt good work habits, even with reinforcingpolicies (composed of the relevant items of Table 3, alpha =.787, ), are not directlyassociated with reducing bad management work habits, though they do support them.The total standardized effect of increasing problem awareness on bad managementwork habits is -.302, and the total standardized effect of the aggregate index variable ofexhorting good work habits (with reinforcing policies) on bad management workhabits is -.182. A plausible explanation for the sequencing of these two variables in themodel is that, in practice, managers may begin their change efforts targeting managerswith bad habits by exhorting them that they ought to adopt better work habits, only tooften realize that sustained efforts are also needed to obtain awareness and acceptanceamong managers with bad work habits that such work habits are indeed bad andtherefore must come to an end.45

Beyond these key relationships, Figure 2 also shows other results worthmentioning. Entitlement attitudes significantly reduce perceptions of goodmanagement work habits and organizational performance; the respective totalstandardized effects are -.203 and -.332. Such negative impacts support numerousstudies about the detrimental effect of entitlement attitudes in organization; this studyadds to such findings that reducing bad management work habits can reduceentitlement attitudes, too. Figure 2 also shows that the presence of good managementwork habits and organizational performance are negatively associated with efforts toexhort managers into adopting good management work habits (with reinforcingpolicies).46 Such negative relationships make sense, of course (why increase attentionon work habits when management work habits and performance are perceived aspositive?), but it also suggests the possibility of a vicious cycle; when things are goingwell, the possibility exists of giving insufficient attention to bad management workhabits which, however minimal, sow the seeds of subsequent performance problems.Finally, regarding other control variables, the type of respondent, city manager or othersenior manager, is also not significant in this study. Alternative specifications, considerimpacts on other key study variables, are also not significant.47 Figure 2 also shows thateffects of the city manager form of government and codes of conduct are positivelyassociated with increased organizational performance, consistent with abovementioned studies. City size is not significant, reflecting that really small cities, that maylack resources, are not part of this sample which includes only cities over 50,000.


Conclusion and DiscussionThis article finds that managerial work habits affect organizational performance – goodmanagerial work habits raise performance, and bad managerial work habits reduce it.Bad management work habits are significantly present in about 45% of cities, and mostcommonly involve managers being overly passive, judgmental, defensive, intimidating,closed-minded or tardy in their performance. While these behaviors vary amongsettings, the presence of just a few managers with bad work habits significantly,negatively affects both organizational performance and the presence of goodmanagement work habits. While good management work habits are widely reported,such as being pro- active and setting high standards, bad management work habitsalmost cancel out gains from good management work habits, and are therefore arelevant concern. This study also finds that a key step to reducing bad managementwork habits is to promote awareness and acceptance among managers with such badwork habits that these habits are indeed occurring, that they are undesirable, and thatthey need to subside. Bad management work habits cannot be overcome by merelyexhorting such managers to change their work habits, even when such efforts arereinforced by appropriate policies and evaluation criteria.

This study contributes to the literature in several ways. First, it augments theliterature on managerial professionalism by examining the nature and extent of good aswell as bad management work habits. Intimidation, closed-mindedness and tardinessare serious matters that are shown to be systematic, and regularly occurringphenomena of management, rather than the aberrant issues of isolated individualmanagers, as they are sometimes portrayed. Both good and bad work habits exist inabundance. While studies of professionalism and ethics already consider unethical andillegal behavior as detrimental to performance and stakeholder relations, this studyshows that bad management work habits should also be included among thosebehaviors to be arrested. Results here suggest that at some point progress towardprofessionalism and organizational performance is significantly hindered, if not stifled,by the presence of sometimes just a few managers with significantly bad work habits.

Second, this study contributes to literatures that discuss the importance ofinformal and on-going supervisory management practices. Studies of 360-degreefeedback and psychological contracts suggest in different ways the need to providemore feedback than that merely given through annual or bi-annual performanceappraisal and to provide consequences for work habits through promotion, reward anddiscipline processes. Findings from both the survey and interviews show thesignificance of frequent and repeated feedback from top managers to subordinatemanagers and supervisors about their work habits. The fact that managers often worktogether and frequently talk with each other does not mean that they talk about badwork habits, too. It is time to start thinking about standards and better practices formanaging the managers in pursuit of organizational effectiveness.

