issued march 2004 report no. 18 the response census · pdf filethe project manager was kevin a...

Census 2000 SynthesisReport No. 18

TR-18

Issued March 2004

Results Fromthe ResponseMode andIncentiveExperimentin 2000

U.S.Department of CommerceEconomics and Statistics Administration

U.S. CENSUS BUREAU

Census 2000 Testing, Experimentation, and Evaluation Program

The Census 2000 Evaluations Executive SteeringCommittee provided oversight for the Census 2000Testing, Experimentation, and Evaluations (TXE)Program. Members included Cynthia Z. F. Clark,Associate Director for Methodology and Standards;Preston J. Waite, Associate Director for DecennialCensus; Carol M. Van Horn, Chief of Staff; TeresaAngueira, Chief of the Decennial ManagementDivision; Robert E. Fay III, Senior MathematicalStatistician; Howard R. Hogan, (former) Chief of theDecennial Statistical Studies Division; Ruth AnnKillion, Chief of the Planning, Research and EvaluationDivision; Susan M. Miskura, (former) Chief of theDecennial Management Division; Rajendra P. Singh,Chief of the Decennial Statistical Studies Division;Elizabeth Ann Martin, Senior Survey Methodologist;Alan R. Tupek, Chief of the Demographic StatisticalMethods Division; Deborah E. Bolton, AssistantDivision Chief for Program Coordination of thePlanning, Research and Evaluation Division; Jon R.Clark, Assistant Division Chief for Census Design ofthe Decennial Statistical Studies Division; David L.Hubble, (former) Assistant Division Chief forEvaluations of the Planning, Research and EvaluationDivision; Fay F. Nash, (former) Assistant Division Chieffor Statistical Design/Special Census Programs of theDecennial Management Division; James B. Treat,Assistant Division Chief for Evaluations of the Planning,Research and Evaluation Division; and VioletaVazquez of the Decennial Management Division.

As an integral part of the Census 2000 TXE Program,the Evaluations Executive Steering Committee char-tered a team to develop and administer the Census2000 Quality Assurance Process for reports. Past andpresent members of this team include: Deborah E.Bolton, Assistant Division Chief for ProgramCoordination of the Planning, Research and EvaluationDivision; Jon R. Clark, Assistant Division Chief forCensus Design of the Decennial Statistical StudiesDivision; David L. Hubble, (former) Assistant DivisionChief for Evaluations and James B. Treat, AssistantDivision Chief for Evaluations of the Planning, Researchand Evaluation Division; Florence H. Abramson,Linda S. Brudvig, Jason D. Machowski, andRandall J. Neugebauer of the Planning, Research and Evaluation Division; Violeta Vazquez of theDecennial Management Division; and Frank A.Vitrano (formerly) of the Planning, Research andEvaluation Division.

The Census 2000 TXE Program was coordinated by thePlanning, Research and Evaluation Division: Ruth AnnKillion, Division Chief; Deborah E. Bolton, AssistantDivision Chief; and Randall J. Neugebauer andGeorge Francis Train III, Staff Group Leaders. KeithA. Bennett, Linda S. Brudvig, Kathleen HaysGuevara, Christine Louise Hough, Jason D.Machowski, Monica Parrott Jones, Joyce A. Price,Tammie M. Shanks, Kevin A. Shaw,

George A. Sledge, Mary Ann Sykes, and CassandraH. Thomas provided coordination support. FlorenceH. Abramson provided editorial review.

This report was prepared under contract by RachelCaspar of the Research Triangle Institute. The projectmanager was Kevin A. Shaw of the Planning, Researchand Evaluation Division. The following authors andproject managers prepared Census 2000 experimentsand evaluations that contributed to this report:

Planning, Research and Evaluation Division:Jennifer Guarino Tancreto

Statistical Research Division:Lawrence A. Malakhoff

Independent contractor:David Cantor, Westat, Inc.

Greg Carroll and Everett L. Dove of the Admin-istrative and Customer Services Division, and WalterC. Odom, Chief, provided publications and printingmanagement, graphic design and composition, and edi-torial review for print and electronic media. Generaldirection and production management were providedby James R. Clark, Assistant Division Chief, andSusan L. Rappa, Chief, Publications Services Branch.

Acknowledgments

U.S. Department of CommerceDonald L. Evans,

Secretary

Vacant,Deputy Secretary

Economics and Statistics AdministrationKathleen B. Cooper,

Under Secretary for Economic Affairs

U.S. CENSUS BUREAUCharles Louis Kincannon,

Director

Census 2000 SynthesisReport No. 18


RESULTS FROM THERESPONSE MODE AND

INCENTIVE EXPERIMENTIN 2000

TR-18

Issued March 2004

Suggested Citation

Rachel CasparCensus 2000 Testing,

Experimentation, and EvaluationProgram Synthesis

Report No. 18, TR-18, Results From the Response Mode and

Incentive Experiment in 2000, U. S. Census Bureau,

Washington, DC 20233 ECONOMICS

AND STATISTICS

ADMINISTRATION

Economics and StatisticsAdministration

Kathleen B. Cooper,Under Secretary for Economic Affairs

U.S. CENSUS BUREAU

Charles Louis Kincannon,Director

Hermann Habermann,Deputy Director and Chief Operating Officer

Cynthia Z. F. Clark,Associate Director for Methodology and Standards

Preston J. Waite, Associate Director for Decennial Census

Teresa Angueira, Chief, Decennial Management Division

Ruth Ann Killion, Chief, Planning, Research and Evaluation Division

For sale by the Superintendent of Documents, U.S. Government Printing Office

Internet: bookstore.gpo.gov Phone: toll-free 866-512-1800; DC area 202-512-1800

Fax: 202-512-2250 Mail: Stop SSOP, Washington, DC 20402-0001

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 iii

Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .v

Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

1.1 Experiment components . . . . . . . . . . . . . . . . . . . . . . . . .5

1.2 Research questions to be answered . . . . . . . . . . . . . . . . .5

2. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

2.1 Research plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

3. Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11

4. Major Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13

4.1 Effect of the incentive on response rates—initialmailout component . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13

4.2 Effect of the response mode on response rates . . . . . . . .14

4.3 Item nonresponse rates by mode of response . . . . . . . . .15

4.4 Results from the Interactive Voice RecognitionQuestionnaire Satisfaction Survey . . . . . . . . . . . . . . . . . .15

4.5 Results from the nonresponse component of the Response Mode and Incentive Experiment . . . . . . . . . . . .16

4.6 Results from the Internet Usage Survey . . . . . . . . . . . . . .18

5. Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21

5.1 Recommendations based on the Response Modeand Incentive Experiment . . . . . . . . . . . . . . . . . . . . . . . .21

5.2 Recommendations for future research . . . . . . . . . . . . . . .23

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25

APPENDIX A: Response Mode and Incentive Experiment Design .A-1

APPENDIX B: Inserts Included With RMIE Mailings . . . . . . . . . . .B-1

APPENDIX C: Final ASQ 2000 Short Form Script . . . . . . . . . . . . .C-1

APPENDIX D: ASQ Usability Questions . . . . . . . . . . . . . . . . . . .D-1

APPENDIX E: Telephone Interviewing Specificationsfor Short Form . . . . . . . . . . . . . . . . . . . . . . . . . . .E-1

APPENDIX F: Internet Census Form . . . . . . . . . . . . . . . . . . . . . .F-1

APPENDIX G: Internet Usage Survey Questionnaire . . . . . . . . . . .G-1

Contents

iv Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

LIST OF TABLES

Table 1 Mode Specific Response Rates, Sample Sizes, and Response Rate Differences Among Modes and Across Incentive Groups . . . . . . . . . . . . . . . . . . . . . . . .16

Table 2 Mode Specific Response Rates, Sample Sizes, and Pairwise Differences Between Incentive and No Incentive Groups Within and Across Response Modes . . .16

Table 3 Logistic Regression Coefficients Predicting the Log Odds of Responding to the Census Through the Assigned Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

Table 4 Logistic Regression Coefficients Predicting the Log Odds of Receiving the Incentive Among Respondents . . .17

Table 5 Overall Response Rates, Sample Sizes, and Pairwise Differences Between Incentive and No Incentive GroupsWithin and Across Response Modes . . . . . . . . . . . . . . . .18

Table 6 Internet Access Rates by Location . . . . . . . . . . . . . . . . .19

Table 7 Reasons Respondents Did Not Use the Internet by Coverage Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19

Table 8 Percentage of Respondents Not Aware of/Not Offered the Incentive Who Would Use the Internet by Incentive Amount . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19

List of Figures

Figure 1 Assigned Mode Response Rate: Combined Panels . . . . .13

Figure 2 Internet Access Rates Among Mail Respondentsby Coverage Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 v

The Census 2000 Testing, Experimentation, and Evaluation Programprovides measures of effectiveness for the Census 2000 design,operations, systems, and processes and provides information on the value of new or different methodologies. By providing measuresof how well Census 2000 was conducted, this program fully sup-ports the Census Bureau’s strategy to integrate the 2010 planningprocess with ongoing Master Address File/TIGER enhancements andthe American Community Survey. The purpose of the report that follows is to integrate findings and provide context and backgroundfor interpretation of related Census 2000 evaluations, experiments,and other assessments to make recommendations for planning the 2010 Census. Census 2000 Testing, Experimentation, andEvaluation reports are available on the Census Bureau’s Internet siteat: www.census.gov/pred/www/.

Foreword

This page intentionally left blank.

The Response Mode and IncentiveExperiment investigated the impactof three computer-assisted datacollection techniques – Computer-Assisted Telephone Interviewing,Internet, and Interactive VoiceResponse – on the response rateand data quality in Census 2000.Households participating in thestudy were randomly assigned tosix panels and to a control group.The households in the six panelswere given the choice of providingtheir Census 2000 data via theusual paper forms or by one of thealternate computer-mediatedresponse modes. Half of the panelswere offered an incentive, a tele-phone calling card good for 30minutes of calls, for using thealternate response mode.

In addition, the experiment includ-ed a nonresponse componentdesigned to assess the effects of apromised incentive and alternativeresponse mode options onresponse among a sample of cen-sus households who failed toreturn their census forms by April26, 2000. The intent of the nonre-sponse component was not to testincentives or response modeoptions as possible nonresponseconversion techniques for the cen-sus. Rather, the experiment wasdesigned to test the effect of thesefactors on response among agroup representing those who aretraditionally difficult to enumerate.

A final component of the experi-ment involved interviewing house-holds assigned to the Internetmode (both with and without theincentive) who opted to completethe traditional paper census form.

The purpose of the interview wasto determine why these house-holds did not use the Internet.

Results from the initial mailoutportion of the Response Mode andIncentive Experiment show that:

• Computer-assistedTelephone Interviewingbrought about a small butstatistically significantimprovement in the overallresponse rate. It also had alow item nonresponse rate.However, in the context of thisexperiment, it entailed substan-tial cost for hardware, software,and programmer and interview-er labor.

• The Internet mode yieldedrelatively high data quality.The primary additional costassociated with this modeinvolved the development andmaintenance of the softwareand hardware. The benefits ofthis data collection method mayoutweigh these costs.

• The implications of thisstudy are complex for theuse of the Interactive VoiceResponse technology. Dataquality was the lowest for thismode. Respondents appeared todislike lengthy surveys with thismethod and some respondentsub-groups (mixed race respon-dents and Hispanics) were morelikely to report confusion withthe task. Nonetheless, thismode is an appealing way toreach persons with limited liter-acy skills. The costs associatedwith this mode included the

hardware, programming, speechrecognition software, and tele-phone expenses.

• The calling card was veryeffective in promoting theuse of the alternativeresponse mode. However,rather than encouraging morehouseholds to participate, theincentive tended to redirecthouseholds that would haveresponded by mail to the alter-nate computer-mediatedresponse mode. This effect maybe partially attributable to thecolorful inserts in the householdmailing that directed attentionto the calling card.

• The impact of the callingcard may not justify its cost.In the Internet and computer-assisted telephone interviewingconditions, the incentive mayhave brought about an increasein responding via the alternatemode, but this increase was off-set by decreases in respondingby mail.

Results from the nonresponse com-ponent of the Response Mode andIncentive Experiment show that:

• Computer-assisted telephoneinterviewing elicited thehighest response fromCensus nonrespondents (7.8 percent) followed by the Interactive VoiceRecognition Questionnaire(4.8 percent) and theInternet (3.7 percent). Thiscomparison is confounded bythe fact that Internet access may

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 1

Executive Summary

be especially problematic forthis target population.

