employee voice: does union membership matter?

20
Employee voice: does union membership matter? John Benson, School of Management, University of South Australia Michelle Brown, Department of Management and Marketing, University of Melbourne Human Resource Management Journal, Vol 20, no 1, 2010, pages 80–99 Industrial relations research has traditionally viewed trade unions as the primary mechanism for employee voice. With the decline in unionism in many advanced industrial economies over the past two decades, new direct non-union voice mechanisms have been introduced by employers. This focus on the mechanisms for employee voice, however, fails to take account of employees’ perceptions of voice. We suggested that employee perceptions of voice vary between the different levels of an organisation and proposed that trade union membership will be more likely to enhance individual employee perceptions of voice at the wider organisational level. Contrary to our expectations, our analysis of 2,949 employees of a public sector scientific research organisation found that union membership had a significant negative impact on employee voice at the organisational level. The article concluded by offering possible explanations for this unexpected finding and the implications for theory, management and future research. Contact: John Benson, School of Management, University of South Australia, North Terrace, Adelaide, SA 5001, Australia. Email: [email protected]INTRODUCTION H istorically, trade unions have been seen as the major way to provide workers with ‘a channel to voice their grievances and to ensure due process in the workplace’ (Gospel and Wood, 2003: 2; see also Lewin and Mitchell, 1992; Haynes et al., 2005; Freeman et al., 2007). This ‘collective’ or union form of employee voice, it has been argued, enables employees to transmit their concerns to management, something they are less likely to do as individuals as they could be penalised for expressing grievances (Freeman and Medoff, 1984). Over the past two decades, the collective union voice view has been challenged as research has broadened to include direct voice mechanisms within a variety of non-union settings (McCabe and Lewin, 1992; McLoughlin and Gourlay, 1994; Terry, 1999; Benson, 2000; Gollan, 2003, 2006; Butler, 2005; Dietz et al., 2005; Dundon et al., 2005; Haynes, 2005; Machin and Wood, 2005; Taras and Kaufman, 2006; Bryson and Freeman, 2007; Dundon and Gollan, 2007). Contemporary research findings contest the accepted wisdom in the industrial relations literature that unions are the primary mechanism of employee voice through their representative role (Freeman and Rogers, 1993; Lansbury et al., 1996; Kaufman and Taras, 1999; Bryson and Freeman, 2007). As Addison and Belfield (2004: 564) argued, the collective voice model is deficient for ‘uncritically equating collective voice with autonomous unionism’. While a small, although growing, body of research has emerged that focuses on individual perceptions of voice (Withey and Cooper, 1989; Leck and Saunders, 1992; Luchak, 2003; Avery and Quiñones, 2004; Bryson, 2004), little attempt has been made to understand how trade union membership impacts these perceptions. doi: 10.1111/j.1748-8583.2009.00116.x HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 80 © 2009 Blackwell Publishing Ltd. Please cite this article as: Benson, J. and Brown, M. (2010) ‘Employee voice: does union membership matter?’. Human Resource Management Journal, 20: 1, 80–99.

Upload: john-benson

Post on 23-Jul-2016

217 views

Category:

Documents


5 download

TRANSCRIPT

Page 1: Employee voice: does union membership matter?

Employee voice: does union membership matter?

John Benson, School of Management, University of South AustraliaMichelle Brown, Department of Management and Marketing, University ofMelbourneHuman Resource Management Journal, Vol 20, no 1, 2010, pages 80–99

Industrial relations research has traditionally viewed trade unions as the primary mechanism foremployee voice. With the decline in unionism in many advanced industrial economies over the past twodecades, new direct non-union voice mechanisms have been introduced by employers. This focus on themechanisms for employee voice, however, fails to take account of employees’ perceptions of voice. Wesuggested that employee perceptions of voice vary between the different levels of an organisation andproposed that trade union membership will be more likely to enhance individual employee perceptions ofvoice at the wider organisational level. Contrary to our expectations, our analysis of 2,949 employees ofa public sector scientific research organisation found that union membership had a significant negativeimpact on employee voice at the organisational level. The article concluded by offering possibleexplanations for this unexpected finding and the implications for theory, management and futureresearch.Contact: John Benson, School of Management, University of South Australia, North Terrace,Adelaide, SA 5001, Australia. Email: [email protected]_116 80..99

INTRODUCTION

Historically, trade unions have been seen as the major way to provide workers with ‘achannel to voice their grievances and to ensure due process in the workplace’ (Gospeland Wood, 2003: 2; see also Lewin and Mitchell, 1992; Haynes et al., 2005; Freeman et al.,

2007). This ‘collective’ or union form of employee voice, it has been argued, enables employeesto transmit their concerns to management, something they are less likely to do as individualsas they could be penalised for expressing grievances (Freeman and Medoff, 1984). Over the pasttwo decades, the collective union voice view has been challenged as research has broadened toinclude direct voice mechanisms within a variety of non-union settings (McCabe and Lewin,1992; McLoughlin and Gourlay, 1994; Terry, 1999; Benson, 2000; Gollan, 2003, 2006; Butler, 2005;Dietz et al., 2005; Dundon et al., 2005; Haynes, 2005; Machin and Wood, 2005; Taras andKaufman, 2006; Bryson and Freeman, 2007; Dundon and Gollan, 2007).

Contemporary research findings contest the accepted wisdom in the industrial relationsliterature that unions are the primary mechanism of employee voice through theirrepresentative role (Freeman and Rogers, 1993; Lansbury et al., 1996; Kaufman and Taras, 1999;Bryson and Freeman, 2007). As Addison and Belfield (2004: 564) argued, the collective voicemodel is deficient for ‘uncritically equating collective voice with autonomous unionism’. Whilea small, although growing, body of research has emerged that focuses on individualperceptions of voice (Withey and Cooper, 1989; Leck and Saunders, 1992; Luchak, 2003; Averyand Quiñones, 2004; Bryson, 2004), little attempt has been made to understand how trade unionmembership impacts these perceptions.

doi: 10.1111/j.1748-8583.2009.00116.x

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201080

© 2009 Blackwell Publishing Ltd.

Please cite this article as: Benson, J. and Brown, M. (2010) ‘Employee voice: does union membership matter?’. Human Resource ManagementJournal, 20: 1, 80–99.

Page 2: Employee voice: does union membership matter?

This article addresses this deficiency by focusing on the question ‘does union membershipenhance employee perceptions of voice?’ The answer to this question has importantimplications for voice theory and for practising managers as it is ultimately an employee’sperception of their work environment that will influence their workplace behaviour (Fishbeinand Ajzen, 1975; Ajzen and Fishbein, 1977). The article commences with a discussion of theindustrial relations’ view of collective union voice and the literature on alternative forms ofdirect non-union voice. We argue that past research has failed to distinguish between employeeperceptions of voice and the mechanisms that offer employees the opportunity to express voice.This analysis provides the point of departure in the conceptualisation of voice from anemployee’s perspective. A measure of employee perception of voice is then developed andtested using data from a large survey of employees in a public sector scientific researchorganisation. This measure is then used to assess trade unions’ contribution to employeeperceptions of voice.

UNIONS AND EMPLOYEE VOICE

Unions in developed economies have a primary aim of maintaining and improving the termsand conditions of employment of their members (Crouch, 1982). To achieve this aim, unionshave mobilised their resources at the local, national and, in some cases, the international level.As a consequence, unions, through their collective representation role (Webb and Webb, 1897),have often been able to provide a wage premium and better working conditions for theirmembers, influence the wider distribution of income within society and provide support forworkers in developing nations (Flanders, 1952; Bloom and Northrup, 1965).

This focus on the collective representation role has meant that the industrial relationsliterature has generally equated employee voice with unions (see, for example, Freeman et al.,2007). A collective voice clearly has value to employees but as Freeman and Medoff (1984)contended, voice through unions may also have value for employers. Their argument, whichdrew on the work of Hirschman (1971), is based on the notion that unions have two faces: themonopoly face associated with their monopolistic power to raise wages and the collective voiceface related to their representation of employees within an organisation. According to Freemanand Medoff (1984), it is this second face of unionism that can promote greater efficiency withinthe firm: by having a collective voice in their dealings with management, employees are morelikely to remain with the organisation than to exit by quitting. These lower levels of labourturnover will make unionised firms more selective in who they employ and more willing toinvest in training. As a consequence, this stable and well-educated workforce will be moreproductive and efficient, generating substantial cost savings for the organisation (Freeman andMedoff, 1984). However, as Hirsch (2004) points out, there is limited empirical support for thisproposition. This collective voice does, however, provide some protection for workers asresearch has found individual employees who filed grievances received lower performanceratings (Lewin and Peterson, 1988; Klaas and DeNisi, 1989) and had fewer promotionalopportunities (Lewin and Peterson, 1988).