This study also has many practical implications. First, it tantalizingly provides anexplanation for the failure of many management training exercises to improveperformance and address bad work habits. This study finds that such habits require


active management, not merely training seminars. A distinguishing feature of “highperformance” organizations is that managers talk about ‘what is really going on,’ andare thus able to benefit more from training seminars and the like.48 Second, awarenessof bad habits is key, but many managers are not well-trained to identify various ways inwhich bad work habits may occur. At the very least, they need some checklists,distinctions and rating criteria, but the complexity becomes clear when consideringthat some managers have deep-seated problems, and that these managers may be quiteadept at camouflaging their bad work habits, at least for a time (below). Third,managers need better training and support regarding discipline. Many managers finddiscipline and termination to be among their most difficult tasks. The literature andpractice suggests that more support and training on this topic would be helpful ifmanagers are going to be asked to address the bad habits of those that report to them.

Finally, the consequences of bad habit management may well be more than thoseexamined in this study. For example, workplace stress is exacerbated by the bad habitsof managers. Many times, managers think that intimidating their employees andplacing more demands on them somehow guarantees a more productive workenvironment, but stress in the workplace can lead to higher turnover rates, high levelsof absenteeism, and overall poor work performance. Bad habits can also lead to conflictamong managers and staff. For example, when managers fail to provide support,assistance or understanding to those under their supervision, such habit-based workpractices can induce stress, ignite conflict and harm performance. Similarly, managerswho habitually fail to communicate sufficiently or effectively, who are unwilling todelegate tasks or responsibilities, or who have poor social skills, are a hindrance tosubordinates in doing their job, which adds to their stress. Intervention strategies areneeded in such cases, whether through formal mentoring or leadership developmentprograms, to address poor workplace habits that heighten stress, foment conflict anddampen productivity. As one astute manager told us, “If an organizational cultureallows bad habits to continue, it is a deficiency of leadership, especially if this has anegative impact on organizational performance.”

Future research needs to begin by acknowledging that systems-basedimprovement and reforms are dependent on those who execute them. By virtue oftheir position and power, the judgments of managers matter. We need to designsystems that better account for the behaviors and orientations that managers mayexhibit. Regarding bad management habits, achieving measurable improvements iscomplex, but successful change is possible. Future research needs to emphasize theimportance of awareness and communication. Open communication allows for topmanagement and employees to give corrective feedback to managers with bad habits.Case studies, field reports, survey-based research and other quantitative or qualitativemethodologies can be used to further identify the ways in which habit-based workpractices impact productivity or stress in the government workplace and themanagerial strategies which successfully reinforce positive work habits and correctnegative habits. Achieving a better understanding of the role of workplace habits iscrucial if we are to achieve the robust, high performing public organizations—alert,nimble, flexible, and adaptable--extolled in the academic and professional literature.


Notes1 Berman, E. & West, J. (2003)a. What is managerial mediocrity? Definition, prevalence, andnegative impact (Part 1). Public Performance & Management Review 27, 9-29; Light, P. (1999).Does management matter? Government Executive 31, 6; Light, P. (2005). The four pillars ofhigh performance: How robust organizations achieve extraordinary results. NY: McGraw-Hill;Nutt, P. (2002). Why decisions fail. San Francisco, CA: Berrett-Koehler; Field, T. (2002). Theserial bully. www.bullyonline.worldwidebully/serial.htm (Retrieved 12/14/06).