• Respondents to theInteractive Voice Responsemode are significantlyyounger and reside inhouseholds with, on aver-age, fewer people than bothmail and computer-assistedtelephone interview respon-dents. Computer-assisted tele-phone interview respondentsare disproportionately Blackwith more households residingin low coverage areas comparedto Internet respondents.

• The calling card incentiveincreased response to thealternative modes by 1.9percent across all responsemodes.

• Person 1 in householdsreceiving the incentive dueto alternative responsemode participation tended tobe younger than Person 1 inhouseholds not receiving theincentive.

• Contrary to past research,the increase in response dueto the incentive is not statis-tically different in areas withhigh concentrations of theBlack and Hispanic popula-tions and renters (1.9 per-cent) from other areas (2.0percent).

• When total response to anexperimental second mailingis considered, no significantincentive effect remains.That is, when mail responsesare included as respondents, theincentive group (13.8 percent) isno more likely to respond thanthe non-incentive group (13.2percent). Similar to the initialmailout experiment, it appearsthat the incentive merely redi-rects responses that would have

otherwise been obtained by mailto alternative modes.

• Irrespective of the experi-mental treatments, around13 percent participation wasobtained from cases that didnot initially return the ques-tionnaire or returned thequestionnaire late.Replacement questionnaireswere not included in the secondmailing, implying that respon-dents who returned a mail form,around six to nine percent, usedtheir original questionnairemailed in March 2000.

Finally, results from the InternetUsage Survey indicate that:

• Approximately 63 percent ofthe Internet Usage Surveysample reported havingaccess to the Internet. Thus,access does not appear to be amajor reason why these censusrespondents did not opt to com-plete their census form via theInternet.

• Nearly half of the InternetUsage Survey respondentsreported they were unawarethat the Internet was anoption for completing theircensus forms.

• Among respondents whowere aware of the Internetoption, 35 percent reportedthat they believed the papercensus form would be easierto complete. Other reasons fornot using the Internet included:no access to a computer, con-cerns about privacy, forgot theInternet was an option, andinsufficient knowledge of theInternet.

• Respondents reported thatan incentive to complete thecensus via the Internetwould have encouragedthem to use this alternative

mode. About 41 percent ofrespondents who were notoffered the incentive or wereunaware of the offer said theywould fill out their census formvia the Internet if they wereoffered a 30 minute calling card.Another nine percent indicatedthey would do it for a 60 minutecalling card, and an additional12 percent would be willing if a90 minute calling card wasoffered.

Based on the findings of theResponse Mode and IncentiveExperiment, the following recom-mendations are made:

• The Internet is an attractivealternative data collectionmode for the decennial cen-sus. Although no formalcost/benefit analysis was com-pleted, it seems likely that thecost of developing and support-ing a web-based application forCensus 2010 would be less thanthe costs associated with thedata processing required for thepaper forms that would bereturned from households whowould have been willing to pro-vide their data via the Internet.As internet accessibility andusage continues to expand,additional savings could be real-ized.

• The use of an incentive wasan effective means of pro-moting the use of the alter-native response modes.Comparisons between the incen-tive and no-incentive conditionsin the initial mailout experimentreveal that the incentive wasassociated with three to four-fold increases in the rate ofusing the alternative mode.However, some of this effectmay be attributable to the useof the insert which drew therespondent’s attention to the

2 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

availability of the alternativemode.

• Data quality was improvedfor the computer-assistedtelephone interviewing mode(as compared with mail).However, this mode entails sub-stantial cost investments forhardware, software, and pro-grammer and interviewer time.

• Without significant improve-ments in the voice-userinterface, the InteractiveVoice Response technologyis probably not a viablealternative for Census 2010.Data quality was the lowest ofall the response modes. Thisoccurred primarily as a result ofrespondents hanging up beforethey had provided completedata. When this occurred, eventhe partial information that hadbeen provided was deleted,resulting in a significant loss ofdata. In addition, the costsassociated with developing thistype of system are sizeable.

• The use of alternativeresponse modes does notincrease overall responserates to the census. Rather, itshifts households who wouldrespond via the paper census tothe other modes. This patternholds true for groups who aretraditionally difficult to enumer-ate in the census, as evidencedby the results of the nonre-sponse component of thisexperiment.

Results from the Response Modeand Incentive Experiment suggestseveral areas worthy of futureresearch:

• Research is needed to deter-mine the best ways to pres-ent the response mode alter-natives, as it appears thatsome respondents assignedto the no-incentive treat-ments did not read the letterthat accompanied theirpaper census form inform-ing them of the alternativemode option. The use of a

colorful mailing insert, irrespec-tive of whether an incentive isoffered may be enough toattract respondents to an alter-native census mode. However,this information cannot bedetermined from the dataobtained from this experiment.

• Research is needed to deter-mine whether recentadvances in speech recogni-tion software can improvethe voice user interface toincrease data quality andeliminate some of the dissat-isfaction voiced by respon-dents who answered theInteractive Voice RecognitionQuestionnaire satisfactionsurvey.

• The choice of incentiveshould be revisited. Basedon the number of respondentswho never used their callingcard once they were activated, itappears that the card may nothave been a powerful incentive.



1. Introduction

The potential benefits of usingInternet, Computer-assistedTelephone Interviewing (CATI),and/or Interactive Voice Response(IVR) surveys for the census canonly be realized if large numbersof respondents are willing toanswer survey questions usingthese computer-assisted data col-lection methods. The objective ofthe Response Mode and IncentiveExperiment (RMIE) was to investi-gate the effect of these technolo-gies in Census 2000.

The specific goals of the RMIEstudy were:

• To assess the public’s willing-ness to provide census datausing these computer-mediateddata collection methods;

• To evaluate the quality of thedata collected using these meth-ods; and

• To study the ability of incen-tives, in the form of telephonecalling cards, to promote theuse of these computer-mediatedmethods.

1.1 Experimentcomponents

The RMIE has three basic compo-nents. The first is the initialmailout. Census 2000 forms weredelivered to all households in theUnited States beginning in mid-March of 2000. A sample of thehouseholds that received the shortform were randomly selected, priorto the mailout, for the RMIE. Thissample was stratified into one oftwo areas based on the geographi-cal location of the household.

Some of the households in the ran-dom sample served as the CensusControl Group (CCG); each of thesehouseholds received a form andletter identical to those used in thenational Census 2000 mailing. Therest of the households in the sam-ple received special instructions,giving them the choice of provid-ing their census data either by fill-ing out the paper form, or byusing a computer-assisted method:

• One subsample of the house-holds was given the option ofproviding their census data via aCATI.

• A second subsample was giventhe option of providing theircensus data via an IVR system.

• A third subsample was given thechoice of providing their dataon a web-based survey.

Half of the households in each ofthese three experimental condi-tions were offered telephone call-ing cards as an incentive to usethe computer-assisted method toreport their census data.

The second component of the RMIEwas an operation to follow up withthe nonrespondents of the CCG.Households in the CCG that failedto mail back their census forms—that is, the nonrespondents to theinitial mailout—were given theopportunity to provide their censusdata using one of the three com-puter-assisted methods. Half ofthese nonrespondents were offeredthe calling card incentive to use acomputer-assisted method. Thus,the design of this nonresponse(NR) phase of the RMIE was very

similar to the design of the initialmailout component. Appendix Aprovides a layout of the RMIEdesign for these first two compo-nents (sample sizes are shown inparentheses).

The third component of the RMIEwas an Internet Usage Survey (IUS).This telephone survey involved asample of the households thatwere offered the opportunity to fillout the Internet version of the cen-sus short form in the initial mailoutbut either mailed in their data onthe paper form or called the opera-tor assistance (OA) number andprovided their census data to atelephone interviewer. The Internetusage survey explored the reasonswhy these households chose not toprovide their information using theweb-based survey.

The advance letter and reminderpostcard to RMIE households wereincluded in the nationwide mailing.RMIE households that requested aspecial language form were exclud-ed from the RMIE data analysis.

1.2 Research questionsto be answered

The RMIE was designed to addressthe following research questions:

• What effect does an incentivehave on census response behav-iors (both overall response aswell as item response)?

• What effect does an alternativeresponse mode have on censusparticipation rates (both overallresponse as well as itemresponse)?

• What effect does an incentive

have on census response by

alternative electronic response

modes for typical census nonre-

spondents?

• What effect does an incentive

have on census participation

across the various response

mode options and subpopula-

tions that historically differ with

regard to census participation?

• What reasons do respondents

give for choosing to provide

their census information using

the paper form rather than viathe Internet?

A fuller discussion of the goals andobjectives of the RMIE can befound in the Program Master Planprepared by Malakhoff and Sanders(2000).

The RMIE was appropriatelydesigned to allow the researchersto determine the independenteffect of an incentive and an alter-native response mode on participa-tion rates and data quality.Assigning nonrespondents to theCCG to treatment groups allowed

for a further understanding of therole that incentives and alternativeresponse modes play in persuadingtraditionally reluctant censushouseholds to participate. Finally,the inclusion of the IUS allows fora fuller understanding of the barri-ers, both actual and perceived, thatmust be overcome to make theInternet a viable option for thenext census. Given the likely costreductions that could be realized infielding the census if a significantproportion of households respond-ed via the Internet, the results ofthe IUS are especially important.


2.1 Research plan

A total of 35,377 households wererandomly selected for this studyfrom the Decennial Master AddressFile (DMAF) developed for Census2000. All of these households werefrom the 94.3 million householdsin mailout/mailback areas(Households that were selected forthe Accuracy and CoverageEvaluation initial and final sampleswere not eligible for selection.). Allhouseholds selected were sched-uled to receive the short form.

Of the households selected for thisstudy, 15,738 were randomly dis-persed among six panels in a threeby two, fully factorial design toform the initial mailout componentof the RMIE experiment. The firstfactor, response mode, had threelevels: CATI, IVR, and Internet. Thehouseholds were given the choiceof providing their census dataeither via U.S. mail on the usualpaper forms, or via their assignedcomputer-assisted response mode.

The second factor, the incentive,had two levels: incentive and noincentive. Households in the incen-tive condition were rewarded forusing a computer-assistedresponse mode to provide theircensus data, while those in the no-incentive condition were not. Thereward was a telephone callingcard.

The six panels and the number ofhouseholds assigned to each wereas follows:

Panel 1:CATI with no incentive 2,621

Panel 2: IVR with no incentive 2,621

Panel 3: Internet with no incentive 2,627

Panel 4: CATI with incentive 2,622

Panel 5: IVR with incentive 2,623

Panel 6: Internet with incentive 2,624

2.1.1 Mailings

The Census Bureau mailed a shortform for Census 2000 and a coverletter to each household in thisstudy at the same time censusforms were mailed to all house-holds in the nation. Appendix Bcontains copies of the RMIE mail-ings. The cover letter explainedthat the household could providecensus data in either of two ways.First, the household could mail inthe data in the usual way, usingthe paper form. Alternatively, thehousehold could use a computer-assisted method. The cover lettersto panels 1 and 4 explained thatthe household could provide dataover the telephone by dialing atoll-free number. The cover lettersto panels 2 and 5 also explainedthat the household could providedata by telephone by calling a toll-free number. Neither letter men-tioned how the data would be col-lected once the household placedthe call. The cover letters to panels3 and 6 explained that the house-hold could provide data via a web-based questionnaire available atwww.2000.census.gov.

The mailings to panels 4, 5, and 6

(the incentive panels) contained an

insert, printed in color on heavy

stock paper. The calling card was

attached to this insert. The cover

letter and insert explained that if

the household provided its census

data using the computer-assisted

method, the calling card would be

activated, giving it a value worth

30 minutes of domestic calls.

The paper census forms sent to

the households in all six panels

provided a toll-free number for any

questions. This number was differ-

ent from the toll-free help line

number that appeared on the stan-

dard Census 2000 forms received

by households that were not

assigned to the RMIE. This source

of help and information was called

“Operator Assistance” or simply

“OA.” Operators were available at

that number to answer questions

both about this study and about

Census 2000 generally.

Mailed questionnaires were

returned to the Jeffersonville Data

Capture Center (DCC) at the

National Processing Center (NPC).

At the initial barcode reading,

these questionnaires were identi-

fied and automatically sorted to

the special data processing unit in

NPC. Members of this unit were

responsible for keying the census

data directly from the paper forms.

This differs from the method of

data capture used for the regular

census forms which employs

image data capture.