Two problems, however, exist with using the presence of a union as a proxy for employeevoice. First, not all unions have an equal capacity to represent members. These differences wereillustrated by Callus et al. (1991), who classified Australian workplaces as non-unionised (nounion members), inactive (no union activity) and active (senior delegate present, membershipmeetings held or management–union negotiations). While some 85.5 per cent of workplaceswere classified as unionised, only 41 per cent of these workplaces were rated as active unionworkplaces. When collective bargaining was added to the criteria this percentage fell to 30 per

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 81

© 2009 Blackwell Publishing Ltd.

Page 3: Employee voice: does union membership matter?

cent (Callus et al., 1991: 156). That is, less than one third of Australian unionised workplaces in1991 had a union structure that could provide for effective employee voice. By 1995 thesefigures had fallen to 24 per cent and 15 per cent, respectively (Morehead et al., 1997: 329).

Second, it can no longer be asserted that unions represent the only avenue for collectivevoice. The substantial decline in unionism in many advanced industrial economies over thepast two decades raised the question as to whether employees were losing their voice. Millwardet al. (2000), using the UK Workplace Industrial Relations Survey (WERS) data, concluded thatemployees had not lost their voice but that the nature of the voice channels had changedsignificantly over the period from 1980 to 1998. Representative and indirect mechanisms, suchas trade unions, had declined, but more direct non-union mechanisms such as meetingsbetween management and employees had significantly increased. These changes, it has beenargued, represent the acceleration of a trend towards direct non-union voice mechanisms thatstarted back in the early part of the 20th century (see Kaufman and Taras, 2000; Taras andKaufman, 2006; Willman et al., 2007).

Some management-sponsored voice mechanisms are also collective in nature. Batt et al.(2002) found that problem-solving groups and self-directed teams had the same effect as unionson quit rates; the presence of these mechanisms significantly reduced quits, although they didnot detract from the significance of unions. Dundon et al. (2004: 1160) also found that somemanagement-sponsored activities, such as attitude surveys, project teams and joint consultationalso served as a collective form of employee voice. These findings are important as theychallenge the notion that trade unions are the only source of collective voice and also illustratethe possibility that particular collective non-union voice mechanisms may be more appropriateto particular levels of the organisation (McCabe and Lewin, 1992; Benson, 2000).

We contend, however, that it will be the employees’ perceptions of voice, rather than theexistence or features of a voice mechanism that will determine whether employees will try tovoice their concerns to management, remain silent and/or exit the organisation. Yet, mostresearch has focused on the existence of voice mechanisms and therefore tells us little about theefficacy of these mechanisms (see Cox et al., 2006; Marchington, 2006). One exception was thestudy by Cox et al. (2006) that found that the more embedded voice practices were the morelikely employees would be committed to the organisation and be satisfied with their job. Asecond exception is the research of Bryson (2004) that explored employees’ perceptions of‘managerial responsiveness’ to various forms of union and non-union voice. Bryson (2004:225–227) found that employees perceived management to be most responsive when the voicemechanisms were direct and of the non-union variety.

Researchers have sought to develop measures of employees’ perceptions of voice. Guest andConway (2004) developed two categories of voice based on employee responses to questionsconcerning how frequently they had discussed issues with management or whethermanagement had sought their views. They found that union membership was negativelyrelated to both voice on personal issues and voice on workplace issues, although theserelationships were only statistically significant for public sector employees (Guest and Conway,2004: 113).

The work of Hirschman on the Nigerian railways provides further direction for thedevelopment of a model of employee voice. In this study, Hirschman (1971) argued thatcustomers who were dissatisfied with the service had two options; discuss the issue withmanagement in an attempt to change the situation or cease to use the service. The choice wastherefore between voice and exit. There were, however, a number of conditions attached to thisdecision. First, if a viable exit option does not exist, then there is no decision to be made.Second, assuming exit is a viable option, the decision to use voice will depend on the

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201082

© 2009 Blackwell Publishing Ltd.

Page 4: Employee voice: does union membership matter?

willingness to accept the uncertainties of the voice option and a perception that such anapproach would be effective (Hirschman, 1971: 33–38). The willingness to accept theuncertainties of the voice option was, it was argued, related to ‘that special attachment to anorganization known as loyalty’ (Hirschman, 1971: 77). In other words, loyalty would serve tomoderate the relationship between voice and exit.

In applying this exit–voice–loyalty framework to the employment context, Boroff and Lewin(1997) found, however, that employee loyalty, operationalised as organisational commitment,did not translate into a higher probability of using voice, although loyalty did result in a lowerintention to leave the firm. This result provides some support for Hirschman’s (1971) model,although Boroff and Lewin (1997: 60) concluded that ‘employee loyalty translates into“silence” ’, which is ‘loyal employees “suffer in silence” rather than to exercise voice’. Yet,as Van Dyne et al. (2003) suggested, voice and silence are not bipolar opposites but aremultidimensional constructs and that both may represent strategic responses to particularsituations. In a similar vein, Luchak (2003) found that voice and attachment were notone-dimensional constructs and that employees with an ‘affective’ or emotional attachment tothe organisation were more likely to use direct voice, while those with a ‘rational’ or calculatedattachment were more likely to use indirect or representative forms of voice.

HYPOTHESES AND MODEL ESTIMATION

Hypotheses

Voice mechanisms provide employees with the opportunity to express their concerns and toinfluence the actions of management (Millward et al., 2000; Luchak, 2003; Dundon et al., 2004;Bryson et al., 2006). Moreover, as Batt et al. (2002: 577) found, the more effective employeesperceived these mechanisms, in their case, grievance procedures, the more they were likely toexercise the voice option. If the issue potentially impacts the whole workforce, it is likely thatthe individual employee will feel he or she has little voice as the settlement process will occurat higher levels within the organisation, which may also be geographically remote from thebulk of the workforce. On the other hand, employees spend most of their working day withina much smaller subset of the organisation, namely their work group. It is here that employees’participation and proximity to managers is greatest and so perceptions of voice may well be thestrongest at this level (McMillan, 2000: 352). It is therefore hypothesised that:

Hypothesis 1: Employees will perceive their voice to be strongest at the work group level.

In one of the few voice studies using employee data, Guest and Conway (2004) found thatunion membership was negatively related to employee perceptions of voice, although only forpublic sector employees. Similarly, Bryson (2004) found that collective union voice wasperceived by workers to have little impact on managerial responsiveness. The alternativeperspective of Freeman and Medoff (1984) suggests that union collective voice reducesemployees’ job dissatisfaction and their intent to quit their employing organisation. Similarly,as unions provide more breadth in the available voice mechanisms, their presence shouldimprove commitment and job satisfaction (Cox et al., 2006). Collective union voice also has animpact on the effectiveness of direct non-union voice. Kessler et al. (2004: 528) found that in theabsence of collective voice mechanisms, employees were likely to find direct non-union voicemechanisms less useful. Such a result suggests a complementarity between the various formsof voice, and it is therefore hypothesised that:

Hypothesis 2a: Union membership will positively contribute to the strength of employeeperceptions of voice.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 83

© 2009 Blackwell Publishing Ltd.

Page 5: Employee voice: does union membership matter?

Hypothesis 2b: The positive relationship between union membership and employeeperceptions of voice will be strongest at the organisational level.

The relationship between employee perceptions of voice, union membership andorganisational level is complex. As Hirschman (1971) argued, more highly committedemployees will view the voice option more favourably, notwithstanding that they may not usethat option (Boroff and Lewin, 1997). In other words, the relationship between employee voiceperceptions and union membership may be affected by the employee’s degree of organisationalcommitment. It is therefore hypothesised that:

Hypothesis 3: The higher the level of employee commitment, the stronger the relationshipbetween union membership and employee perceptions of voice at both the work group andorganisational levels.

Model estimation

To test hypotheses 2 and 3, the following functional model will be utilised: voice = f (unionmembership, organisational commitment, personal and work-related controls), where voice ismeasured at both the work group and the organisational level. The control variables areincluded to avoid ascribing impact that may be due to the characteristics of the respondentsand their work environment. Organisational controls were not included in the model as theresearch was carried out in one organisation, where all employees were subject to the samehuman resource management (HRM) policies and procedures.