2 Sutton, R. (2007). The no asshole rule. NY: Warner Business Books.

3 The importance of workplace habits among organization leaders was recently documented inWelbourne’s study of more than 300 private-sector executives and senior managers. Whenasked to rate 15 items in terms of the degree to which each might derail execution oforganizational strategy, the number one barrier was the organization’s past and habits. Closebehind in the rating were two factors linked to an organization’s past and habits: organizationalculture and the way employees work together. Getting bogged down in routines is not uniqueto the private sector, of course. Welbourne, T. (2005). Leaders talk about executing strategy.Leadership pulse survey results. (March). www.umbs.leadership.eepulse.com (Retrieved May 1,2006).

4 Stinson, E. & Mueller, D. (1980). Survey of health professionals’ information habits and needs.conducted through personal interviews. JAMA January 11, 243, 2: 140-143.

5 Berman, E. (2006). Performance and productivity in public and nonprofit organizations.Armonk, NY: M.E. Sharpe.

6 Salamon, L. (2002). The tools of government: A guide to the new governance. Oxford, UK:Oxford University Press; Kettl, D. (2002). The transformation of governance. Baltimore, MD:Johns Hopkins University Press.

7 Nutt, P. (2004). Prompting the transformation of public organizations. Public Performance &Management Review 27, 9-33; Ammons, D. (2004). Productivity barriers in the public sector. InM. Holzer and S-H. Lee, eds. Public productivity handbook, 2nd ed. NY: Marcel Dekker, 139-164; Oreg, S. (2003). Resistance to change: Developing an individual differences measure.Journal of Applied Psychology 88, 680-693.

8 Rousseau, D. (1995). Psychological contracts: Understanding written and unwritten contracts.Thousand Oaks, CA: Sage; Berman, E. & West, J. (2003)b. Psychological contracts in localgovernment: A preliminary analysis. Review of Public Personnel Administration 23, 267-285;Guest, D. & Conway, N. (2002). Communicating the psychological contract: An employerperspective. Human Resource Management Journal 12, 22-38.

9 A close alternate for the phrase “work habit” is “work practice.” However, the term ‘practice’also involves chosen patterns of behaviors, sometimes in accordance with group norms (e.g.,‘professional practices’). However, bad management work habits are typically unprofessional,based on individual rather than group norms, and not always reflecting conscious choice.Hence, to avoid any such confusion, we use the phrase work habit. Covey even refers to “beingpro-active” or “thinking win-win” as habits to be developed. This use of the word habit ismisleading, as it involves mindsets, only. However, it, too, shows that ‘habit’ rather than‘habitude’ has become the prevailing usage. Waldroop, J. & Butler, T. (2000)a. Managing awaybad habits. Harvard Business Review 78, 89-98; Covey, S. (1990). The 7 habits of highly effectivepeople. NY: Simon & Schuster.

10 Ohly, S., Sonnentag, S. & Pluntke, F. (2006). Routinization, work characteristics and theirrelationship with creative and proactive behaviors. Journal of Organizational Behavior 27,257-279.

11 Beatty, S. & Kahle, L. (1988). Alternative hierarchies of the attitude-behavior relationship: Theimpact of brand commitment and habit. Journal of the Academy of Marketing Science 16, 1-10;


Kahle, L. & Beatty, S. (1987). The task situation and habit in the attitude-behavior relationship:A social adaptation view. Journal of Social Behavior and Personality 2, 219-232.

12 As noted by one anonymous reviewer, work habits represent a mélange of behaviors, practices,predispositions, proclivities, etc. This, however, does not suggest any inaccuracy of definition,but rather a plethora of synonyms for the phenomena under study.

13 As a further example, smoking is statistically associated with numerous illnesses, and is nowwidely regarded as a bad practice, but, in the moment, smoking also reduces stress which is apositive or productive outcome. Only when practices are seen as, on balance, problematic ordeficient in some way, may reflection and analysis occur, possibly leading to modification.

14 Holzer, M. & S. H. Lee, eds. (2004). Op. cit.

15 Another line of inquiry identifies orientations and behaviors of exceptionally dysfunctionalmanagers, such as those who are highly abusive, grandiose, controlling or antisocial. These aresignificant manifestations of the unproductive orientations and behaviors noted in the text.Finkelstein, S. (2004). The seven habits of spectacularly unsuccessful executives. Ivey BusinessJournal Online (January/February): 1; Lubit, R. (2004). The tyranny of toxic managers: Applyingemotional intelligence to deal with difficult personalities. Ivey Business Journal 68, 1-7;Waldroop, J. & Butler, T. (2000)a. Op. Cit.