2. Methodology

2.1.2 Census control group

The remaining 19,639 householdsthat were selected for this studycomprised the Census ControlGroup (CCG). The CCG receivedmailings that contained a cover let-ter and a census short form. Themailings did not offer the CCGhouseholds the opportunity to pro-vide census data using a computer-assisted response mode, nor didthe mailings offer any type ofincentive. The CCG served as agroup against which the six panelsin this study could be compared. Inaddition, households in the CCGthat failed to provide their censusdata were involved in the secondphase of the RMIE; the nonre-sponse component. Of the CCG, atotal of 6,130 households failed toreturn their census form by April26, 2000 and thus comprised thesample for the nonresponse com-ponent of the RMIE. These house-holds were randomly assigned topanels 7A - 9A and panels 7B - 9Bas shown in Appendix A. A sec-ond mailed package was sent toeach of these households. Thesehouseholds had the option ofanswering Census 2000 via thestandard paper questionnaire origi-nally sent to the household; how-ever, replacement questionnaireswere not included in this secondmailing and calling cards were notactivated for households thatreturned paper questionnaires.

Because the households in the CCGthat failed to provide their censusdata were included in the nonre-sponse component, census formsfor all CCG households listed thespecial OA number for RMIE ratherthan the standard toll-free assis-tance number printed on theCensus 2000 short forms. Exceptfor the OA telephone number, themailings received by the CCG wereidentical to the official Census2000 short form and cover letter.

When CCG households had ques-tions about the nonresponse phaseand called the RMIE OA number,they reached an operator who wasknowledgeable both about RMIEand about Census 2000 generally.As a courtesy, these operatorscould also collect census data ifcallers specifically requested toprovide their information duringthe call.

2.1.3 Stratification

Each household selected for thisstudy was classified as being fromone of two strata: a low coveragearea (LCA) or high coverage area(HCA). The LCA was comprised ofcensus tracts with high concentra-tions of non-White residents andrenters, two groups associatedwith high nonresponse rates.About 19.3 percent of the house-holds in the DMAF in mailout/mail-back areas are in the LCA; the HCAconsists of the remaining house-holds. In RMIE, households wereproportionately selected from thetwo strata; just under one-fifth ofthe households in each panel andin the CCG were in the LCA stra-tum.

2.1.4 Interactive voice recognitionquestionnaire

Only households assigned to pan-els 2 and 5 were informed of theIVR system telephone number inthe initial mailout phase.Therefore, calls to the IVR systemcame only from householdsassigned to those two panels. Theprotocol for the IVR Questionnaireis included as Appendix C. TheIVR Questionnaire was available toreceive calls 24 hours a day.

The IVR Questionnaire closely fol-lowed the paper Census 2000short form. However, unlike thepaper census short form, the IVRQuestionnaire allowed the collec-tion of information about all

members of a household, no mat-ter how many there were. In con-trast, the paper short form askedfor information about only six per-sons in the household; it collectedonly the first and last names of theseventh through the twelfth per-sons, and no information at all forany persons beyond the twelfth.

The respondent answered nearlyall questions in the IVRQuestionnaire by speaking. Theexceptions were the questions ask-ing for the household’s telephonenumber, the 22-digit census identi-fication number, and the ten-digitcalling card number (for panel 5only). The respondents providedthese data by pressing the touch-tone keys on their telephones.However, respondents who werenot using a telephone with touch-tone keys provided this informa-tion verbally.

Immediately after respondentsentered their 22-digit census iden-tification numbers, the systemdetermined whether the respon-dents had called the system previ-ously. If a respondent had calledearlier, the system transferred thecall to a CATI operator who collect-ed any updated information fromthe respondent. The IVR systemalso transferred a call to a CATIoperator if the respondent did anyof the following:

• Failed to provide the 22-digitcensus ID when asked;

• Attempted to enter the censusID with a pulse telephone;

• Entered a census ID that wasnot in the databases for panels2 or 5; or

• Indicated he/she was unable towork with the system properly

The CATI operator helped the callerfind the correct 22-digit number



and then collected the caller’s cen-sus data.

When the speech recognition soft-ware attempted to recognize anutterance, it returned a confidencelevel associated with the recogni-tion attempt. The level wasexpressed as a percentage, gener-ally between 50 and 100.Recognition attempts with highconfidence levels were more cer-tain than attempts with low confi-dence levels.

If the software returned a confi-dence level under 70 percent in anattempt to recognize a “yes” or a“no” response, the system repeatedthe question. If the software stillcould not adequately recognize theresponse in this second attempt,the system transferred the call to aCATI operator, who administeredthe questionnaire. If no CATI oper-ator was available at the time thatthe call was transferred, or if thetransfer occurred after CATI work-ing hours, the respondent heard arecorded message, left a name andtelephone number, and received acall from a CATI operator later.

Some questions in the IVRQuestionnaire, such as “Please tellus the month, day and year thisperson was born” required spokenresponses that were more complexthan a simple “yes” or “no.” Thesystem was not programmed torecognize these responses in “realtime.” Instead, the system record-ed these responses so they couldbe transcribed soon afterward. TheCATI operators transcribed theserecorded responses during periodswhen they were not taking CATIcalls.

At the end of the IVRQuestionnaire, the respondentswere given the opportunity tochange any of their responses toany question. The transcriptionists

listened to these changes andupdated the data accordingly.

The IVR Questionnaire concludedwith a set of questions to assessthe respondent’s satisfaction withthe data collection method. Thesequestions are summarized inAppendix D. In addition, timingdata from the IVR Questionnairewere also retained for analysis.These data included the totalamount of time required for thehousehold to complete the IVRQuestionnaire and the mean timerequired to answer individual sur-vey items.1

2.1.5 Computer-assisted telephoneinterview

Persons from households that wereselected for this study could reacha CATI operator in three ways:

• Calls to the IVR system weretransferred to a CATI operatorwhen the speech recognizercould not adequately recognizethe respondent’s responses tocertain questions, or when therespondent entered a censusidentification number that wasinvalid or that belonged to ahousehold that had already pro-vided data.

• Households in panels 1 and 4could dial the toll-free numberto reach a CATI operator.

• Respondents in households inany panel could call the OA tele-phone line and offer to providetheir data. Even though the OAnumber was offered primarily tohelp respondents with questionsabout this study or about thecensus generally, some respon-dents did call the OA number

and ask to provide their censusdata. The OA operator trans-ferred these calls to a CATI oper-ator who collected the dataregardless of panel assignment.

Callers heard a recorded messageif they reached CATI during the latenight or early morning or when alloperators were unavailable. Themessage asked the callers to leavetheir names, telephone numbers,and the times that they might beavailable for a return call. A CATIoperator later called the respon-dent to collect the census data.

At the start of the interviews, theCATI operators first ascertainedwhether the caller could speakEnglish. If the caller could speakonly Spanish, the operator trans-ferred the call to a bilingual opera-tor. If a respondent who spoke nei-ther English nor Spanish called, theCATI operator could not collect anydata. Since no communication waspossible with these few callers,they were not considered respon-dents, and had no follow-up con-tact. If the caller could speakEnglish, the operator began theinterview by asking the caller toread the 22-digit census identifica-tion number from the mailinglabel. The operator administeredthe CATI interview after verifyingthat the identification number wasfrom a household in this study.The content of the CATI interviewclosely followed the content of theCensus 2000 short form. However,like the IVR Questionnaire, theCATI interview collected completeinformation about all persons inthe household, no matter howmany persons lived there. The pro-tocol for the CATI interview isincluded as Appendix E.

2.1.6 Internet questionnaire

Census Bureau staff developed and provided the Internet-basedquestionnaire for the RMIE.

1This time includes the time required forthe system to play the question, the respon-dent to answer, two seconds to determine ifthe response is completed, and the speechrecognition software to compute theresponse.


Respondents answered multiple-

choice questions in the question-

naire by clicking the appropriate

radio buttons and checkboxes.

They answered text-entry ques-

tions by typing their answers into

response fields. The questionnaire

screens were designed to resemble

the short form paper question-

naire. The screens were not pro-

grammed with any branching logic

or data validity checks. The

Internet survey was available 24

hours a day. A printout of the sur-

vey appears in Appendix F.

2.1.7 Internet usage survey

The sample for the IUS was select-

ed from those respondents in the

internet panels of the RMIE who

responded via mail or CATI2

(through a phone transfer from

OA). The frame from which the

IUS sample was drawn included

293 households from panel 3

(internet, no incentive) and 277

households from panel 6 (internet

with incentive). Since the original

RMIE sample was selected with

proportional allocation to stratum,

it was anticipated that the IUS sam-

ple would be selected in the same

manner. However, this selection

methodology would have resulted

in a very small sample size in the

LCA strata due to nonresponse to

the original mailing. Therefore,

systematic sampling using equal

allocation was conducted. The

resulting sample included 318

cases in the HCA and 252 in the

LCA. The IUS Questionnaire is

included in Appendix G.

2Only seven CATI interviews werereceived in panels 3 and 6. These caseswere selected into this sample with certainty.


As can happen in even the mostcarefully designed experiments,technical problems occurred overthe course of the RMIE which bearmentioning here. The most signifi-cant of these were problems thataffected the representativeness ofthe sample in the IVR panels (pan-els 2 and 5):

• When the IVR system first beganaccepting calls, a software prob-lem in a lookup routine causedthe system to inaccurately clas-sify all of the callers as oneswho had called before. The sys-tem therefore failed to adminis-ter the IVR Questionnaire andinstead directed the calls inap-propriately to the CATI opera-tors immediately after therespondents entered their cen-sus identification numbers. Thisproblem began with the first callto the IVR system and wasresolved within just a few days.The first 115 calls to the IVRsystem (110 from panel 5 andfive from panel 2) were affected.

• Once analysis of the data began,a serious problem was discov-ered. The response rate forpanel 2 (IVR - no incentive)appeared to be very low. Thisinexplicable effect dwarfed allother observed effects andappeared to be an artifact ofsome error. Moreover, the pro-portion of mailings returned asUndeliverable As Addressed(UAA) was much lower for panel2 than for any other panel.Further investigation revealedan apparent problem with themailout for panel 2. With veryfew exceptions, no responses

were received, nor were anymailings returned UAA, forpanel 2 mailings to householdsin Missouri, Kansas, Nebraska,Louisiana, and Arkansas (thefive states whose ZIP Codesstart with 630 to 729), Hawaii,Oregon, Washington, and Alaska(the four states whose ZIPCodes start with 967 to 999),and ZIP Codes 39301 to 39648,60202 to 60490, and 95608 to95833. Similar problems werenot detected for any otherpanel. The Census Bureau inves-tigated this situation and foundthat some responses did in factcome in from households inthese areas, but they arrived toolate to be counted. Apparently,the mailout to these areas waseither delayed or not sent, pre-venting the affected householdsfrom responding before the cut-off date.

• For panels 1 and 3, the propor-tion of UAA returns was morethan two times higher for thestate of Indiana than for anyother state. In panel 1, Indianahad ten responding households,six nonresponding households,and 51 UAAs. In panel 3,Indiana had 13 respondinghouseholds, no nonrespondinghouseholds, and 55 UAAs.These UAA rates were by far thehighest observed for any statein any panel. The UAA rate forthe entire nation for panels 1and 3 were respectively 10.5and 11.0 percent. For Indianaalone the rates were respective-ly 76.1 and 80.9 percent.

The data were examined afterremoving all data from Indiana,Missouri, Kansas, Nebraska,Louisiana, Arkansas, Hawaii,Oregon, Washington, and Alaska,and ZIP Codes 39301 to 39648,60202 to 60490, and 95608 to95833. Chi square analysisrevealed that the UAA rate differedamong the six panels, even withthese areas excluded (chi square =10.13, df = 5, p < 0.073). Furthertests revealed that this effect wasentirely attributable to panel 2.The UAA rate for panel 2 was sig-nificantly lower than the rate for allthe other panels (chi square =7.62, df = 1, p < 0.006). No suchsignificant effect was found for anyother panel. Thus, even withoutthe ten problematic states and thethree problematic ZIP Code areas,the UAA rate for panel 2 was sig-nificantly depressed. This findingsuggests that problems may stillexist with the mailout for panel 2,even after the problematic statesand ZIP Code areas are eliminated.