The personal characteristics included in the model were age, education, gender and negativeaffectivity. Age can also serve as a proxy for tenure and to a lesser extent, job level, both ofwhich have some bearing on voice (Boswell and Olson-Buchanan, 2004). They were, however,excluded because of the high correlations with age (r = 0.66, p < 0.001, r = 0.47, p < 0.001,respectively). Education is related to the ability to articulate employee’s concerns, although itmay also serve to increase dissatisfaction with the current level of communications and voicearrangements. More highly educated employees have been shown to have the highest level ofunionisation (Gunnigle et al., 2005; Geary, 2007), and these employees have also been found tobe more likely to be involved in problem-solving groups and self-directed teams (Batt et al.,2002: 585). Gender was included as women employees may well have a negative view of voicearrangements as they are often excluded from information flows, often occupy low-levelpositions and also are less likely to participate in collective union voice arrangements andcollective non-union problem-solving groups (Batt et al., 2002: 585). While there is a significantbody of research that suggests that women have as strong a preference for unionisation as domen (Leigh and Hills, 1987; Yates, 2006), this research also points out that gender differencesin unionisation rates can be substantially attributed to the male-focused recruitment practicesof unions and the means of communicating with, and representing, female members (Yates,2006).

The fourth personal characteristic to be included in the model was negative affectivity, whichis a measure of the negative emotional state of the employee. Negative affectivity has beenfound to be related to higher levels of stress at work (Chen and Spector, 1991; Spector andO’Connell, 1994), and employees with high negative affectivity are probably less likely todisplay an attachment to an organisation. They are, however, probably less likely to leavebecause of fewer perceived employment alternatives (Iverson and Buttigieg, 1999). If, however,they view the organisation as the source of their negative emotional state, they may developa strong intention to quit rather than voice their concerns or work to rectify the situation.

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201084

© 2009 Blackwell Publishing Ltd.

Page 6: Employee voice: does union membership matter?

The two work-related variables included as controls were role conflict and co-workersupport. Employees who face role conflict in their day-to-day job have grievances over whichthey may wish to exercise voice or quit. The union in these cases offers an alternative path toquitting (Freeman, 1976). Non-union members do not have such a choice, although for allworkers, a successful resolution of the conflict may well depend on the level of supportindividual employees receive from their co-workers. Supportive work groups provide a senseof belonging and a mechanism for expressing views to management. As a consequence,employees are less likely to quit (Newell, 2000). Where co-workers are not supportive, theworker is faced with the choice to quit, to individually take up the matter with management,or if they are a union member, to request assistance from the union.

METHODOLOGY

Setting and subjects

The setting for this study was a large Australian public sector, scientific research organisation(PSR). PSR’s primary roles are to carry out scientific research, to assist Australian industry, andto encourage and facilitate the application and use of scientific research. PSR receives 70 percent of its funding from government with the balance of funds derived from commercialactivities. At the time of the research, PSR employed 6,957 employees at over 50 sites, coveringa range of occupations including tradespersons, technicians, clerical staff, and managerial andscientific research workers. Nearly half the workforce of PSR was unionised by a single unionand union coverage extended to employees in all salary classifications up to and includingmid-level corporate employees.

The data for this research were collected in the period December 1998 to March 1999. Theage of the data is, however, only an issue if the context in which the data were collected haschanged to the extent that it would impact on the key relationships being investigated (seeMarchington, 2006: 242–246; Dundon and Gollan, 2007: 1185–1189; Wilkinson et al., 2007:1291–1293). There have been no significant changes in the context that could undermine thevalue of the results. Australian industrial relations has had little structural change since 1996when the Workplace Relations Act was introduced. Although some further changes (WorkChoices Act) were enacted in 2006, they had virtually no impact on PSR and have nowbeen repealed. Throughout this period, enterprise bargaining was the key approach to thedetermination of employment conditions in unionised enterprises. The unionisation rate inpublic employment over this time remained relatively stable (Australian Bureau of Statistics,2009) and at PSR, has remained at around 50 per cent of the workforce. The labour marketfor scientific and technical workers has remained strong as universities, the major labourcompetitor for PSR, have experienced significant growth during this period. Discussions withPSR senior management and union officials in 2009 indicated that the key corporate objectives,funding and managerial strategies have remained stable.

Limiting the research to this one organisation meant that all employees were covered by thesame union and that all employees worked under the same enterprise agreement, thus reducinga possible source of bias in our results. A case study approach also overcame the problem ofthe diversity that exists among Australian unions (Callus et al., 1991: 156; Morehead et al., 1997:329), which would tend to confound the findings of multi-organisational research. As the PSRunion has a strong internal delegate structure led by senior delegates, holds regular meetingswith members and management, and engages in collective bargaining, it clearly meets theconditions used by Callus et al. (1991: 156) to define an active union. Thus, union membershipin this study refers to membership of an active union. Having only one union present in the

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 85

© 2009 Blackwell Publishing Ltd.

Page 7: Employee voice: does union membership matter?

organisation means the actual impact of union membership on employee perceptions of voiceshould be relatively consistent among all union members.

PSR management and union officials provided strong support for the research project. Thequestionnaire was mailed to each employee’s work address and returned directly to theresearchers. A number of questions seeking information about demographic characteristicswere also included. The survey achieved an overall response rate of 47.9 per cent (n = 3,335).The mean age of the respondents was just over 42 years, wherein 66.2 per cent of therespondents had completed a university degree, with 30 per cent having completed a PhD, and35 per cent of respondents were female. As information was available from PSR on unionmembership by gender and location, we tested the representativeness of the sample on thesecriteria. The sample was representative of the union and non-union population by gender andlocation (t test, p < 0.05). After allowing for missing values, the effective sample size was 2,949.

Measures

Dependent variables Various meanings and definitions of voice had been advanced in theliterature. Dundon et al. (2004: 1153) noted that the ‘precise meaning of the term “employeevoice” ’ was open to question, though there were some commonalities. Voice mechanisms weregenerally seen as a way to provide employees with the ability to articulate concerns and toinfluence the actions of management. Employee voice had commonly been defined as atwo-way process of communication, which is characterised by an exchange of information(Marchington et al., 2001; see also Wilkinson et al., 2004). Employee voice could therefore beseen in terms of three dimensions: the provision of information by management to employees,the willingness of management to listen to employees and management’s preparedness todiscuss work-related problems and issues (see Van Dyne et al., 2003; Marchington, 2006; Brysonand Freeman, 2007; Cannell, 2008).

Included in the PSR data were eight items, measured on a five-point Likert scale thatprovided operational definitions of these three dimensions of employee voice. In particular, asillustrated in Table 1, these items covered employee perceptions of management’s preparednessto discuss problems, management’s willingness to listen to employees, and the degree to which

TABLE 1 Principal components factor analysis (varimax rotation)

Level of voice and items Factor 1 Factor 2

Work group voiceMy supervisor is willing to listen to work-related problemsa 0.84 0.18I feel free to discuss job-related problems with my supervisorb 0.86 0.18I am comfortable expressing my feelings to my supervisorb 0.85 0.16My supervisor can be relied on when things get difficult at worka 0.82 0.19

Organisational voicePSR listens to ideas from employees at my level in the organisation 0.21 0.80Information is available as needed 0.15 0.76I get enough information about PSR 0.08 0.80I can make my views known to PSR management 0.24 0.78

a From Michaels and Spector (1982).b From Tang and Sarfield-Baldwin (1996).

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201086

© 2009 Blackwell Publishing Ltd.

Page 8: Employee voice: does union membership matter?

information is available to employees. These items provided a measure of how employeesperceived their voice possibilities rather than simply relying on the existence of various voicemechanisms. This provides an improvement on many of the measures of employee voice usedin previous research.