16 Waldroop, J. & Butler, T. (2000)b. The 12 bad habits that hold good people back. NY:Doubleday.

17 A wealth of practitioner literature exists in business management, largely non-empiricalcommentaries, identifying critical habits in information technology, safety administration,manufacturing, management, negotiations, financial services, insurance, and human resources.A few practitioner-oriented articles in public administration focus on developing effective habitsof elected city councils, and on habits of public managers in the United Kingdom, but nonedeal with senior managers in U.S. local governments. Datz, T. (2003). 6 habits of highly effectiveCIOs. CIO 16, 17 (June 15): 72; Kincaid, W. (1996). Habits of effective safety managers.Occupational Hazards. (November): 41-43; Pollock, T. (1997). July. Bad habits that may beholding you back. Automotive Manufacturing & Production. 109, 13-14; Haddock, P. (1997).Secrets for getting things done: Develop work habits that stress success, Not Ineffectiveness.Incentive. 171, 120-121; Barrier, M. (1999). Discovering the habits of well-regarded CEOs.Nations Business 87, 3-15; Finkelstein, S. (2004). Op. cit.; Sebenius, J. (2001). Six habits ofmerely effective negotiators. Harvard Business Review 79, 87-95; Leeds, D. (1996). How towork smarter, not harder. National Underwriter Life & Health-Financial Services Edition. 41,21; Van Zijl, C. (1999). Sharpening up social skills. Credit Management (November): 22-23;Groner, D. (1997). The habits of highly ethical insurance professionals. National UnderwriterLife & Health-Financial Services Edition. 101, 19, 21; Rosner, B. (1999). Building trust is a goodHR habit. Workforce. 78, 25-7; Neu, C. (1997). 10 habits of highly effective councils. PublicManagement. 79, 4-9; Vogelsang-Coombs, V. (1997). Governance education: Helping citycouncils learn: Public Administration Review 57, 490-500; Brookfield, D. (2000). Managementstyles in the public sector. Management Decision 38, 13-18.

18 This is yet another likely reason for being skeptical of many consulting training efforts as thebasis for achieving lasting change. It is anecdotally observed that training alone seldom suffices.

19 Ashworth, K. (2001). Caught between the dog and the fireplug, or How to survive in publicservice. Washington, D.C.: Georgetown University Press; Berman, E. & West, J. (2003)a. Op. cit.

20 This can include mental health or alcohol/substance. While the latter are presumably infrequentamong managers, the discussion about standards suggests that even just a few of such instanceswill have negative impacts on organizational performance.

21 McDonald, T. (1998). Eye-opener—New circumstances require new ways of thinking. SuccessfulMeetings (July): 20.

22 Lewin, K. (1947). Frontiers in group dynamics. Human Relations 1:5-41.


23 These four steps are similar to the more comprehensive learning program initiated at GeneralElectric by Jack Welch who was concerned about bad habits in the company. GE’s ChangeAcceleration Process (CAP) focused on “making change stick” and included a seven-step modelof change: leading change (requires a committed champion), creating a shared need(communicates the rationale for change), shaping a vision (conveys desired behavioraloutcomes), mobilizing commitment (building support), making change last (developing shortplus long-term plans), monitoring progress (installing metrics), and changing systems orstructures (reinforcing change). These various step-by-step approaches can help to guide thehabit management initiatives of organizations seeking improved performance. Vogelsang-Coombs, V. (1997). Op. cit.; Garvin, D. (2000). Learning in action: A guide to putting thelearning organization to work. Boston, MA: Harvard Business School Press.