Based on these findings, theCensus Bureau decided that twosets of analyses would be complet-ed. Method 1 involved analyzingdata for only four of the six pan-els; panel 2 is excluded because ofthe mailout problems, and panel 5,the other IVR panel, is also beexcluded to maintain a balanced,factorial design. All households inthe remaining four panels wereincluded in this analysis. The prob-lem for Indiana in panels 1 and 3is ignored. Insomuch as theIndiana problem involves UAArates, not nonresponse rates, theimpact of the problem on the

3. Limitations


response rates should be relativelyminor.

Method 2 involved analyzing thedata from all six panels. However,households from the ten problem-atic states and the three problem-atic ZIP Code ranges are excluded

from the analyses. The assumptionunderlying this analysis is thatdata errors are eliminated byexcluding these households. Thatassumption may not be correct;the depressed UAA rate for panel 2suggests that problems may stillexist even when the ten states and

three ZIP Code areas are eliminat-ed. These analyses do not involvea truly national sample, since somuch of the country is excludedfrom the sample. Thus, theseresults should not be generalizedto the entire nation.


As noted in Section 3, several tech-nical problems created limitationsin the way the RMIE could be ana-lyzed and interpreted. As a result,a decision was made to analyzethe data in two different ways.One of the two approaches,Method 1, restricted the usabledata to only that collected by theCATI or Internet modes. In con-trast, analyses completed usingMethod 2 allows all threeresponse modes to be compared,though not for a sample that canbe generalized to the entire nation.As the response mode is a criticalcomponent of the RMIE the resultsof this sub-national analysis arepresented in this report. The inter-ested reader can review the analy-ses completed using Method 1 inthe report entitled, Response Modeand Incentive Experiment forCensus 2000 (Westat, 2002).

Throughout this section two differ-ent response rates will be dis-cussed. These two rates are com-puted as follows:

• The first computation considersall responses, regardless of theresponse mode. This includesresponses using the paper formand any responses using theInternet or CATI. The responserates calculated this way arecalled the Overall ResponseRates (ORR).

• The second computation consid-ers only the alternative comput-er-mediated response modesthat were offered in the mailingsto the respective panels. Thus,the response rates for panels 1and 4 include only those cases

that responded via CATI.Similarly, the response rates forpanels 3 and 6 include onlythose cases that responded viathe Internet. The response ratescalculated in this manner arecalled the Assigned ModeResponse Rates (AMRR).

With either method, householdswere considered nonrespondents ifthey failed to respond at all, or ifthey provided data with too manyomissions to meet the Census2000 criteria for a completeresponse.

4.1 Effect of the incentiveon response rates – initialmailout component

4.1.1 Overall Response Rates

The ORR of the no-incentive panels(72.55 percent) and the incentivepanels (71.01 percent) were not

significantly different (chi square =2.49, df = 1, not significant (n.s.)).3

A logistic regression analysis wascarried out to reveal any significantinteractions between the incentiveand the two other factors—response mode and coverage area.The results showed that the incen-tive factor did not attain statisticalsignificance either by itself or inany interaction with the other fac-tors.

4.1.2 Assigned mode responserate

Figure 1 reveals that the incentivewas associated with a largeincrease in the AMRR.

A logistic regression analysisrevealed a significant (p < .001)

4. Major Findings

0

20

40

60

80

100

No incentive panels (Panels 1, 2, and 3)

Incentive panels (Panels 4, 5, and 6)

2.05

17.26

Figure 1.Assigned Mode Response Rate: Combined Panels

3 At the time this report was prepared,detailed response rate data for the high coverage and low coverage areas were not available.

chi square = 868.15, p <.001


interaction between the incentivefactor and the response mode fac-tor. The difference between theincentive and no-incentive condi-tions was greater for the IVR andCATI response modes than it wasfor the Internet response mode.

Chi square analyses were carriedout to illustrate the manner inwhich the incentive affected theAMRR. The results show that theAMRR in the incentive householdswere significantly (p<.001) higherthan those in the no-incentivehouseholds, regardless of whetherthe households were in the CATI,IVR, or Internet response modeconditions. The AMRR increasedfrom 1.4 to 17.9 percent for CATI;from 0.8 to 18.0 for the IVR, andfrom 4.0 to 15.9 for the Internet.Based on the logistic regression,this difference between the incen-tive and no incentive conditionwas larger for the CATI and IVRconditions than it was for theInternet condition.

The logistic regression alsorevealed a significant interactionbetween the incentive factor andthe coverage area factor. The dif-ference between the incentive andno-incentive conditions wasgreater in the high coverage areathan in the low coverage area.However, the incentive increasedthe AMRR, regardless of whetherthe households were in the high orlow coverage area (from 2.1 to19.2 percent for the HCA and from1.7 to 9.8 for the LCA).

4.1.3 Summary of results for theeffect of the incentive

The effect of the incentive in theanalyses involving all six panelsand a sub-national sample can besummarized as follows:

• The incentive offered to thehouseholds for responding via an alternative, computer-

mediated response mode had nosignificant effects on the ORR.

• The incentive increased the like-lihood that the householdswould choose the alternativeresponse mode.

• The incentive increased theAMRR most for the IVR and CATIresponse modes, and least forthe Internet response mode.

• The incentive increased theAMRR more for the high cover-age area than for the low cover-age area.

One finding regarding the choiceof incentive is of interest as well.Although the incentive increasedreporting via the alternativemodes, a large number of respon-dents never (or least not withinseven months) used the callingcard once it was activated. Of the862 cards that were activated andfor which data were available, athird had not been used. An addi-tional 38 percent had been partial-ly used, and about 28 percent hadbeen fully used.

4.2 Effect of the responsemode on response rates

4.2.1 Overall response rates

The ORR for the CATI panels(72.33 percent), IVR panels (70.67percent) and Internet panels (72.35percent) were not significantly different (chi square = 4.32, df = 2,n.s.).

The logistic regression analysisdescribed in Section 4.1.1 alsoshowed a significant interactionbetween the response mode factorand the coverage area factor.Respondents in the high coveragearea were more likely to use CATIthan the Internet. Chi squareanalyses were run to further illus-trate the relationship between theresponse mode factor and the cov-

erage area factor. The results sug-gested that the overall responserates differed among the threeresponse mode conditions in thehigh coverage area (chi square =7.05, df = 2, p < .03) but not inthe low coverage area (chi square= 2.30, df = 2, n.s.). For high cov-erage area households, the overallresponse rate was lower in the IVRcondition (73.6 percent) than ineither the CATI condition (76.2 per-cent, chi square = 6.23, df = 1, p <.02), or the Internet condition(75.4 percent, chi square = 3.21,df = 1, p < .08). No significant dif-ference was found in the high cov-erage area households between theoverall response rates in the CATIand Internet conditions (76.2 per-cent and 75.4 percent respectively,chi square = 0.64, df = 1, n.s.).

A logistic regression analysis wasrun that included the CATI no-incentive panel, the IVR no- incen-tive panel, the Internet no-incen-tive panel, and the CCG. Theoutcome variable was a responseindicator. The predictor variableswere the response mode, the cov-erage area, and all of the interac-tion terms. None of the interactionterms was statistically significant.

To further illustrate the patternacross response modes, chi squareanalyses compared the overallresponse rates of the CCG (71.1percent) with those of the CATI no-incentive (72.33 percent), IVR no-incentive (70.67 percent), andInternet no-incentive (72.35 per-cent) panels. The overall responserate of the CCG was lower thanthat of the CATI no-incentive panel(chi square = 2.89, df = 1, p < .09),and the Internet no-incentive panel(chi square = 4.29, p < .04). Theoverall response rates of the CCGand the IVR no-incentive panel didnot differ (chi square = 0.26, df =1, n.s.).


4.2.2 Effect of the response modeon the assigned mode responserate

A three by two chi square testcompared the AMRR of the CATIpanels (9.65 percent), IVR panels(9.30 percent) and Internet panels(10.0 percent). The differenceswere not significantly different (chisquare = 1.53, df = 2, n.s.).

As noted in Section 4.1.2, a logisticregression analysis revealed a sig-nificant interaction between theincentive factor and the responsemode factor. This interaction sug-gests that the incentive increasedthe AMRR in the CATI and IVR con-ditions more than in the Internetcondition. The results of a chisquare analysis suggest that in theno-incentive condition, the Internetpanel had the greatest AMRR (ver-sus the CATI panel, chi square =27.09, df = 1, p < .001; versus theIVR panel, chi square = 61.01, df =1, p < .001). The AMRR of the CATIand IVR Questionnaire panels didnot differ (chi square = 2.64, df =1, n.s.).

For the incentive condition,Internet panel had the lowestAMRR (versus the CATI panel, chisquare = 4.77, p < .03; versus theIVR Questionnaire panel, chi square= 2.98, p < .09). Again the AMRRof the CATI and IVR panels did notdiffer (chi square = 0.00, df = 1,n.s.).

4.2.3 Summary of results for theeffect of the response mode

The effect of the response mode inthe analyses involving all six pan-els and a sub-national sample canbe summarized as follows:

• The ORR did not differ acrossthe CATI, IVR, and Internet con-ditions.

• In the high coverage area, theORR in the IVR condition was

lower than that for the CATI orInternet conditions.

• The CATI no-incentive and theInternet no-incentive panels hada higher ORR than the CCG.

• The CCG’s ORR was not signifi-cantly different from that of theIVR no-incentive panel.

• In the no-incentive condition,the Internet panel had the great-est AMRR.

• In the incentive condition, theInternet panel had the lowestAMRR.

4.3 Item nonresponserates by mode of response

The highest item nonresponserates occurred when the data werecollected using the IVRQuestionnaire, up to 11.8 percentfor the race of Person 1 in thehousehold, and nearly that high forage and date of birth (10.0 percentand 10.5 percent respectively).Much lower rates occurred whenthe data were collected by theother modes. Among the othermodes, the mail had the highestitem nonresponse rate, with theInternet and CATI having the low-est rates.

The amount of missing data forthe IVR Questionnaire has impor-tant implications for the feasibilityof this mode for the decennial cen-sus. A large proportion of themissing data was due to IVRrespondents hanging up the tele-phone before the end of the inter-view. Most of these hang-upsoccurred early in the interview.Some comments from respondentsindicated impatience with the paceof the interview. This reaction mayhave been exacerbated by the typeof information that was collectedat the beginning of the interview,when the respondents were askedto enter their 22-digit identification

numbers and telephone numberswith touch-tone buttons, and tosay and spell the names of every-one in the household. These tasks,along with the speed with whichthe questions were administered,may have played a role in therespondents’ decision to terminatethe interview prematurely.

Some of the missing data in theIVR mode may be attributable toproblems respondents encounteredproviding data within the time con-straints allotted by the computerprogram. The system was pro-grammed to repeat the questionwhen it encountered two secondsof silence. Even given this repeti-tion, respondents sometimes couldnot report the information forsome items. Future IVR question-naires may need to give therespondents more time to beginanswering before it repeats thequestion. A longer wait time hasrelatively little cost (e.g., it doesnot increase the length of time tofill out the questionnaire for thosethat provide answers right away)and could result in capturing datafrom some of the respondentswho, for whatever reason, couldnot initiate their answers withintwo seconds.

4.4 Results from the Interactive VoiceRecognition (IVR)Questionnaire Satisfaction Survey

Briefly, the results from the IVRQuestionnaire Satisfaction Surveyindicate the following:

• Hispanic respondents tended tospend more time per item thanothers. Respondents fromhouseholds with more than oneHispanic member tended tohave relatively long calls andfound the questionnaire moreconfusing.


• Female respondents tended togive the system higher overallsatisfaction ratings.

• Older respondents tended togive the system higher overallsatisfaction ratings and to findthat the IVR Questionnaireafforded the appropriate amountof time to answer.

• White respondents tended not tofind the IVR Questionnaire con-fusing, and to spend less timeanswering the individual items.Black respondents tended togive the system higher overallsatisfaction ratings. However,respondents who identifiedthemselves with a race otherthan white or black tended tofind the IVR Questionnaire to beconfusing. Racial complexity ofthe household also affected howrespondents rated the IVRQuestionnaire. Respondents inmixed race households tendedto find the IVR Questionnaireconfusing and to have longercalls.

4.5 Results from thenonresponse component of the Response Mode and Incentive Experiment(RMIE)

As described earlier, the nonre-sponse component of the RMIEinvolved assigning the CCG nonre-spondents to one of six treatmentgroups parallel to the six panelsincluded in the main RMIE (referback to Appendix A). This nonre-sponse study was not conductedas a means to test the utility ofincluding nonresponse conversionincentives for the 2010 census.Rather, the goal was to test theeffect of an incentive and alterna-tive response modes as a means toimprove response from groupswho are traditionally difficult toenumerate.