The eight potential voice items were subjected to factor analysis as this allowed theunderlying pattern of relationships among the items to be investigated. The principalcomponents technique was used as it ‘does not require any assumptions about the generalnature of the variable’ (Kim, 1975: 479) and is concerned with ‘patterning all the variation ina set of variables, whether common or unique’ (Rummel, 1970: 355). To assist in distinguishingthe pattern of loadings, varimax rotation was performed. Two clear, identifiable factors resultedfrom this analysis (eigenvalue > 1). The first factor (Factor 1) included four items that involvedthe relationship between the employee and their work group supervisor over such matters asthe ability to express and discuss work-related problems. All four items in this factor had factorloadings of 0.82 or above. The reliability between and across the four items was high, with thealpha reliability coefficient (Cronbach, 1951) being 0.88. The second factor (Factor 2) includedfour items that involved the perceptions of the employee of the availability of information onthe organisation and the ability to express their views to senior management within PSR. Theitems in this factor had factor loadings of 0.76 or above. The reliability between and across thefour items was high, with the alpha reliability coefficient being 0.82. In total, the two factorsexplained 69.6 per cent of variance of the eight items. The eight items and the results of thefactor analysis are presented in Table 1.

These results demonstrate that employee voice can be conceptualised at two levels withinthe organisation. The first scale was named ‘Work Group Voice’ and was composed of the fouritems that loaded heavily on Factor 1 being summated and divided by four. The second scalewas named ‘Organisational Voice’ and was composed of the four items that loaded heavily onFactor 2 being summated and divided by four.

Independent variables Union membership refers to whether the respondent belonged to theunion. To ensure each respondent was categorised correctly, the questionnaire was sent outseparately to union and non-union members. Fifty-four per cent of all the respondents wereunion members at the time of the survey. Organisational commitment was measured by thenine-item version of Porter’s organisational commitment scale (Porter et al., 1974; Mowdayet al., 1979). This scale was used because of its reliability and validity (Mowday et al., 1979),and its widespread use and acceptance (Morris et al., 1993). A five-point Likert scale wasused and individual scores were calculated by averaging the responses to the items(a = 0.86).

Control variables Six control variables covering personal and work-related factors wereincluded in the analysis. Age was measured in years, education was categorised into nine levelsbased on the number of years of schooling: year 10 or lower, year 11 or 12, apprenticeship,technical certificate, diploma, 3-year degree, honours degree or fourth year, masters’ degree andPhD/doctorate. Gender was coded 1 for female and 0 for male. Negative affectivity was basedon the three-item scale by Watson et al. (1987) (a = 0.86). The two work-related variablesincluded were role conflict and co-worker support, which were both measured on a five-pointLikert scale. Role conflict was measured by three items from Kahn et al. (1964) (a = 0.74).Co-worker support was measured by three items from House (1981) (a = 0.86).

Details of each variable, along with descriptive statistics, are presented in Table 2. In thisstudy, the reliability of the scales ranged from 0.74 for role conflict to 0.88 for work group voice.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 87

© 2009 Blackwell Publishing Ltd.

Page 9: Employee voice: does union membership matter?

As all scales exceeded 0.7 they were considered to be reliable measures of the constructs underconsideration (Cortina, 1993).

Statistical analysis and model estimation

Hypothesis 1 will be tested using a t-test to ascertain the significance of the difference betweenemployees’ perception of work group and organisational voice. Hypotheses 2 and 3 will betested using hierarchical ordinary least squares (OLS) regression techniques (Hair et al., 1998).The dependent variables were work group voice and organisational voice. Both employee

TABLE 2 Variable definitions

Variable Definition and construction Mean (standarddeviation)

Dependent variablesWork group voice Employees’ perceptions of voice at the work group

level as measured by four items as indicated inTable 1, alpha = 0.88. Five-point scale where 5 = highlevel of perceived voice within the work group.

3.74 (0.87)

Organisational voice Employees’ perceptions of voice at the organisationallevel as measured by four items as indicated inTable 1, alpha = 0.82. Five-point scale where 5 = highlevel of perceived voice within the organisation.

3.14 (0.77)

Independent variablesUnion Dichotomous variable where 1 = union member. 0.54 (0.50)Commitment The extent to which an employee believes in and

accepts an organisation’s goals and values, togetherwith a willingness to exert considerable effort forthat organisation as measured by nine items fromPorter et al. (1974), alpha = 0.86. Five-point scalewhere 5 = highly committed.

3.38 (0.62)

Control variablesAge Age of respondent in years. 42.02 (9.96)Education Nine levels of education ranging from year 10 or

lower = 1 to PhD = 9.6.32 (2.44)

Gender Dichotomous variable where female = 1. 0.35 (0.48)Negative affectivity The extent to which an individual experiences aversive

emotional states over time and across situations asmeasured by three items from Watson et al. (1987),alpha = 0.086. Five-point scale where 5 = highnegative affectivity.

2.77 (0.89)

Role conflict Degree to which an employee’s role expectations areincompatible. Role conflict is measured by threeitems from Kahn et al. (1964), alpha = 0.74. Five-pointscale where 5 = high role conflict.

2.78 (0.79)

Co-worker support Degree of consideration expressed by co-workers.Co-worker support is measured by three items fromHouse (1981), alpha = 0.86. Five-point scale where5 = high level of co-worker support.

3.54 (0.76)

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201088

© 2009 Blackwell Publishing Ltd.

Page 10: Employee voice: does union membership matter?

perceptions of voice measures were composed of four items where the individual score for eachitem ranged from 1 to 5. When the four items were summated and divided by four, voice scoresranged from 1 to 5 with 17 different values. As these scores approximated interval data, OLSregression was selected as this technique is sufficiently robust to handle this type of data (Berryand Feldman, 1985). As employee perceptions of voice are measured at two levels, twoequations will be generated for each hypothesis. Step 1 involved entering the six controlvariables into the model (control model). Step 2 involved entering the control variables and thetwo independent variables, union membership and organisational commitment (full model).The predicted directions of each variable entered into the model in steps 1 and 2 are providedin parentheses after the name of each variable in Tables 4 and 5. Step 3 involved entering allthe variables contained in the full model plus an interaction term made up of unionmembership and organisational commitment. If the impact of union membership on the twolevels of voice is modified by the level of organisational commitment, then this will be indicatedby a statistically significant interaction term. The change in the explained level of variancebetween these models (DR2) is also presented to ascertain whether the explanatory power of themodel had improved substantially by the addition of the various hypothesised predictors ofemployee perceptions of voice.

The possibility of multicollinearity (Campbell and Friske, 1959) rendering the estimatesunreliable appeared low. The correlations between all the independent variables in the modeldid not exceed 0.43 (see Table 3), well below the 0.80 figure at which multicollinearity maybe considered a problem (Studenmund and Cassidy, 1987). Further investigation wasundertaken in accordance with the assessment procedure recommended by Hair et al. (1998:220–221). First, a condition index was computed for each independent variable contained ineach step of the two models. In both models, one variable (in steps 2 and 3) had a conditionnumber above 30, the most commonly used threshold value (Hair et al., 1998: 220).Decomposing the variance showed that in no case did these variables have a varianceproportion, with the other independent variables above 90 per cent. Second, a varianceinflation (vif) test was run for all the models reported in Tables 4 and 5, with the exceptionof the interaction model. The mean vif did not exceed 1.12 and no individual vif was greaterthan 1.52; substantially below the level at which multicollinearity would be a problem(Chatterjee et al., 2000).

TABLE 3 Correlation matrix (all employees)

1 2 3 4 5 6 7 8 9 10

1. Work group voice 1.002. Organisational voice 0.41 1.003. Age -0.01 0.07 1.004. Education -0.00 0.05 0.01 1.005. Gender 0.01 -0.04 -0.21 -0.25 1.006. Negative affectivity -0.25 -0.23 -0.04 0.04 -0.02 1.007. Role conflict -0.34 -0.25 -0.02 -0.01 -0.09 0.31 1.008. Co-worker support 0.43 0.31 -0.06 0.00 0.04 -0.22 -0.21 1.009. Union membership -0.06 -0.12 0.27 0.07 -0.08 0.09 0.03 -0.03 1.00

10. Organisationalcommitment

0.31 0.41 0.07 -0.03 -0.01 -0.18 -0.17 0.24 -0.05 1.00

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 89

© 2009 Blackwell Publishing Ltd.

Page 11: Employee voice: does union membership matter?

RESULTS

It was hypothesised that employees will perceive their voice to be strongest at the work grouplevel (H1). This was the case in PSR. Work group voice had a mean of 3.74 and standarddeviation of 0.87, while the mean and standard deviation of organisational voice was lower at3.14 and 0.77, respectively. A t-test of the difference between these means indicated that thedifference was statistically significant (t = 37.84, p < 0.01). Clearly, PSR employees perceivedthey had significantly more voice at the work group level than they did at the higher levels ofthe organisation. Hypothesis 1 was therefore accepted.