24 Berman, E. (2006). Performance and productivity in public and nonprofit organizations.Armonk, NY: M.E. Sharpe.

25 Waldroop, J. & Butler, T. (2000)a. Op. cit.; Waldroop, J. & Butler, T. (2000)b. Op. cit.

26 Buckingham, M. & Coffman, C. (1999). First, break all the rules. NY: Simon-Schuster; Guy, M.(2004). The human side of productive work environments. In M. Holzer and S-H. Lee, Op. cit.,pp. 447-462; Olshfski, D. & Levine, H. (2004). Organizational change and innovation. In M.Holzer and S-H. Lee, Op. cit., pp. 593-609; Shafritz, J., Rosenbloom, D., Riccucci, N., Naff, K., &Hyde, A. (2004). Personnel management in government. 5th ed. NY: Marcel Dekker chapter 8;Van Wart, M. (2004). Training and development for productivity. In M. Holzer & S-H. Lee, Op.cit., 529-549; Van Wart, M. (2005). Dynamics of leadership in public service. NY: M.E. Sharpe.

27 ICMA. (2005). Municipal yearbook 2005. Washington, D.C.: International City/CountyManagement Association.

28 In practice, it is nearly impossible to achieve higher response rates for surveys of this lengthwhich are necessary for providing the kind of detailed study that this is. The challenge for anyrealistic response rate is to show that the sample is representative of the sampling frame. Tothis end, we use two approaches that have now become standard in the discipline. We comparethe sample against the sampling frame, and compare a sample of non-respondents againstrespondents. The results of these tests are reported below, and we conclude that the surveyresults can be generalized to the study population. We are very concerned that some mightjudge this matter solely by response rate; even surveys with, say, a 60% response rate can havesubstantial sample bias; the purpose of the above methods is to assess whether such bias exists.Hays, S. & Kearney, R. (2001). Anticipated changes in human resource management: Views fromthe field. Public Administration Review 61, 485-597; Berman, E. & Korosec, R. (2005). Planningto coordinate and coordinating the plan: Evidence from local governments. American Reviewof Public Administration 35, 380-401; Sheehan, K. (2001). E-mail survey response rates: Areview. Journal of Computer-Mediated Communication 6, 2.http:jcmc.indiana.edu/vol6/issue2/Sheehan.html (Retrieved on 4/9/07).

29 Specifically, the survey sample is broadly representative of the sampling frame, which is a listingof all U.S. cities with populations over 50,000. As is common, survey participation is a bit lowerin the Northeast than elsewhere in the country.

30 We also think that asking others to report on the bad habits of others is more accurate thanasking people to report on their bad habits which, as noted in the Framework section, peopleare apt to be at least somewhat in denial about.

31 While this study takes great pains to further measurement validity, we acknowledge thelimitations of the data as noted in the introduction. On the one hand, the estimates of mangersmay be considered as a conservative estimate of bad habits because (i) some bad habits ofsubordinate managers and supervisors may only be revealed to their employees, and not totheir superiors, who are the respondents in this study, and (ii) respondents do not have aroundthe clock access to subjects as a basis for their observations. On the other hand, respondentsare subject to Fundamental Attribution Error Theory, which is the tendency to explain the


behavior of others on the basis of disposition or character, rather than context or the actions ofthird parties. This may cause some overestimation of bad habits. While the net effect of thesedifferent considerations cannot be known with certainty, of course, we note that no data insocial science are ever known in ways that are completely free from all possible distortions,errors and biases. We caveat our perceptual data for these possibilities, as is conventionalpractice.

32 A sample of four randomly selected items was used. These items, along with tau-c and p-values for testing significant differences between respondents of the mail survey and theabbreviated phone survey among non-respondents, are: “I have challenged the manager tochange work habits” (tau-c = .020, p=.635), “set explicit goals for new work habits managersneed to achieve” (tau-c= .080, p=.101), “provided a coach to work one-on-one to support newwork habits” (tau-c= .020, p==.653) and “included work habits as a criterion in performanceappraisal” (tau-c=.079, p=.097). Demographic variables of phone survey respondents werecollected, too, but these do not significantly affect the above relationships.