With regard to the effect of thealternative modes on response, thestudy found that CATI consistentlyelicited the highest response rate(see Table 1). The IVR does notgain higher response than theInternet. There is some evidenceto suggest that these findings maybe due to difficulties in using theIVR system. Feedback from censusIVR Questionnaire testers revealedthat the system was somewhat dif-ficult to use. Moreover, the level ofresponse does not differ betweenCATI and IVR when calls and

rollovers to CATI are permittedfrom households assigned to IVR,suggesting that usability issuesrather than mode preference areresponsible for the IVR and CATIdifference.

In order to assess the effect of theincentive within and acrossresponse modes, response rates inTable 2 were computed for eachexperimental treatment along withpairwise differences between theincentive and non-incentive groupswithin and across each responsemode.

Table 1.Mode Specific Response Rates, Sample Sizes,1 andResponse Rate Differences Among Modes and AcrossIncentive Groups

Mode Mode specificresponse rate Difference**

CATI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.8%(1656)

*2.9%IVR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8%(1555)

CATI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.8%(1656)

*4.1%Internet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7%

(1717)

IVR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8%(1555)

1.2%Internet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7%

(1717)

1Undeliverables and late mail returns are excluded from this analysis.* Statistically significant when the familywise error rate is controlled using Bonferroni

at a=.1 for all comparisons.** Note that the numbers in the difference column may be slightly different from the

computations using the rates presented due to rounding error.

Table 2.Mode Specific Response Rates, Sample Sizes,1 andPairwise Differences Between Incentive and No IncentiveGroups Within and Across Response Modes

ModeMode specific response rate

DifferenceIncentive No incentive

CATI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.8% 6.7% 2.1%(875) (781)

IVR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4% 3.4% *3.0%(753) (802)

Internet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9% 3.4% .5%(867) (850)

Total . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4% 4.5% *1.9%

1Undeliverables and late mail returns are excluded from this analysis.* Indicates statistical significance when a=.1.


Results in Table 2 show that theincentive increases mode specificresponse compared to no incentivewhen rates are computed acrossresponse modes. The incentiveeffect is not significant within CATIand Internet, but is significant inthe IVR.

Table 3 presents logistic regressioncoefficients when the mode specif-ic response rate is regressed onthe experimental treatments aswell as some control variables.The Simple Model investigates theeffect of the incentive on responsewhile controlling for strata (as aproxy for socioeconomic status)under the assumption that the

effect is consistent within each

response mode. The interaction

model reveals whether the incen-

tive effect differs based on the

stratum to which it is adminis-

tered.

Tests of parameter estimates in the

Simple Model confirm that CATI

obtains higher response than the

Internet and IVR while controlling

for the incentive treatment, and

that the incentive effect holds

while simultaneously controlling

for response mode and stratum.

The Interaction Model in Table 3

helps to determine if the incentive

is more effective in increasing

response in low coverage areas(high Black and Hispanic and renterconcentration) compared to highcoverage areas. The test of thisinteraction (Incentive*HighCoverage Area = -.253) indicatesthat the effect of the incentive onresponse is not significantly differ-ent between high and low cover-age areas. This finding contradictspast literature that showed a morepronounced incentive effect amonglower socio-economic populationscompared to other populations(Kulka,1994; Singer,2002). Thereare at least two possible reasonsfor this discrepancy. First, strata,while a good indicator of censusresponse, is based on 1990 tractlevel data and may not be a suit-able proxy variable for socio-eco-nomic status. Secondly, legalityand sponsorship differencesbetween the U.S. decennial censusand surveys may explain this dis-crepancy. Certain people, such asillegal immigrants and fugitives,may deliberately avoid the census.If low coverage areas contain ahigher concentration of these peo-ple than high coverage areas, it ispossible that these results reflectthat fact that the incentive doesnot increase response from thosewho are intentionally avoiding thecensus.

Finally, logistic regression coeffi-cients in Table 4 allow an assess-ment of the effect of the incentiveon the demographics of respon-dents. Specifically, this regressionmodel includes all respondents,regardless of their experimentalpanel assignment, in an attempt todetermine which factors are associ-ated with households that per-formed the prescribed behavior toreceive the incentive.

The model suggests that Person 1 in households receiving the

Table 3.Logistic Regression Coefficients Predicting the Log Oddsof Responding to the Census Through the Assigned Mode

Predictor variables Simple modelIncentive-Strata

Interaction model

Mode:

Internet = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . *–.302 .012CATI = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . *.496 *.717IVR = 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . – –

Incentive:

Incentive = 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . *.374 *.888

Census Area (strata):

High Coverage Area = 1 . . . . . . . . . . . . . . . . . . . . . . . . *.567 *.725

Interactions:CATI * Incentive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . –.365Internet* Incentive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . *–.534Incentive *Strata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . –.253

Intercept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . –3.616 –3.934

* Indicates statistical significance when a = .1


incentive due to alternativeresponse mode participation tendsto be younger than Person 1 inhouseholds not receiving theincentive. This finding may sug-gest that the incentive is moreattractive to younger persons.Conversely, since the incentive wasonly activated for those who trieda new response mode, perhapsyounger people are more likely touse new technology. It is impossi-ble to control for the effects ofmode in this study given that analternative mode response was

required in order for a household

to receive the incentive. However,

an age comparison of mail and

electronic mode respondents

reveals that mail respondents are

on average older (50.4) than elec-

tronic mode respondents (42.1),

suggesting that the proposed

incentive effect on younger people

may be due to more willingness to

try a new mode. Otherwise, while

controlling for age, sex, and

household size there is no evi-

dence to suggest that incentives

disproportionately recruit non-whites or renters.

In Table 2, the increase in modespecific response due to the incen-tive is significant when the threeresponse modes are combined, yetthe effect of the incentive isinsignificant when overall responseto the second mailing is consid-ered (see Table 5). This findingsuggests that the incentive redi-rects response to alternativemodes, but does not encourageresponse from those with no inten-tion of responding.

4.6 Results from theInternet Usage Survey

Of the respondents contacted forthis study, 8.2 percent (6.8 percentin HCA, 8.6 percent in LCA) did notunderstand or have any knowledgeof the concept of the Internet.Interviews with these respondentswere terminated as soon as thislack of understanding was revealedsince the remaining survey ques-tions probe for reasons theInternet was not used.

Somewhat surprisingly, 62.9 per-cent of respondents had Internetaccess at one or more locationseven though they responded to thecensus by mail or phone whengiven the option of providing cen-sus data via the Internet (seeFigure 2). After this informationwas gathered, interviews withrespondents who did not haveInternet access were terminated.

Table 6 provides the percentage ofrespondents in each stratum andthe full sample who had Internetaccess at various locations.

A large number of respondentswere unaware that the option ofreplying to the census by theInternet was available. Nearly half(48.2 percent) of respondents whoreceived the calling card as an

50

54

58

62

66

70

74

78

82

HCA LCA Population

Overall

70.7

60.1

62.9

Figure 2.Internet Access Rates Among Mail Respondents byCoverage Area (Percent)


incentive to use the Internet wereunaware of the Internet option,despite the colorful brochure print-ed on heavy stock paper includedin their questionnaire package con-taining the calling card as well asan announcement of Internet avail-ability. Over half (54.9 percent) ofnon-incentive respondents report-ed that they did not know theycould have used the Internet torespond.

Table 7 provides data on the rea-sons respondents with Internetaccess gave for completing thepaper census form rather than theInternet version.

Of those who received the incen-tive in the initial mailing, 57.3 per-cent claimed that they wereunaware of the offer to receive afree calling card. When those whowere unaware of the incentiveoffer or did not receive the offerwere asked if they would use theInternet if they were given a 30-minute calling card to do so, 41.2percent indicated that they would.Those who continued to declinethe Internet option were asked ifthey would use the Internet if thevalue of the calling card was dou-bled or tripled. Table 8 summa-rizes the findings from these questions.


5.1 Recommendationsbased on the ResponseMode and IncentiveExperiment

The results of the RMIE can helpguide future use of computer-medi-ated response modes and incen-tives in the decennial census. Theresults address the questions:

• Can offering alternate, comput-er-mediated response modesincrease overall response rates?

• Do respondents using alternate,computer-mediated responsemodes tend to provide goodquality data?

• Are the costs involved in offer-ing alternative response modescommensurate with any advan-tages they offer?

Overall response rates didincrease when respondentswere offered the CATI andInternet alternative modes, ascompared with the controlgroup. The increase in overallresponse rates was small andoccurred only when the respon-dents were not offered an incen-tive. When an incentive wasoffered, overall response rateswent down slightly, to about thesame level as that of the controlgroup. These alternative responsemodes also seemed to reduce theamount of missing data for partic-ular items; that is, the item nonre-sponse rates tended to be higherfor mail questionnaires as com-pared with CATI and Internet ques-tionnaires.

The major drawback to theCATI mode is its cost. CATI

involves a number of expensesthat the other modes do notrequire, such as the costs associat-ed with the interviewers, CATIequipment and software, and the800 telephone line. The interview-er costs are increased by the timethat they must spend unoccupied,waiting for calls. However, CATIalso involves some cost savingswithin the context of a large-scalecensus data collection effort. CATIdata collection saves the costs forreturn postage and data captureassociated with mail surveys. Also,CATI did seem to improve someaspects of data quality; CATI didhave less missing data than themail survey on certain items.However, this difference was notextremely large and probably doesnot justify the increase in coststhat this mode would likelyinvolve.

It is difficult to assess these trade-offs precisely. However, it is likelythat CATI poses a significantincrease in cost relative to the cur-rent census procedures, unlessthese costs can be offset by a largeincrease in the response rate. TheRMIE results suggest that offeringa CATI response mode alternativedoes not bring about such a largeincrease in the response rate.

Like CATI, the Internet modeyielded relatively high dataquality. There was also a rela-tively low rate of missing dataon key items. When an incentiveand insert were not included, theresponse rate was approximatelyone to two percentage points high-er than that of the CCG. Relative tothe census mail procedure, the

costs of fielding a web survey arelikely to be relatively modest. Theprimary additional cost associatedwith the Internet, relative to mail,involves the development andmaintenance of the software andhardware. However, this cost isfixed and does not increase asmore data are collected. Web sur-veys also have lower postage andprocessing costs than mail surveysdo. Data quality could be improvedfurther with the introduction ofautomated edits.

Based on conservative assumptionsand the data from RMIE, one mightsave between one and six milliondollars in postage costs alone ifbetween three percent and 15 per-cent of the sample uses the webrather than the mail survey. Thisestimate assumes that the postageto mail back the short form is 37cents and 110 million householdsmust be enumerated (3 percent x110 million households x 37 centspostage = $1.2 million; 15 percentx 110 million households x 37cents = $6.1 million). This savingswould more than offset the costsrequired to design, develop andmaintain the web survey. Ofcourse, the web survey would alsoproduce savings related to reducedprocessing (receipt and scanning).Given this crude calculation, it isanticipated that the Internet wouldbe cost-effective even if a relativelysmall proportion of respondentsused it. Offering a web surveywould also provide additional costsavings if it increased the overallresponse rate, as it did in RMIE, asfewer followup field interviewswould be required.

5. Recommendations


The implications of this experi-ment for the use of the IVRQuestionnaire are complex. Dataquality was the lowest for thisresponse mode, both in terms ofresponse rate and missing dataitems. Much of these missing datawere due to individuals hanging uprelatively early during the inter-view. With respect to costs, the IVRsystem has fixed costs related topurchasing the hardware, develop-ing the software and maintainingthe data collection site. There areother costs if operator assistanceis provided for those individualswho cannot complete the question-naire using the IVR system. Thereare also additional data-processingcosts because of the need to tran-scribe information that the speechrecognizer could not code.Therefore, an IVR Questionnaire ismore costly than an Internet sur-vey. It is unclear how IVR costscompare to those of CATI or mailquestionnaires. An additionalissue is whether (and how) toinform respondents that theywould be providing their data to acomputer. The RMIE mailings didnot notify IVR households that thetelephone number was for an IVRQuestionnaire. Some of the nega-tive reaction to the IVRQuestionnaire may have beenavoided if respondents made thecall with the expectation that theywould be interacting with an auto-mated system.