This dual categorisation of voice was used in the subsequent regression analysis (Tables 4and 5). For work group voice, three of the control variables entered in step 1 (control model)were statistically significant (p < 0.01) in the direction predicted (see column 1 of Table 4). Thismodel explained 25.9 per cent of the variance in work group voice. When union membershipand organisational commitment were entered into the model (step 2 – full model), onlyorganisational commitment was statistically significant (p < 0.01). The level of explainedvariance increased to 29 per cent with the increase of 3.1 per cent being statistically significant(F = 63.58, p < 0.001). Much of the increase was due to the inclusion of the organisationalcommitment variable. This can be seen in the minor loss of explained variance when the unionvariable was deleted from the full model (0.05 per cent). The three control variables that weresignificant in the control model remained significant in the full model. Results of that analysis

TABLE 4 Hierarchical OLS regression for work group voice

Variable Step 1 (control) Step 2 (full) Step 3 (interaction)

Intercept 3.37 (0.13) 2.64 (0.14) 2.77 (0.17)Control variables

Age (+) -0.00 (0.00) -0.00 (0.00) -0.00 (0.00)Education (+) -0.00 (0.01) -0.00 (0.01) -0.00 (0.01)Gender (–) -0.06 (0.03) -0.06 (0.03) -0.06 (0.03)Negative affectivity (–) -0.10** (0.02) -0.08** (0.02) -0.08** (0.02)Role conflict (–) -0.27** (0.02) -0.25** (0.02) -0.25** (0.02)Co-worker support (+) 0.41** (0.02) 0.36** (0.02) 0.36** (0.02)

Independent variablesUnion membership (+) -0.05 (0.03) -0.27 (0.15)Organisational commitment (+) 0.25** (0.03) 0.22** (0.03)

Interaction termUnion*commitment 0.07 (0.04)

Adjusted R2 0.259 0.290 0.290D Adjusted R2 – 0.031 0.000F-test (D Adjusted R2) 174.49** 63.58** 2.32Mean variance inflation (vif) 1.10 1.11 –Highest individual vif 1.39 1.45

* p < 0.05, ** p < 0.01.Note: unstandardised coefficients; standard errors in parentheses.DR2 (Adjusted) is the additional variance explained compared to the previous model.

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201090

© 2009 Blackwell Publishing Ltd.

Page 12: Employee voice: does union membership matter?

are presented in columns 1 and 2 of Table 4 and show that employee perceptions of work groupvoice relied more on organisational commitment. Union membership only served to weakenthose perceptions.

A different situation appeared in relation to organisational voice (Table 5). In this case, all sixcontrol variables entered in step 1 were statistically significant (p < 0.01) in the directionpredicted (see column 1 of Table 5). This model explained 14.9 per cent of the variance inorganisational voice. When union membership and organisational commitment were enteredinto the model (step 2 – full model), both variables were statistically significant (p < 0.01).Contrary to our hypotheses, union membership was significantly negatively related toorganisational voice. In other words, union members had a lower perception of theirorganisational voice than did those employees who did not belong to the union. The inclusionof these two variables increased the level of explained variance to 25.7 per cent. This increaseof 10.8 per cent was statistically significant (F = 230.79, p < 0.001). This substantial increase inexplained variance again, however, was primarily due to the inclusion of the organisationalcommitment variable. When the union variable was deleted from the model, the level ofexplained variance fell by only 0.15 per cent. In other words, it was the level of an employee’sorganisational commitment that was the major contributor to their perceptions of organisationalvoice. All six control variables remained statistically significant in the full model. Results of theanalysis are presented in columns 1 and 2 of Table 5.

Taken collectively, the results presented in Tables 4 and 5 provided no support forHypothesis 2a (that union membership will positively contribute to the strength of employee

TABLE 5 Hierarchical ordinary least squares regression for organisational voice

Variable Step 1 (control) Step 2 (full) Step 3 (interaction)

Intercept 2.67 (0.12) 1.42 (0.13) 1.50 (0.15)Control variables

Age (+) 0.01** (0.00) 0.01** (0.00) 0.01** (0.00)Education (+) 0.01** (0.01) 0.02** (0.01) 0.02** (0.01)Gender (–) -0.06* (0.03) -0.07** (0.03) -0.07* (0.03)Negative affectivity (–) -0.10** (0.02) -0.07** (0.01) -0.07** (0.01)Role conflict (–) -0.15** (0.02) -0.12** (0.02) -0.12** (0.02)Co-worker support (+) 0.25** (0.02) 0.19** (0.02) 0.18** (0.02)

Independent variablesUnion membership (+) -0.17** (0.03) -0.33* (0.13)Organisational commitment (+) 0.40** (0.02) 0.38** (0.03)

Interaction termUnion*commitment 0.05 (0.04)

Adjusted R2 0.149 0.257 0.257D Adjusted R2 – 0.108 0.000F-test (D Adjusted R2) 88.75** 230.79** 1.62Mean variance inflation (vif) 1.10 1.12 –Highest individual vif 1.39 1.52

* p < 0.05, ** p < 0.01.Note: unstandardised coefficients; standard errors in parentheses.DR2 (Adjusted) is the additional variance explained compared to the previous model.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 91

© 2009 Blackwell Publishing Ltd.

Page 13: Employee voice: does union membership matter?

perceptions of voice), although union membership was found to have a significant relationshipto organisational voice. The finding that this was a negative relationship, however, means thatHypothesis 2b (the relationship between unions and employee perceptions of voice will bestrongest at the organisational level) was also rejected. This finding challenges the argument ofFreeman and Medoff (1984) that union membership enhances employee perceptions of voice,although our results are consistent with the findings of Guest and Conway (2004) for publicsector employees. The significant relationship between employee perceptions of voice (bothwork group and organisational voice) and organisational commitment provided some supportfor the argument of Hirschman (1971) that more loyal (committed) employees are more likelyto perceive they have voice. Our final hypothesis (Hypothesis 3) proposed that organisationalcommitment will moderate the relationship between employee perceptions of work group andorganisational voice, and union membership. As shown in Tables 4 and 5, the interaction termwas not statistically significant for either work group or organisational voice and did notcontribute to the level of explained variance. This means that perceptions of voice are notdifferent between unionists and non-unionist based on different levels of organisationalcommitment. As a consequence, Hypothesis 3 was rejected.

DISCUSSION

In this study, we examined employees’ perception of voice. We hypothesised that employeeperceptions of voice would be strongest at the work group level as this is where employeesspend their time and how they typically connect with the organisation. This hypothesis wassupported by the data. PSR is a relatively large organisation and the decentralised approach todecision-making appears to have promoted employee perceptions of work group voice. Wethen turned our attention to the role of union membership on employee perceptions of voiceat the organisational level. We hypothesised that union membership would be positivelyassociated with employee perceptions of organisational voice as unions are most visible at thislevel. For example, unions seek to establish the terms and conditions of employment for allthe members of the organisation and represent employees’ interests during negotiationsabout organisation-wide issues such as redundancies, performance management systemsand organisational change programmes. Finally, we also hypothesised that organisationalcommitment would moderate the relationship between union membership and employeeperceptions of work group and organisation voice. These two hypotheses were rejected. Themost notable result, however, was that that union membership was negatively related toemployees’ perceptions of organisational voice.

There are a number of possible explanations for this unanticipated result. First, unionmembers may have a higher level of dissatisfaction with the organisation and managementprocesses (Freeman and Medoff, 1984: 21; Bryson et al., 2004), including communications andparticipative arrangements, and therefore perceive less favourably their degree of influencewithin the organisation (Kleinman, 2000: 403–404) or the degree of managerial responsiveness(Bryson et al., 2006). Bryson (2004) found that union members had a higher level of jobdissatisfaction than non-union workers but that after an extensive array of control variableswere included, no significant negative relationship was found. This finding, they suggested,may well be due to a selection bias on the part of employees. Employees who had higheraspirations for their working life may be more likely to join a union. This proposition resonateswell in the case of PSR, where some 63.8 per cent of unionised employees indicated that a majorreason for joining the union was because they wanted to have a say in matters affecting theirworking life. Unions themselves may also contribute to this dissatisfaction by raising employee

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201092

© 2009 Blackwell Publishing Ltd.

Page 14: Employee voice: does union membership matter?

expectations that they are unable to meet (see also discussion on ‘consciousness raising’ inGuest and Conway, 2004: 115–116) and by providing their members with considerableinformation about problems within the organisation and the problems they are encounteringwith management. This is likely to be part of the explanation for the findings of this researchas at the time of the survey, PSR management were keen to introduce a performance-relatedpay scheme which had led to an active campaign of opposition by the union.