33 Note that the item lead-in is worded differently in Tables 1 and 2. Table 1 concerns good habitsamong at least one manager, whereas Table 2 concerns bad habits among managers that reportto the respondent. This is appropriate, because, as noted in the framework, standards forevaluating good and bad habits differ. The prevalence of just a few managers with bad habitscan have a strong impact on senior managers’ team, whereas good habits are widely seen asrequired among all or most members for the purpose of high performance.

34 In fact, very few disagree to any extent with these statements. The percentage of those whostrongly agree, agree or somewhat agree is at least 95% for each of these items.

35 The categories are based on theoretical considerations, and are supported by measures ofinternal reliability (alpha) shown in Table 2. Construct validity is evidenced as follows. Forexample, the index of ideas & information is associated with such items as “makingconsequential decisions unilaterally” (tau-c=.358, p < .01), the index of collaboration isassociated with “being uncomfortable in social situations” (tau-c=.415, p <.01), and the indexof self-management is correlated with “managers are not well attuned to their feelings” (tau-c=.278, p <.01). While none of these individual items constitute aspects of the index measures,they do correlate with them as might be expected, hence, providing construct validity to themeasures.

36 Also, among the 340 possible pairs of correlations between items in Tables 1 and 2, 270 pairsare statistically significant.

37 If one (1) somewhat agree is included as constituting a jurisdiction in which managers have badhabits (including of course, any cities that agree or strongly agree), then an additional 11.5%should be added for a total of 74.3% of cities. If three (3) somewhat agree are required, then7.2% should be subtracted, yielding 55.6% of cities. Beyond this, different definitions can alsoshift percentages among categories, of course, but sensitivity analyses do not suggest changesby more than about 6%.

38 A variety of statistical tests all support these conclusions. Here, ANOVA is used, as thedependent variable is continuous, and the independent variables are categorical with three ormore categories. Df = 160 to 164.

39 West, J. and Berman, E. (1997). Administrative creativity in local government. PublicProductivity & Management Review 20, 446-457.

40 While by themselves codes of conduct are unlikely to affect performance, they are thought tobe part of other strategies which collectively do aim to further performance. Here, they areviewed as an indicator of such broader efforts. This, too, is reflected in the survey data. Forexample, having a code of conduct is positively associated with the statement “unethicalconduct is dealt with harshly around here” (tau-c= .168, p <.01), which in turn is associatedwith the measure of performance (tau-c=.249, p <.010).


41 West, J. & Berman, E. (1997). Op. cit.

42 While the literature is silent on this matter, city managers might have a more positiveassessment because their role includes strong elements of marketing the city and its staff. Thevariables city size and manager-council form of government were gathered from the ICMA.2005. Municipal yearbook 2005. Washington, D.C.: International City/County ManagementAssociation.

43 In an alternative model specification, excluding all of the above control variables, the totalstandardized effects of these relationships are, respectively -.426 and -.140, which supports thetext conclusions. However, this alternate model has only three degrees of freedom and istherefore not shown.

44 Another provided an example of this: “We had one manager who was here for many years, butdid not adapt well to the changing conditions in the city. All city employees know that theyshould stay at work until 5:00 p.m., but he would leave early routinely. Others felt he wasdoing it, so why couldn’t they leave when they wanted to as well. Eventually everyone in theunit was leaving early. To deal with this problem the manager was reassigned to a job where hehad a flexible schedule and the rest of the staff returned to working the approved schedule.”

45 This explanation is consistent with a rational approach to managing that sequences subsequentefforts, starting with strategies of least effort, to those that involve more effort. Many medicaltreatments follow the same strategy, starting with interventions that are least invasive.

46 While these relationships are not statistically significant in Figure 2, bivariate results aresignificant. Good management work habits are associated with exhorting good managementwork habits, r = -.23, p < .01, as is organizational performance, r = -.24, p <. 01). This seemsto be a good reason for bringing this point up here.

47 We also examined whether the age or type of respondent is associated with perceptions oforganizational performance, but find no such relationship.