Another concern revolves aroundthe design of the IVR interview.Several tasks were difficult to com-plete or took more time thandesired with the IVR Questionnaire.This likely affected the quality ofthe data with this mode. Issuesthat may have led to problemsinclude: (1) entering a 22-digit ID,(2) reporting and spelling out thenames of all persons in the house-hold and (3) reporting race using

information printed on the paperquestionnaire.

Some of these issues were a func-tion of the special nature of thisexperiment within Census 2000.For example, shortening the IDmay be possible if a crosswalkcould be developed between thefull 22-digit census number and ashorter number that would be easi-er to enter. Also, the IVRQuestionnaire may become easierto use as the technology of speechrecognition becomes more sophis-ticated. For example, the IVRQuestionnaire did not rely on rec-ognizing the responses to everyquestion. The responses to thequestions on race and certainother topics were recorded andlater transcribed. Improved capa-bilities to recognize speech, espe-cially words embedded within asentence (e.g., reports of multipleraces), would allow for easier inter-action between the respondent andthe computer.

The RMIE results show that theinclusion of a calling card withan insert was extremely effec-tive in promoting the use ofthe alternative response mode.Comparisons between the incen-tive and no-incentive conditionsreveal that the incentive was asso-ciated with three to four-foldincreases in the rate of using thealternative mode.

At least some portion of this effect is probably attributable to the insert, which drew therespondents’ attention to the avail-ability of the alternative mode. Thenon-incentive condition relied sole-ly on the census cover letter toinform respondents about theavailability of the computer-mediat-ed mode. Many respondents in theno-incentive panels probably didnot read the letter. The insert, bycontrast, prominently called the

respondents’ attention to the com-puter-mediated alternative mode.The insert and calling card mayaccount for some of the effectsobserved in the incentive condi-tion.

However, this increase seemed tocome at some cost to the overallresponse rate with one to two per-cent fewer people respondingwhen an incentive was offered. Inboth the CATI and Internet condi-tions, the overall response rates,once factoring in the mail respons-es, were lower in the incentivepanels than in the no-incentivepanels. This reduction may be dueto the fact that the calling cardincentive makes the response taskmore complicated. If the alterna-tive modes are not available at thetime the respondent tries to usethem, the respondent may not fol-low up in all cases to complete thequestionnaire at a later time. Theone advantage of a mail question-naire is that it can be filled out themoment the package arrives.Completing a CATI questionnairerequires the use of a telephoneand the availability of a CATI oper-ator. A web survey requires accessto a computer that has Internetaccess. If these are not available atthe time the respondent attemptsto fill out the questionnaire, thensome persons may simply neverrespond.

This result may also be indicativeof a relatively weak effect of thecalling card as an incentive. In fact,many respondents whose callingcards were activated never usedthem, suggesting that the callingcard may not have been a univer-sally powerful incentive.

With respect to the nonre-sponse component of the RMIE,an examination of the responsemode alternatives reveals thatCATI obtains the highest level


of response compared to IVRand the Internet. However, itshould not be inferred that thepeople prefer CATI over theInternet for data collection.Internet accessibility limitationsamong the population in this non-response component confound theresponse rate comparisons amongthe modes. As Internet accesscontinues to span the United Statespopulation, experiments testingthe feasibility of this method forcensus data collection should con-tinue to be tested.

Consistent with past findings,the use of an incentive in thisnonresponse componentincreases response to thealternative modes; however,the effects disappear whentotal response to the secondmailing is examined. Therefore,the incentive in this experiment issuccessful in transferring responsethat would have otherwise beenobtained by mail to a differentmode, but not in recruiting house-holds who would otherwise notrespond.

In contrast to past incentive lit-erature, there is no evidence ofincreased incentive effectswithin areas of low census cov-erage (with high proportionsof non-whites and renter units)compared to high coverageareas, which may be due to thefact that coverage area is not agood proxy for socio-economicstatus. Moreover, there is no evi-dence that incentives are morepowerful at increasing response inthe absence of an interviewer as amotivator. It is possible that IVRdifficulties as well as Internetaccessibility issues confound theincentive effect within each mode.Moreover, the interviewer was onlya motivating source in keeping therespondent from discontinuing theinterview, since the initial contact

was respondent-initiated. Perhapsincentives would prove to be mosteffective in the self-administeredmodes if the cases assigned to theCATI mode were contacted directlyby the interviewer as in a tradition-al survey setting.

Comparisons of respondentdemographics reveal that theincentive seems to attractyounger respondents; however,this finding is confounded withthe influence of the alternativeresponse mode options. Thereis some evidence to suggest thatyounger persons may be influ-enced by the chance to use a newmode.

5.2 Recommendations forfuture research

Given the success of the insert andincentive to promote the use of analternative mode to respond to thecensus, this option should be con-sidered in future research. Thisresearch should carefully considerboth the role the insert and incen-tive separately play in the respon-dent’s decision to participate. Itwould be useful to better under-stand the relative effects of thecalling card incentive and theinsert on the respondents’ decisionto use the alternative responsemodes. The use of just an insert,without any incentive, has a num-ber of economical and logisticaladvantages for the census.Research is needed into the bestways to present the alternativesthrough either the letter or aninsert.

As many respondents inter-viewed as part of the IUSreported that they wereunaware that either an alterna-tive response mode or anincentive was offered, futureresearch should also be direct-ed at how best to convey thisinformation to respondents. It

is possible that when the packageof materials arrives at the house-hold, one person opens the pack-age, saves what appears to be nec-essary (the actual form and thereturn envelope) and throws therest away. Then, when a memberof the household is actually readyto complete the census formhe/she no longer has the informa-tion explaining these aspects ofthe data collection process.Perhaps finding a way to providethis information directly on thepaper form would further increasethe percent of respondents whoprovide their data through somealternative response mode.

There is also some indication thatthe calling card incentive may nothave been a particularly effectivemotivator. Only 28 percent ofrespondents fully used the callingcard and a third of respondentsnever used their cards at all.

While the calling card has theadvantage of being usable any-where in the country (whichstore gift certificates, forexample, would not be), futureresearch should investigateother types of incentives thatmight be valued by a greaterpercentage of respondents.The incentive and alternativeresponse modes were not effectivetools for increasing responseamong typical census nonrespon-dents as evidenced by the resultsof the nonresponse component ofthe RMIE. The incentive, whilesomewhat effective in directingresponse to a particular mode, hasno overall effect on total responseto the census. Moreover, theresponse mode comparisons inthis study are confounded due toInternet access limitations as wellas IVR system technology limita-tions. Therefore, further testing isneeded prior to the 2010 census.Obviously we are likely to see

increased access to the Internet inthe years to come. With increasedaccess may come an increasedacceptance of the use of theInternet for collecting importantinformation such as that collectedin the census. Similarly, it is likelythat enhancements will continue tobe made in the speech recognitionsoftware used in the IVRQuestionnaire. Future researchshould continue to monitor theprogress of this software. A more

“user-friendly” system mightincrease response rates for thismode as well as reduce theamount of missing data thatoccurred in this mode.

Finally, future research should seekto gain a more detailed under-standing of the costs associatedwith providing each of the alterna-tive response modes. This knowl-edge would further inform thedecision to provide these alterna-

tive modes in the future. In addi-tion, such information would allowresearchers to understand the true“cost” of providing an incentive inthe census. If the costs associatedwith mailouts and data processingcould be sufficiently reduced byoffering an incentive for respon-dents to provide their data throughan alternative response mode, thenan incentive might pay for itself.



References

Kulka, R. A. (1994) “The Use ofIncentives to Survey Hard-to-ReachRespondents: A Brief Review ofEmpirical Research and CurrentPractice,” Paper from seminar,“New Directions in StatisticalMethodology,” COPAFS, Bethesda,MD, May 25 - 26.

Malakhoff, L. and D. Sanders(2000) Program Master Plan forthe Response Mode and IncentiveExperiment. Report prepared forthe Research and ExperimentationProgram Steering Committee of theU.S. Census Bureau.

Singer, E. (2002) “The Use ofIncentives to Reduce Nonresponse

in Household Surveys,” in SurveyNonresponse, Robert Groves, DonDillman, John Eltinge, and RoderickLittle (editors), pp. 163 - 178.

Westat (2002) Response Mode andIncentive Experiment for Census2000. Final Report prepared forthe U.S. Bureau of the Census.

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 A-1

Appendix A

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 B-1

Appendix B

Inserts Included With RMIE Mailings

B-2 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 B-3

B-4 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

Final ASQ 2000 ShortForm Script

Test line: 1-877-286-3119

Revised March 7, 2000

Notes:

Panel 2 = ASQ Control Panel

Panel 5 = ASQ with calling card incentive

Panel 8A= ASQ with calling card, NRFU

Panel 8B = ASQ, no calling card, NRFU

All responses must be recorded for playback and veri-fication and transcribed if necessary.

Feedback to the respondent is done by the recordedaudio clips.

Some responses do not have to be recognized in realtime. Spelled and spoken names will be processed bySpeechWorks in post-processing and ship the resultsback to Westat. These entries are noted by:

<record, transcribe, recognize (SpeechWorks)>

Other responses may be processed in batch mode atthe end of the project to obtain information about therecognition confidence level needed for the ASQusability analysis. These entries are noted by:

<record, transcribe, batch recognition later>

Open-ended responses from the satisfaction survey aremarked with this entry:

<record, transcribe>

###########################

[chime] You have reached the Census Bureau’sComputerized Questionnaire.

[testing] The data you provide is not confidential andwill be used for software development.

(March 13, 2000 and later) Your answers are protectedby law.

(April 1, 2000 and earlier)

You will be asked to provide information about your-self and persons living in your household on April 1,2000, including:

(April 2, 2000 and later)

You will be asked to provide information about your-self and persons who were living in your household onApril 1, 2000, including:

* last name, first name and middle initial;

* sex

* date of birth

* age on April 1, 2000

* origin

* race

* relationship

[chime] We will record your information. When youhear this beep <tone> please speak and keep youranswers brief. Please keep the form with your ques-tionnaire ID at hand to assist you with some of thequestions. We will now begin.

Do you have a telephone with number buttons, yes orno? <tone>

if silence, then

Do you have a telephone with number buttons?Please say yes or no AFTER the beep. <tone>

if silence, thentransfer to CATI

Your questionnaire ID number is located above youraddress on the form mailed to you.

<If yes, then>

<all touchtone digits scenario>

buttons = true

ID22:

Please enter all 22 digits of your questionnaire id withthe pushbutton keys on your telephone after you hearthe beep.<tone2>

<accept input>

You entered $$$$$ dash $$$$$$$ dash $$ dash $$$dash $$$ dash $$, Is this correct, yes or no? <tone>

If yes, then

go to VERIFY 1

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 C-1

Appendix C

If no, then

go to AGAIN

if silence, then

You entered $$$$$ dash $$$$$$$ dash $$ dash $$$ dash $$$ dash $$.

Is this correct? Please say yes or no after the beep. <tone>

If yes, then

go to VERIFY 1

If no OR silence, then

go to AGAIN

<end all touchtone digits scenario>

<begin punctuated touchtone digits scenario>

ID5:

Please enter the first five digits of your QuestionnaireID with the pushbutton keys on your telephone afteryou hear the beep.<tone2>

<accept input>

You entered xxxxx. Is this correct, yes or no? <tone>

if silence, then

You entered xxxxx. Is this correct? Please sayyes or no after the beep <tone>

If no, then

go to ID5

ID7:

Please enter the next seven digits of your question-naire ID with the pushbutton keys on your telephoneafter you hear the beep. <tone2>

<accept input>

You entered xxxxxxx. Is this correct, yes or no?<tone>

if silence, then

You entered xxxxxxx. Is this correct? Please sayyes or no after the beep <tone>

If no then go to ID7

ID2:

Please enter the next 2 digits of your questionnaire IDwith the pushbutton keys on your telephone after youhear the beep. <tone2>

<accept input>

You entered xx. Is this correct, yes or no? <tone>

if silence, then

You entered xx. Is this correct? Please say yes orno after the beep <tone>


ID3: Please enter the next 3 digits of your question-naire ID with the pushbutton keys on your telephoneafter you hear the beep. <tone2>

<accept input>

You entered xxx. Is this correct, yes or no? <tone>

if silence, then

You entered xxx. Is this correct? Please say yesor no after the beep <tone>


ID32:

Please enter the next 3 digits of your questionnaire IDwith the pushbutton keys on your telephone after youhear the beep. <tone2>

<accept input>

You entered xxx. Is this correct, yes or no? <tone>

if silence, then



ID222:

Please enter the last two digits of your questionnaireID with the pushbutton keys on your telephone afteryou hear the beep. <tone2>

<accept input>

You entered xx. Is this correct, yes or no? <tone>

if silence, then


If no, then go to ID222

C-2 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

go to VERIFY 2

<end punctuated touchtone digits scenario>

<If no, then>

<all spoken digits scenario>

IDV22:

Please say all 22 digits of your questionnaire ID with-out pausing after you hear the beep.<tone>

<accept input>

You said $$$$$ dash $$$$$$$ dash $$ dash $$$ dash$$$ dash $$, Is this correct, yes or no? <tone>

If yes, then

go to VERIFY 1

If no, then

go to AGAIN

if silence, then

You entered $$$$$ dash $$$$$$$ dash $$ dash$$$ dash $$$ dash $$.