A second explanation is that union members may have higher expectations of voice.Employees join unions for a variety of reasons (Peetz, 1998), but overall, they expect unions tomake a difference to their working lives. As Bryson and Freeman (2007: 84) found, ‘unionisedworkers reported more problems with management’ than non-union workers. In the case ofPSR, the major reasons for employees joining the union were a belief in unions, wanting to havea say in things that affect their working life and a belief that unions generate better wages andconditions. Clearly, these union members had high expectations concerning voice. Yet, it wasalso the case that these expectations were not being met. Improved wages and workingconditions were becoming harder to achieve as the enterprise bargaining system, with itsemphasis on productivity improvements, had led to trade-offs involving redundancies, adecline in wage relativities with the private sector, and the possibility of a new performance-based pay scheme being introduced.

A third explanation is that the overall focus within PSR at the time of the research hadshifted from curiosity-led to commercially-driven research with an emphasis on productivityand efficiency. For employees of PSR, most of who were accustomed to the more protectedworking environment of the public sector, this meant that external factors were now drivingresearch, and the scientific arguments used in the past for justifying research projects werebecoming less important. In this environment, the value of union membership was becomingmore marginal as union members had less protection than that of the past in the case ofredundancies and adverse performance appraisals. In addition, other benefits of unionmembership such as access to grievance procedures were available to all employees regardlessof their union status. It may also have been the case that management’s approach to reform,for example the introduction of a performance-based pay scheme that ran counter to the wishesof union members and past practices, may have led many unionists to believe they were beingtargeted. In this context, the tense state of the relationship between management and the unionmay have created a negative perception of collective union voice (Freeman and Medoff, 1984).

Implications for theory

Industrial relations researchers in studying voice have traditionally been concerned with thepresence or absence of a union. This literature has tended to focus on union activity at anorganisational level. By collecting up the issues of workers across the organisation and takingthese up with management, unions are able to protect individual workers and improve theterms and conditions of work of their members (Crouch, 1982). The rise of HRM brought withit an interest in direct non-union voice mechanisms. Direct communications between individualemployees and their supervisor is seen to result in higher trust and cooperation between theparties (Waring, 1999). Yet, the mere presence of a voice mechanism does not tell us whetherthat mechanism actually increases the individual employee’s perception of voice (Marchington,2006). By focusing on the individual as the unit of analysis, this article has opened up thepossibility of developing the voice concept in a way that more closely connects with howemployees feel and what influences their decisions concerning work effort and theircontinuance in the workplace.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 93

© 2009 Blackwell Publishing Ltd.

Page 15: Employee voice: does union membership matter?

Combining the industrial relations and HRM literature with the findings from our researchsuggests several important implications for the further development of theory. First, voicetheory should accommodate the perceptions of individual employees and recognise that theseperceptions are likely to vary according to the level being considered. Employees in this studyperceived they had more voice at the work group level, while part of the explanation for thisfinding is the proximity to colleagues and their supervisor (Middlemist and Hitt, 1988), ofimportance is the move by organisations to decentralised decision-making structures, whichhave created more avenues through which employees can express voice. At the organisationallevel, these conditions do not generally exist, which leads employees to feel remote from thevoice processes. However, limited bargaining power or access to information may also limitemployees’ perceptions of work group voice.

Second, models of employee voice have generally assumed union membership would resultin employees perceiving greater voice. In our research we found that union membership wasassociated with lower perceptions of organisation voice. This has important implications asFreeman and Medoff (1984) contend that voice would be associated with higher productivity.However, empirical studies have produced inconsistent findings on the relationship betweenvoice and productivity (Hirsch, 2004). Our research suggests that this inconsistency may be dueto the operationalisation of voice. This is to say it is not about the existence of voicemechanisms but how workers perceive their level of voice. Future researchers need to rethinkthe conceptualisation of voice to take into account the level at which voice occurs, as well asemployee perceptions of voice.

Third, the research demonstrated that commitment to an organisation is a significant andsubstantial influence on employees’ perceptions of voice at both the work group andorganisation level. This provides some support for the arguments of Hirschman (1971: 78) thatloyalty (commitment) ‘holds exit at bay and activates voice’. According to Hirschman (1971: 80),it does this by ‘raising the cost of exit’ and thus inducing employees to exercise voice. Theresearch also found a range of employee characteristics impacted on employee perceptions ofvoice. These antecedents of voice included negative affectivity, role conflict and co-workersupport for work group voice, and these three variables plus age, education and gender fororganisation voice. This finding suggests that models of employee perceptions of voice can nolonger focus solely on voice mechanisms but should also consider the characteristics ofemployees.

Implications for policy and research

The findings of the study have important implications for unions as members are less likely toreport an organisation voice, relative to non-members. The challenge for unions is to encouragetheir members to regard the involvement of full-time officials in organisational decision-makingas providing strong employee voice at the organisation level. Our results suggest that unionsshould devote more resources to communicating with members about their activities in theorganisation on behalf of members. The growth of new technologies, such as the internet andemail, offer unions low-cost ways to communicate more regularly with members about theiractivities on behalf of its members.

The findings of the study also have important implications for industrial relations and HRmanagers, particularly around issues of the design of employee involvement schemes.Employee involvement in decision-making is a technique that can be used by management toenhance and maintain a competitive advantage (Pfeffer, 1994). In our study, employees(irrespective of union status) perceived work group voice. This represents a good foundationfor the development of more formal schemes of organisational participation as work group

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201094

© 2009 Blackwell Publishing Ltd.

Page 16: Employee voice: does union membership matter?

level schemes can address issues of direct and immediate importance to workers and have thecapacity to engender higher levels of commitment to the outcomes (Boot et al., 1979). Theextension of work group employee voice may, however, create challenges for line managerswho may see their decision-making authority as being undermined by a more participativestructure (Hammer, 2001).

Industrial relations and HR managers could further promote the perception of employeevoice through the provision of training of employees and line managers in a range ofcommunication skills. Employees who are able to effectively articulate their concerns to a linemanager skilled in communications are more likely to perceive they have high work groupvoice, which in turn, may influence their perceptions of organisational voice. Line managers canalso play a key role in influencing employee perceptions of voice by being receptive toemployees’ concerns and also tailoring managerial policies to enhance commitment (Guest,1995; Becker et al., 1996). While there are some doubts as to whether commitment can bemanaged (Guest, 1992), the research showed that employee perceptions of voice at both levelswas significantly related to commitment, which in turn, was significantly related to co-workersupport. Line managers could therefore focus their attention on building strong systems ofco-worker support through such mechanisms as teams.

Our research has generated two reliable measures of employee perceptions of voice. Weencourage future researchers to apply these new measures to a variety of contexts and sectorsto further our conceptual understanding of the determinants of employee perceptions of voice.We suggest that researchers examine employee perceptions of voice in the private sector wherethere are typically lower levels of union membership (Ebbinghaus, 2002). In PSR, about half ofthe respondents were union members, and the union had a relatively good relationship withmanagement. In the private sector, organisations may be less accepting of unions, so membersmay perceive greater organisation voice as the union is able to achieve improvements formembers in the face of significant opposition. We also suggest that researchers should expandtheir measures of unionism to accommodate the level of union activities within theorganisation. For example, situations where a union does little to keep members informed orwhere management–union relations are poor may lead employees to believe that managementhas withheld information or are not keen on engaging in voice mechanisms. In these cases, anegative relationship between union membership and employee perception of voice would beexpected.

CONCLUSION

The general conclusion that can be drawn from this research is that unions appear to have someeffect in reducing employees’ perceptions of voice but only at the organisational level. Such afinding was made possible by the adoption of a more individually focused view of employeevoice, which allowed for the impact of unions to be more clearly distinguished. The resultsdemonstrate an alternative way forward in the development of the voice concept, and thedevelopment of reliable measures of employee perceptions of voice at the work group and theorganisational level have improved the measurement of voice used in earlier research.

Acknowledgement

This research was funded by grants received from the Faculty of Economics and Commerce,University of Melbourne. We thank the management of PSR and officials of the Community andPublic Sector Union for their ongoing support of the project. We also thank the employees ofPSR for their willingness to participate in the research.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 95

© 2009 Blackwell Publishing Ltd.

Page 17: Employee voice: does union membership matter?