48 Kearney, R. & Berman, E., eds. (1999). Public sector performance. Boulder, CO: WestviewPress; Holzer, M. & S-H. Lee, (2004). Op. cit.; Cohen, S. & Eimicke, W. (2002). The new effectivepublic manager. San Francisco, CA: Jossey-Bass.


AuthorsJonathan P. West, PhD

Department of Political ScienceP.O. Box 248047University of MiamiCoral Gables, FL 33124(305) [email protected]

Professor Evan M. Berman, PhD

National Chengchi UniversityInternational Doctoral Program in Asia-Pacific StudiesGeneral Building of Colleges, 12th Floor, North WingNo. 64 (Section 2) ZhiNan Road, Wenshan DistrictTaipei [email protected]

Dr. Jonathan P. West is a professor of political science and director of the graduate pub-lic administration program at the University of Miami. He has published eight books andmore than 100 scholarly articles and book chapters. He teaches undergraduate and gradu-ate courses in American politics, public management, and human resource managementand is managing editor of Public Integrity journal.

Dr. Evan M. Berman is the distinguished university professor at the National ChengchiUniversity, in Taipei, Taiwan. He has published eight books and more than 100 publica-tions, many of which appeared in the leading journals of the discipline. Berman teachesHRM and quantitative methods, and has interests in productivity. His recent booksinclude HRM in Public Service (2010, 3/e, with West, Bowman and Van Wart), PublicAdministration in East Asia: China, South Korea, Japan and Taiwan (2010), and Peo-ple Skills at Work (forthcoming with Dira Berman).


2011 IPMA-HR

INTERNATIONAL TRAINING CONFERENCE & EXPO

September 24-28, 2011Marriott Downtown Chicago Magnificent Mile Hotel | Chicago, IL

International Public Management Association for Human Resources

The premier

event of the

year for public

sector HR

professionals!

Register Today! www.ipma-hr.org or call 703-549-7100 to learn more.

Conference Highlights:� Choose from more than 35 educational sessions

presented by leaders in the field of Human Resources.

� See the latest cutting-edge products and services for thepublic sector HR professional at the conference expo featuring employee benefit packages, professional development programs, training programs and materials,and much more.

� Earn up to eight points toward your IPMA-CP or IPMA-CSrecertification.

� Choose from multiple full-day pre-conference workshopsproviding an in-depth look at the leading topics and issues important to public sector human resource professionals.

IPMA2011_Full Page Ad:Layout 1 1/21/11 9:54 AM Page 1

DATE: December 20, 2006Questions about electronic art:

Amy DaglianoClayton Design Group7702 Leesburg Pike, Suite 300Falls Church, VA 22043

PHONE: 703-448-1920FAX: 703-448-3820EMAIL: [email protected]

PUBLICATION: IPMA-HR News

FILE NAME: IPMA NewsAd FP.pdf

FILE TYPE: Press-ready pdf

AD SIZE: Half-page, 9" x 12", no bleed

COLORS: 2C: K and PMS 653

CONTACT: Carrie Hoover, 703-549-7100

Professional Development Courses for Public Sector HR Professionals

Bring Public Sector HR Professional Development to Your Agency!

Trusted professional development designed specifically for public sector HR professionals that:• Develops a broader range

of competencies • Enhances job performance • Improves quality

• Elevates professionalism• Instills pride • Advances the HR profession

Visit IPMA-HR at www.ipma-hr.org or call 703-549-7100 to schedule your professional development course today!

2688 IPMA News Ad.indd 1 12/20/06 2:23:15 PM

Someone who will burn bright. Let us illuminate your path to �nding the brightest candidates. With high quality tests at low per-candidate costs, we can help meet your agency’s hiring and promotional needs every step of the way. To �nd out how our tests, publications and assessment tools can guide you, visit us on the web at www.ipma-hr.org/assessment.

www.ipma-hr.org | 800.381.TEST

public personnel management - ipma-mc · public personnel management subscription order u.s. -...

Documents