Is this correct? Please say yes or no after thebeep. <tone>

If yes, then

go to VERIFY 1

If no OR silence, then

go to AGAIN

<end all spoken digits scenario>

<punctuated spoken digits scenario>

IDV5:

Please say the first five digits of your Questionnaire IDafter you hear the beep. <tone>

<accept input>

You said xxxxx. Is this correct, yes or no? <tone>

if silence, then

You entered xxxxx. Is this correct? Please sayyes or no after the beep <tone>

If no, then

go to IDV5

IDV7:

Please say the next seven digits of your questionnaireID after you hear the beep. <tone>

<accept input>

You said xxxxxxx. Is this correct, yes or no? <tone>

if silence, then

You entered xxxxxxx. Is this correct? Please sayyes or no after the beep <tone>

If no then go to IDV7

IDV2: Please say the next 2 digits of your question-naire ID after you hear the beep. <tone>

<accept input>

You said xx. Is this correct, yes or no? <tone>

if silence, then



IDV3:

Please say the next 3 digits of your questionnaire IDafter you hear the beep. <tone>

<accept input>

You said xxx. Is this correct, yes or no? <tone>

if silence, then



IDV32:

Please say the next 3 digits of your questionnaire IDafter you hear the beep. <tone>

<accept input>

You said xxx. Is this correct, yes or no? <tone>

if silence, then



IDV222: Please say the last two digits of your ques-tionnaire ID after you hear the beep. <tone>

<accept input>


You said xx. Is this correct, yes or no? <tone>

if silence, then



go to VERIFY 2

<end punctuated spoken digits scenario>

:VERIFY 1

<verify ID>

<If no match, then>

The number you entered is not in our records.

:AGAIN

if buttons, then

go to ID5

else

go to IDV5

:VERIFY 2

<verify ID>

<if ID used before, then

if complete, then

We see from our records that you already providedyour Census information. We are transferring you toan operator who will answer your questions.

else

We see from our records that you entered some infor-mation into this system. We are transferring you nowto an operator who will take your information.

<transfer to OA>

>

<If no match, then>

See bailout specification

#####HOME_OWNER

[chime] We will now ask you about this property.

Is this property owned by you or someone in thishousehold free and clear, without mortgage, yes orno? <tone>

if silence, then

Is this property owned by you or someone in thishousehold free and clear without a mortgage? Pleasesay yes or no after the beep <tone>

<if no, then>

Is this property owned by you or someone in thishousehold with a mortgage or loan, yes or no? <tone>

if silence, then

Is this property owned by you or someone in thishousehold with a mortgage or loan? Please say yes orno after the beep. <tone>

<if no, then>

Is this property rented for cash, yes or no?<tone>

if silence, then

Is this property rented for cash? Please say yes or noafter the beep. <tone>

<if no, then>

Is this property occupied without payment of cashrent, yes or no? <tone>

if silence, then

Is this property occupied without payment of cashrent? Please say yes or no after the beep. <tone>

#######NAME & TELEPHONE

We will need your name and telephone number in casewe need to contact you to understand or clarify ananswer. Please say your first name after thebeep.<tone>


if silence, then

Please say your first name AFTER you hear the beep.<tone>


Please say your last name after the beep. <tone>


if silence, then

Please say your last name AFTER you hear the beep.<tone>


Ok ... now


short = false

silence1 = false

Phone:

If buttons, then

if not short, then

Please enter your phone number, area code first, withthe number buttons on your telephone now. <tone2>

else

if short or silence1, then

Please enter all ten digits of your phone numberAFTER you hear the beep.<tone2>

else

if not short, then

Please say your phone number, area code first, byspeaking one digit at a time now. <tone>

else

if short or silence1, then

Please say all ten digits in your phone number withoutpausing AFTER you hear the beep.<tone>

We have xxx <pause> xxx <pause> xxxx as your tele-phone number. Is this correct, yes or no? <tone>

telephone = false

if silence, then

silence1 = true

go to Phone

If no, then

If count(digits) < 10, then short = true

go to Phone

If yes, then telephone = true

##### NAME_LIST


Next, you will be asked to list any other persons livingat this address on April 1, 2000.


Next, you will be asked to list any other persons wholived at this address on April 1, 2000.

Certain persons will be counted at other places, so DONOT INCLUDE anyone who is:

away at college,

OR in a correctional facility, nursing home, or men-tal hospital on April 1, 2000,

OR in the Armed Forces and living somewhere else,

OR staying at another place most of the time.

In addition to yourself, are there any other householdmembers that need to be counted, yes or no? <tone>

if silence, then

In addition to yourself, are there any other householdmembers that need to be counted? Please say yes orno after the beep. <tone>

If yes, then

roster = 2

Please say the first name of person 2 after the beep.<tone>


if silence, then

Please say the first name of person 2 AFTER youhear the beep. <tone>


Please say the last name of person 2 after the beep.<tone>


if silence, then

Please say the last name of person 2 AFTER youhear the beep. <tone>


Are there any other household members that need tobe counted, yes or no? <tone>

if silence, then

Are there any other household members that need tobe counted? Please say yes or no after the beep.<tone>

...(repeat for all members of household)

#####

PERSON 1


#####

[chime]

if roster > 1, then

For each of the persons on your list, we will now askyou a series of questions starting with yourself.

else

We will now ask you a series of questions aboutyourself.

#####NAME

:FN

Please spell your first name after the beep. <tone>


If silence, then

Please spell your first name AFTER you hear thebeep. <tone>


:LN

Please spell your last name after the beep.<tone>


If silence, then

Please spell your last name AFTER you hear thebeep.<tone>


:MI

Please say your middle initial. If there is no middle ini-tial, say “none.” Answer after the beep. <tone>


If silence,

Please tell us your middle initial. If there is no middleinitial say “none”. Answer AFTER you hear the beep.<tone>


#####SEX

What is your sex, female or male? <tone>


if silence, then

What is your sex? Please answer either female or maleAFTER you hear the beep. <tone>


#####AGE & DATE_OF_BIRTH

[chime] We will now ask about your age and date ofbirth.

#####AGE


What is your age on April 1, 2000? Please answer afterthe beep. <tone>


if silence, then

What is your age on April 1, 2000? Please answerAFTER you hear the beep. <tone>



What was your age on April 1, 2000? Please answerafter the beep. <tone>


if silence, then

What was your age on April 1, 2000? Please answerAFTER you hear the beep. <tone>


Please tell us the month, day, and year of your birthafter the beep. <tone>


if silence, then

Please tell us the month, day, and year of your birth.Please answer AFTER you hear the beep. <tone>


#####ORIGIN

[chime] We will now ask about your origin.

Are you of Spanish or Hispanic origin, yes or no?<tone>

if silence, then

Are you of Spanish or Hispanic origin? Please say yesor no after the beep. <tone>

<if yes, then>


Which of the following best describes your origin:

Mexican, Mexican American, Chicano, Puerto Rican,Cuban or other? <tone>

if silence, then

Which of the following best describes your ori-gin:

Mexican, Mexican American, Chicano, Puerto Rican,Cuban or other? Please answer AFTER you hear thebeep. <tone>

<if unrecognized, then go to OH>

<if other, then>

Okay, to what other Spanish or Hispanic group doyou belong? <tone>


if silence, then

Please say what other Spanish or Hispanic group youconsider yourself a member AFTER the beep. <tone>


<Argentinean, Colombian, Dominican, Nicaraguan,Salvadoran, Spaniard, other Hispanic>

:OH

Please spell that after the beep.<tone>


if silence, then

Please spell the name of the Spanish or Hispanicgroup AFTER the beep. <tone>


Do you belong to any other Spanish or Hispanicgroups, yes or no? <tone>

if silence, then

Do you belong to any other Spanish or Hispanicgroups? Please say yes or no after the beep. <tone>

<if yes, then>

Please tell us the name or names of these groupsafter the beep.<tone>


<Mexican, Mexican American, Chicano, Puerto Rican,Cuban, Argentinean, Colombian, Dominican,Nicaraguan, Salvadoran, Spaniard, other Hispanic>

if silence, then

Please tell us the name or names of any other Spanishor Hispanic group you consider yourself a memberAFTER the beep. <tone>


#####RACE

[chime] We will now ask about your race.

<If panel 2 or 5, then>

Do you belong to one or more of the races printedunder question 8 on page 1 of the questionnaire, yesor no? <tone>

<If panel 8A or 8B, then>

Do you belong to one or more of the races printedunder item 9 inside the brochure, yes or no? <tone>

if silence, then

goto RACELIST 1

<If yes, then>

Please say the name of the race or races you belong towith a short pause between each name after thebeep.<tone>


if silence, then

go to RACELIST1

else

go to CONFIRM 1

<If no, then>

Please say the name of the other race or races youbelong to with a short pause between each name afterthe beep. <tone>


if silence, then

go to RACELIST1

else

go to CONFIRM 1

:RACELIST1

Are you White? Please say yes or no after the beep.<tone>

<if yes, then>


Do you belong to any other races, yes or no? <tone>

if silence, then

Do you belong to any other races? Please sayyes or no after the beep. <tone>

<if no, then>

skip to CONFIRMATION 1

Are you Black, African American, or Negro, yes or no?<tone>

if silence, then

Are you Black, African American, or Negro? Pleasesay yes or no after the beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you an American Indian or Alaska Native, yes orno? <tone>

if silence, then

Are you an American Indian or Alaska Native? Pleasesay yes or no after the beep. <tone>

<if yes, then>

Please say the name of your tribe after the beep.<tone>


<Cherokee, Blackfoot, Navajo, Chickasaw, Chippewa,Potawatomi, Sioux, Tohono O’Odham, Choctaw, Pima,Pueblo, Tlingit, Apache, Seminole, Iroquois, AlaskanAthabaskans, Lumbee, Cheyenne, Creek, Comanche,other tribe>

if silence, then

Please say the name of your tribe AFTER youhear the beep. <tone>


Please spell the name of your tribe after thebeep.<tone>


if silence, then

Please spell the name of your tribe AFTER youhear the beep. <tone>



if silence, then


<if no, then>


Are you an Asian Indian, yes or no? <tone>

if silence, then

Are you an Asian Indian? Please say yes or no afterthe beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you Chinese, yes or no? <tone>

if silence, then

Are you Chinese? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you Filipino, yes or no? <tone>

if silence, then

Are you Filipino? Please say yes or no after the beep.<tone>

<if yes, then>



if silence, then


<if no, then>


Are you Japanese, yes or no? <tone>

if silence, then

Are you Japanese? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you Korean, yes or no? <tone>

if silence, then

Are you Korean? Please say yes or no after the beep.<tone>

<if yes, then>


if silence, then


<if no, then>


Are you Vietnamese, yes or no? <tone>

if silence, then

Are you Vietnamese? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you from some other Asian race, yes or no?<tone>

if silence, then

Are you from some other Asian race? Please say yesor no after the beep. <tone>

<if yes, then>

Please say the name of your race after the beep.<tone>


if silence, then

Please say the name of your race AFTER youhear the beep. <tone>


<Cambodian, Hmong, Indonesian, Pakistani, Laotian,Thai, other Asian race>

Please spell the name or your race after thebeep.<tone>


if silence, then

Please spell the name of your race AFTER youhear the beep. <tone>



if silence, then


<if no, then>


Are you Native Hawaiian, yes or no? <tone>

if silence, then

Are you Native Hawaiian? Please say yes or no afterthe beep. <tone>

<if yes, then>


if silence, then



<if no, then>


Are you Guamanian or Chamorro, yes or no? <tone>

if silence, then

Are you Guamanian or Chamorro? Please say yes orno after the beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you Samoan, yes or no? <tone>

if silence, then

Are you Samoan? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Are you from some other Pacific Islander race, yes orno? <tone>

if silence, then

Are you from some other Pacific Islander race? Pleasesay yes or no after the beep. <tone>

<if yes, then>



if silence, then



<Fijian, Palauan, Tahitian, Tongan, other PacificIslander>

Please spell the name of your race after thebeep.<tone>


if silence, then



Do you belong to some other race, yes or no?<tone>

if silence, then

Do you belong to some other race? Please sayyes or no after the beep. <tone>

<if no, then>


<if yes, then>



if silence, then



Please spell the name of your race after thebeep.<tone>


if silence, then



####CONFIRMATION 1


[chime] We will now summarize the information yourecorded about yourself.