REFERENCES

Addison, J. and Belfield, C. (2004). ‘Union voice’. Journal of Labor Research, 25: 4, 563–596.Ajzen, I. and Fishbein, M. (1977). ‘Attitude–behavior relations: a theoretical analysis and review of

empirical research’. Psychological Bulletin, 84: 5, 888–918.Australian Bureau of Statistics (2009). Employee Earnings, Benefits and Trade Union Membership, August

2008, media release, 17 April, Canberra: Australian Bureau of Statistics.Avery, D. and Quiñones, M. (2004). ‘Individual differences and the voice effect: the moderating role

of value of voice’. Group & Organization Management, 29: 1, 106–124.Batt, R., Colvin, A. and Keefe, J. (2002). ‘Employee voice, human resource practices and quit rates:

evidence from the telecommunications industry’. Industrial & Labor Relations Review, 55: 4, 573–594.

Becker, T., Billings, R., Eveleth, D. and Gilbert, N. (1996). ‘Foci and bases of employee commitment:implications for job performance’. Academy of Management Journal, 39: 2, 464–482.

Benson, J. (2000). ‘Employee voice in union and non-union Australian workplaces’. British Journal ofIndustrial Relations, 38: 3, 453–459.

Berry, W. and Feldman, S. (1985). Multiple Regression in Practice, Beverly Hills, CA: Sage.Bloom, G. and Northrup, H. (1965). Economics of Labour Relations, Homewood, IL: Irwin.Boot, R., Cowling, A. and Stanworth, M. (1979). Behavioural Sciences for Managers, Melbourne: Edward

Arnold.Boroff, K. and Lewin, D. (1997). ‘Loyalty, voice, and intent to exit a union firm: a conceptual and

empirical analysis’. Industrial & Labor Relations Review, 51: 1, 50–63.Boswell, W. and Olson-Buchanan, J. (2004). ‘Experiencing mistreatment at work: the role of grievance

filing, nature of mistreatment, and employee withdrawal’. Academy of Management Journal, 47: 1,129–139.

Bryson, A. (2004). ‘Managerial responsiveness to union and non-union worker voice in Britain’.Industrial Relations, 43: 1, 213–242.

Bryson, A. and Freeman, R. (2007). ‘What voice do British workers want?’, in R. Freeman, P. Boxalland P. Haynes (eds), What Workers Say: Employee Voice in the Anglo–American Workplace, Ithaca, NY:Cornell University Press, pp. 72–96.

Bryson, A., Cappellari, L. and Lucifora, C. (2004). ‘Does union membership really reduce jobsatisfaction?’ British Journal of Industrial Relations, 42: 3, 439–459.

Bryson, A., Charlwood, A. and Forth, J. (2006). ‘Worker voice, managerial response and labourproductivity: an empirical investigation’. Industrial Relations Journal, 37: 5, 438–455.

Butler, P. (2005). ‘Non-union employee representation: exploring the efficacy of the voice process’.Employee Relations, 27: 3, 272–288.

Callus, R., Morehead, A., Cully, M. and Buchanan, J. (1991). Industrial Relations at Work, Canberra:Australian Government Printing Service.

Campbell, D. and Friske, D. (1959). ‘Convergent and discriminant validation by themultitrait–multimethod matrix’. Psychological Bulletin, 56: 2, 81–105.

Cannell, M. (2008). Employee Voice, London: Chartered Institute of Personnel and Development.Chatterjee, S., Hadi, A. and Price, B. (2000). Regression Analysis by Example, 3rd edn, New York: John

Wiley and Sons.Chen, P. and Spector, P. (1991). ‘Negative affectivity as the underlying cause of correlation between

stressors and strains’. Journal of Applied Psychology, 76: 3, 398–407.Cortina, J. (1993). ‘What is coefficient alpha? An examination of theory and applications’. Journal of

Applied Psychology, 78: 1, 98–104.Cox, A., Zagelmeyer, S. and Marchington, M. (2006). ‘Embedding employee involvement and

participation at work’. Human Resource Management Journal, 16: 3, 250–267.Cronbach, L. (1951). ‘Coefficient alpha and the internal structure of tests’. Psychometrika, 16: 3,

297–334.Crouch, C. (1982). Trade Unions: The Logic of Collective Action, London: Fontana.

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201096

© 2009 Blackwell Publishing Ltd.

Page 18: Employee voice: does union membership matter?

Dietz, G., Cullen, J. and Coad, A. (2005). ‘Can there be non-union forms of workplace partnership?’Employee Relations, 27: 3, 289–306.

Dundon, T. and Gollan, P. (2007). ‘Re-conceptualizing voice in the non-union workplace’. InternationalJournal of Human Resource Management, 18: 7, 1182–1198.

Dundon, T., Wikinson, A., Marchington, M. and Ackers, P. (2004). ‘The meaning and purpose ofvoice’. International Journal of Human Resource Management, 15: 6, 1149–1170.

Dundon, T., Wikinson, A., Marchington, M. and Ackers, P. (2005). ‘The management of voice in thenon-union organizations: managers’ perspectives’. Employee Relations, 27: 3, 307–319.

Ebbinghaus, B. (2002). ‘Trade unions’ changing role: membership erosion, organisational reform, andsocial partnership in Europe’. Industrial Relations Journal, 33: 5, 465–483.

Fishbein, M. and Ajzen, I. (1975). Belief, Attitude, Intention, and Behavior: An Introduction to Theory andResearch, Reading, MA: Addison-Wesley.

Flanders, A. (1952). Trade Unions, London: Hutchinson.Freeman, R. (1976). ‘Individual mobility and union voice in the labor market’. American Economic

Review, 66: 361–368.Freeman, R. and Medoff, J. (1984). What Do Unions Do?, New York: Basic Books.Freeman, R. and Rogers, J. (1993). ‘Who speaks for us? Employee representation in a nonunion labor

market’, in B. Kaufman and M. Kleiner (eds), Employee Representation: Alternatives and FutureDirections, Madison, WI: Industrial Relations Research Association, pp. 13–18.

Freeman, R., Boxall, P. and Haynes, P. (eds) (2007). What Workers Say: Employee Voice in theAnglo–American Workplace, Ithaca, NY: Cornell University Press.

Geary, J. (2007). ‘Employee voice in the Irish workplace: status and prospect’, in R. Freeman, P. Boxalland P. Haynes (eds), What Workers Say: Employee Voice in the Anglo–American Workplace, Ithaca, NY:Cornell University Press, pp. 97–124.

Gollan, P. (2003). ‘All talk and no voice: employee voice at the Eurotunnel call centre’. Economic andIndustrial Democracy, 24: 4, 509–541.

Gollan, P. (2006). ‘A process of transition – employer strategies and outcomes of employee voice atEurotunnel’. Socio-Economic Review, 4: 2, 337–351.

Gospel, H. and Wood, S. (2003). ‘Representing workers in modern Britain’, in H. Gospel and S.Woods (eds), Representing Workers: Trade Union Recognition and Membership in Modern Britain,London: Routledge, pp. 1–14.

Guest, D. (1992). ‘Employee commitment and control’, in J. Hartley and G. Stephenson (eds),Employment Relations, Oxford: Blackwell, pp. 111–135.

Guest, D. (1995). ‘Human resource management, trade unions and industrial relations’, in J. Storey(ed.), Human Resource Management: A Critical Text, London: Routledge, pp. 110–141.

Guest, D. and Conway, N. (2004). ‘Exploring the paradox of unionized worker dissatisfaction’.Industrial Relations Journal, 35: 2, 102–121.

Gunnigle, P., Collings, D. and Morley, M. (2005). ‘Exploring the dynamics of industrial relations inUS multinationals: evidence from the Republic of Ireland’. Industrial Relations Journal, 36: 3,241–256.

Hair, J., Anderson, R., Tatham, R. and Black, W. (1998). Multivariate Data Analysis, 5th edn, UpperSaddle River, NJ: Prentice Hall.

Hammer, T. (2001). ‘Industrial democracy’, in M. Poole and M. Warner (eds), The IEMB Handbook ofHuman Resource Management, London: Thompson Learning, pp. 143–152.

Haynes, P. (2005). ‘Filling the vacuum? Non-union employee voice in the Auckland hotel industry’.Employee Relations, 27: 3, 259–271.

Haynes, P., Boxall, P. and Macky, K. (2005). ‘Non-union voice and the effectiveness of jointconsultation in New Zealand’. Economic and Industrial Democracy, 26: 2, 229–256.

Hirsch, B. (2004). ‘What do unions do for economic performance?’. Journal of Labor Research, 25: 3,415–455.

Hirschman, A. (1971). Exit, Voice and Loyalty: Responses to Decline in Firms, Organizations and States,Cambridge, MA: Harvard University Press.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 97

© 2009 Blackwell Publishing Ltd.

Page 19: Employee voice: does union membership matter?

House, J. (1981). Work Stress and Social Support, Reading, MA: Addison-Wesley.Iverson, R. and Buttigieg, D. (1999). ‘Affective, normative and continuance commitment:

can the “right kind” of commitment be managed?’ Journal of Management Studies, 36: 3, 307–333.Kahn, R., Wolfe, D., Quinn, R., Snoek, J. and Rosenthal, R. (1964). Organizational Stress: Studies in Role

Conflict and Ambiguity, New York: John Wiley and Sons.Kaufman, B. and Taras, D. (1999). ‘Nonunion employee representation: introduction’. Journal of Labor

Research, 20: 1, 1–8.Kaufman, B. and Taras, D. (2000). Nonunion Employee Representation: History, Contemporary Practice and

Policy, Armonk, NY: M.E. Sharpe.Kessler, I., Undy, R. and Heron, P. (2004). ‘Employee perspectives on communication and

consultation: findings from a cross-national survey’. International Journal of Human ResourceManagement, 15: 3, 512–532.

Kim, J. (1975). ‘Factor analysis’, in N. Nie, C. Hull, J. Jenkins, K. Steinbrenner and D. Brent (eds),Statistical Package for the Social Sciences, New York: McGraw-Hill.

Klaas, B. and DeNisi, A. (1989). ‘Managerial reactions to employee dissent: the impact of grievanceactivity on performance ratings’. Academy of Management Journal, 32: 705–717.

Kleinman, L. (2000). ‘Industrial relations’, in M. Melms (ed.), Encyclopedia of Management, Detroit, MI:Gale Group, pp. 402–406.

Lansbury, R., Davis, E. and Simmons, D. (1996). ‘Reforming the Australian workplace throughemployee participation’. Economic and Labour Relations Review, 7: 1, 29–45.

Leck, J. and Saunders, D. (1992). ‘Hirschman’s loyalty: attitude or behavior?’ Employee Responsibilitiesand Rights Journal, 5: 219–230.

Leigh, D. and Hills, S. (1987). ‘Male–female differences in the potential for union growth outsidetraditionally unionized industries’. Journal of Labor Research, 8: 2, 131–142.

Lewin, D. and Mitchell, D. (1992). ‘Systems of employee voice: theoretical and empiricalperspectives’. California Management Review, 34: 3, 95–111.

Lewin, D. and Peterson, R. (1988). The Modern Grievance Procedure in the United States, New York:Quorum Books.

Luchak, A. (2003). ‘What kind of voice do loyal employees use?’ British Journal of Industrial Relations,41: 1, 115–134.

McCabe, D. and Lewin, D. (1992). ‘Employee voice: a human resource management perspective’.California Management Review, 34: 3, 112–123.

McLoughlin, I. and Gourlay, S. (1994). Enterprises without Unions: Industrial Relations in the Non-unionFirm, Buckingham: Open University Press.

McMillan, A. (2000). ‘Group dynamics’, in M. Melms (ed.), Encyclopedia of Management, Detroit, MI:Gale Group, pp. 352–357.

Machin, S. and Wood, S. (2005). ‘Human resource management as a substitute for trade unions inBritish workplaces’. Industrial & Labor Relations Review, 58: 2, 201–218.

Marchington, M. (2006). ‘Employee voice systems’, in P. Boxall, J. Purcell and P. Wright (eds), TheOxford Handbook of Human Resource Management, Oxford: Oxford University Press, pp. 231–250.

Marchington, M., Wilkinson, A. and Ackers, P. (2001). Management Choice and Employee Voice, London:Chartered Institute of Personnel and Development.

Michaels, C. and Spector, P. (1982). ‘Causes of employee turnover: a test of the Mobley, Griffeth,Hand and Meglino model’. Journal of Applied Psychology, 67: 52–59.

Middlemist, R. and Hitt, M. (1988). Organization Behavior, St. Paul, MN: West Publishing Company.Millward, N., Bryson, A. and Forth, J. (2000). All Change at Work? British Employment Relations as

Portrayed by the Workplace Industrial Relations Survey Series, London: Routledge.Morehead, A., Steele, M., Alexander, M., Stephen, K. and Duffin, L. (1997). Changes at Work,

Melbourne: Longman.Morris, T., Lydka, H. and O’Creevy, M. (1993). ‘Can commitment be managed: a longitudinal analysis

of employee commitment and human resource policies’. Human Resource Management Journal, 3: 3,21–42.

Employee voice

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 201098

© 2009 Blackwell Publishing Ltd.

Page 20: Employee voice: does union membership matter?

Mowday, R., Steers, R. and Porter, L. (1979). ‘The measurement of organizational commitment’.Journal of Vocational Behavior, 14: 2, 224–247.

Newell, S. (2000). ‘Teams’, in M. Melms (ed.), Encyclopedia of Management, Detroit, MI: Gale Group,pp. 929–932.

Peetz, D. (1998). Unions in a Contrary World: The Future of the Australian Trade Union Movement,Cambridge: Cambridge University Press.

Pfeffer, J. (1994). Competitive Advantage through People, Boston, MA: Harvard Business School Press.Porter, L., Steers, R., Mowday, R. and Boulian, P. (1974). ‘Organizational commitment, job

satisfaction, and turnover amongst psychiatric technicians’. Journal of Applied Psychology, 59:603–609.

Rummel, R. (1970). Applied Factor Analysis, Evanston, IL: Northwest University Press.Spector, P. and O’Connell, B. (1994). ‘The contribution of personality traits, negative affectivity, locus

of control and type A to the subsequent reports of job stressors and job strains’. Journal ofOccupational and Organizational Psychology, 67 (March): 1–11.

Studenmund, A. and Cassidy, H. (1987). Using Econometrics: A Practical Guide, Boston, MA: Little,Brown and Company.

Tang, T. and Sarfield-Baldwin, L. (1996). ‘Distributive and procedural justice as related to satisfactionand commitment’. SAM Advanced Management Journal, 61: 3, 25–31.

Taras, D. and Kaufman, B. (2006). ‘Non-union employee representation in North America: diversity,controversy, and uncertain future’. Industrial Relations Journal, 37: 5, 513–542.

Terry, M. (1999). ‘Systems of collective employee representation in non-union firms in the UK’.Industrial Relations Journal, 30: 1, 16–30.

Van Dyne, L., Ang, S. and Botero, I. (2003). ‘Conceptualizing employee silence and employee voiceas multidimensional constructs’. Journal of Management Studies, 40: 6, 1359–1392.

Waring, P. (1999). ‘The rise of individualism in Australian industrial relations’. New Zealand Journalof Industrial Relations, 24: 3, 291–318.

Watson, D., Pennebaker, J. and Folger, R. (1987). ‘Beyond negative affectivity: measuring stress andsatisfaction in the workplace’. Journal of Organizational Behavior Management, 8: 141–157.

Webb, S. and Webb, B. (1897). Industrial Democracy, London: Longman, Green and Co.Wilkinson, A., Dundon, T., Marchington, M. and Ackers, P. (2004). ‘Changing patterns of employee

voice: case studies from the UK and Republic of Ireland’. Journal of Industrial Relations, 46: 3,298–322.

Wilkinson, A., Dundon, T. and Grugulus, I. (2007). ‘Information but not consultation: exploringemployee involvement in SMEs’. International Journal of Human Resource Management, 18: 7,1279–1297.

Willman, P., Bryson, A. and Gomez, R. (2007). ‘The long goodbye: new establishments and the fallof union voice in Britain’. International Journal of Human Resource Management, 18: 7, 1318–1334.

Withey, M. and Cooper, W. (1989). ‘Predicting exit, voice, loyalty and neglect’. Administrative ScienceQuarterly, 34: 521–539.

Yates, C. (2006). ‘Challenging misconceptions about organizing women into unions’. Gender, Work andOrganization, 13: 6, 565–584.

John Benson and Michelle Brown

HUMAN RESOURCE MANAGEMENT JOURNAL, VOL 20 NO 1, 2010 99

© 2009 Blackwell Publishing Ltd.