{Silent responses should be spoken as “blank”}

Name: <first name 1 & last name 1>.

Sex: <sex>

Birthdate: if <month> = silence AND <day> =silence AND <year> = silence, then

say “blank”

else

<month> <day> <year>

Age: <age>

Origin: <origin> (if blank, say “Non-Hispanic”)

Race: <race>

Ownership: <owned free and clear, owned with a mort-gage, rented for cash, occupied with no rent> {NOTE:these phrases are spoken, not synthesized}

Is all of this information correct, yes or no? <tone>

if silence, then

Is all of this information correct? Please say yes orno after the beep. <tone>

<if no, then>

Please tell us which item or items were incorrect andprovide the correct information for each one after thebeep. <tone>


if silence, then

For each item you wish to correct, please tell us theitem and the new information AFTER you hear thebeep. <tone>


#####

PERSON 2..n

#####

[chime] We will now ask you some questions about<first name n> <last name n>.

#####NAME

Please spell the first name of this person after thebeep. <tone>


if silence, then

Please spell the first name of this person AFTER youhear the beep. <tone>


Please spell the last name of this person after thebeep. <tone>


if silence, then

Please spell the last name of this person AFTER youhear the beep. <tone>


Please say their middle initial. If there is no middle ini-tial, say “none”. Answer after the beep. <tone>


if silence, then

Please say their middle initial. If there is no middle ini-tial, say “none”. Answer AFTER you hear the beep.<tone>


#####RELATIVE

Is <first name n> <last name n> related to you, yes orno? <tone>

if silence, then

Is <first name n> <last name n> related to you? Pleasesay yes or no after the beep. <tone>

if yes, then

relation = true

if panel 2 or 5, then

Which one of the items listed under question 2 onpage 2 of the form describes the relationship betweenthis person and yourself? <tone>

if panel 8A or 8B, then

Which one of the items listed under item 5 inside thebrochure describes the relationship between this per-son and yourself? <tone>

if silence, then

Which describes this person’s relationship toyou,


husband or wife,

natural born son or daughter,

adopted son or daughter,

stepson or stepdaughter,

brother or sister,

father or mother,

grandchild, parent-in-law,

son or daughter-in-law, or other relative? Pleaseanswer AFTER you hear the beep. <tone>

if “daughter”, then

:D1 Is this person your natural born daughter, yesor no? <tone>

if silence, then

Is this person your natural born daughter? Please sayyes or no after the beep. <tone>

if yes, then go to SEX

if no, then

:D2 Is this person your adopted daughter,yes or no? <tone>

if silence, then

Is this person your adopted daughter? Please say yesor no after the beep. <tone>


if no, then

:D3 Is this person your stepdaugh-ter, yes or no? <tone>

if silence, then

Is this person your stepdaughter? Please say yes or noafter the beep. <tone>


if “son”, then

:S1 Is this person your natural born son, yes orno? <tone>

if silence, then

Is this person your natural born son? Please say yes orno after the beep. <tone>


if no, then

:S2 Is this person your adopted son, yesor no? <tone>

if silence, then

Is this person your adopted son? Please say yes or noafter the beep. <tone>


if no, then

:S3 Is this person your stepson,yes or no? <tone>

if silence, then

Is this person your stepson? Please say yes or no afterthe beep. <tone>

If yes, then goto SEX

if unrecognized, then goto OREL

if other OR (if D1, D2, D3 are no OR silent) OR (if S1,S2, S3 are no OR silent), then

Please say what other relationship this person has withyou after the beep. <tone>


if silence, then

Please say what other relationship this person has withyou AFTER you hear the beep. <tone>


:OREL

Please spell the name of the relationship this personhas with you after the beep. <tone>


if silence, then

Please spell the name of the relationship this personhas with you AFTER you hear the beep. <tone>


if no, then

if panel 2 or 5, then

Which one of the items listed under question 2 onpage 2 of the form describes the association betweenthis person and yourself? <tone>


if panel 8A or 8B, then

Which one of the items listed under item 5 inside thebrochure describes the association between this per-son and yourself? <tone>

if silence, then

Please say which of the following best describes theassociation between yourself and this person: aroomer, boarder, foster child, housemate, roommate,unmarried partner, other? Answer AFTER you hear thebeep. <tone>

if unrecognized, then go to OASS

if other, then

Please say what other association this person has withyou after the beep. <tone>


if silence, then

Please say what other association this person has withyou after the beep. Answer AFTER you hear the beep.<tone>

<record, transcribe, batch recognitionlater>

:OASS

Please spell the name of the association this personhas with you after the beep. <tone>


if silence, then

Please spell the name of the association this personhas with you. Please answer AFTER you hear the beep.<tone>

<record, transcribe, batch recognitionlater>

#####SEX

What is this person’s sex, female or male? <tone>


if silence, then

What is this person’s sex? Please answer female ormale after the beep. <tone>


#####AGE & DATE_OF_BIRTH

[chime] We will now ask about their age and date ofbirth.


What will this person’s age be on April 1, 2000? Pleaseanswer after the beep. <tone>


if silence, then

What will this person’s age be on April 1, 2000? Pleaseanswer AFTER you hear the beep. <tone>



What was this person’s age on April 1, 2000? Pleaseanswer after the beep. <tone>


if silence, then

What was this person’s age on April 1, 2000? Pleaseanswer AFTER you hear the beep. <tone>


Please tell us the month, day, and year this person wasborn after the beep. <tone>


if silence, then

Please tell us the month, day, and year this person wasborn. Please answer AFTER you hear the beep. <tone>


#####ORIGIN

[chime] We will now ask about their origin

Is this person of Spanish or Hispanic origin, yes or no?<tone>

if silence, then

Is this person of Spanish or Hispanic origin? Pleasesay yes or no after the beep. <tone>

if yes, then

Which of the following best describes their origin:Mexican, Mexican-American, Chicano, Puerto Rican,Cuban, or other? <tone>

if silent, then


Which of the following best describes their origin:Mexican, Mexican-American, Chicano, Puerto Rican,Cuban, or other? Please answer AFTER you hear thebeep. <tone>

if unrecognized, then go to OH2

if other, then

Okay, to what other Spanish or Hispanic groupdo they belong? <tone>


if silent, then

What other Spanish or Hispanic group does this personbelong? Please answer AFTER you hear the beep.<tone>


<Argentinean, Colombian, Dominican, Nicaraguan,Salvadoran, Spaniard, other Hispanic>

:OH2

Please spell that after the beep. <tone>


if silent, then

Please spell the name of the Spanish or Hispanic groupAFTER the beep. <tone>


Does this person belong to any other Spanish orHispanic groups, yes or no? <tone>

if silent, then

Does this person belong to any other Spanish orHispanice groups? Please say yes or no after the beep.<tone>

<if yes, then>

Please tell us the name or names of these groupsafter the beep. <tone>


if silent, then

Please tell us the name or names of any other Spanishor Hispanic group this person belongs to AFTER thebeep. <tone>


<Mexican, Mexican American, Chicano, Puerto Rican,Cuban, Argentinean, Colombian, Dominican,Nicaraguan, Salvadoran, Spaniard, other Hispanic>

#####RACE

[chime] We will now ask about their race.

<If panel 2 or 5, then>

Does this person belong to one or more of the racesprinted under question 8 on page 1 of the question-naire, yes or no? <tone>

if silence goto RACE2

<If panel 8A or 8B, then>

Does this person belong to one or more of the racesprinted under item 9 inside the brochure, yes or no?<tone>

if silence goto RACE2

<If yes, then>

Please say the name of the race or races this personbelongs to with a short pause between each name.<tone>


if silence, then

go to RACE2

else

go to CONFIRMATION 2

<If no, then>

Please say the name of the other race or races thisperson belongs to with a short pause between eachname. <tone>


if silence, then

goto RACE2

else

goto CONFIRMATION 2

:RACE2

Is this person White? Please say yes or no after thebeep. <tone>

<if yes, then>


Does this person belong to any other races, yes orno? <tone>

if silence, then

Does this person belong to any other races? Please sayyes or no after the beep. <tone>

<if no, then>


Is this person Black, African American, or Negro, yesor no? <tone>

if silence, then

Is this person Black, African American, or Negro?Please say yes or no after the beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person an American Indian or Alaska Native, yesor no? <tone>

if silence, then

Is this person American Indian or Alaskan Native?Please say yes or no after the beep. <tone>

<if yes, then>

Please tell us the name of this person’s tribe afterthe beep. <tone>


<Cherokee, Blackfoot, Navajo, Chickasaw, Chippewa,Potawatomi, Sioux, Tohono O’Odham, Choctaw, Pima,Pueblo, Tlingit, Apache, Seminole, Iroquois, AlaskanAthabaskans, Lumbee, Cheyenne, Creek, Comanche,other tribe>

if silence, then

Please tell us the name of this person’s tribe. Pleaseanswer AFTER you hear the beep. <tone>


Please spell the name of this person’s tribe after thebeep. <tone>


if silence, then

Please spell the name of this person’s tribeAFTER you hear the beep. <tone>


Does this person belong to any other races, yesor no? <tone>

if silence, then


<if no, then>


Is this person an Asian Indian, yes or no? <tone>

if silence, then

Is this person an Asian Indian? Please say yes or noafter the beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Chinese, yes or no? <tone>

if silence, then

Is this person Chinese? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Filipino, yes or no? <tone>


if silence, then

Is this person Filipino? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Japanese, yes or no? <tone>

if silence, then

Is this person Japanese? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Korean, yes or no? <tone>

if silence, then

Is this person Korean? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Vietnamese, yes or no? <tone>

if silence, then

Is this Vietnamese? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person from some other Asian race, yes or no?<tone>

if silence, then

Is this person from some other Asian race? Please sayyes or no after the beep. <tone>

<if yes, then>

Please say the name of this person’s race after thebeep. <tone>


if silence, then

Please say the name of this person’s race AFTERyou hear the beep. <tone>


<Cambodian, Hmong, Indonesian, Pakistani, Laotian,Thai, other Asian>

Please spell the name of this person’s race after thebeep. <tone>


if silence, then

Please spell the name of this person’s race AFTERyou hear the beep. <tone>



if silence, then


<if no, then>



Is this person Native Hawaiian, yes or no? <tone>

if silence, then

Is this person Native Hawaiian? Please say yes or noafter the beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Guamanian or Chamorro, yes or no?<tone>

if silence, then

Is this person Guamanian or Chamorro? Please say yesor no after the beep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person Samoan, yes or no? <tone>

if silence, then

Is this person Samoan? Please say yes or no after thebeep. <tone>

<if yes, then>


if silence, then


<if no, then>


Is this person from some other Pacific Islander race,yes or no? <tone>

if silence, then

Is this person from some other Pacific Islander race?Please say yes or no after the beep. <tone>

<if yes, then>

Please say the name of this person’s race after thebeep. <tone>


if silence, then



<Fijian, Palauan, Tahitian, Tongan, other PacificIslander>

Please spell the name of this person’s race after thetone. <tone>


if silence, then



Is this person from some other race, yes or no? <tone>

if silence, then

Is this person from some other race? Please say yes orno after the beep. <tone>

<if yes, then>

Please say the name of this person’s race. <tone>


if silence, then



Please spell the name of this person’s race after thetone. <tone>


if silence, then





####CONFIRMATION 2

[chime] We will summarize the information you record-ed about this person.

{Silent responses should be spoken as “blank”}

Name: <first name n> <last name n>.

Sex: <sex>

Birthdate: if <month> = silence AND <day> =silence AND <year> = silence, then

say “blank”

else

<month> <day> <year>

Age: <age>

if relation, then

Relationship: <relationship>

(if initial response was ambiguous, i.e. son or daugh-ter, then the applicable phrase should be announcedhere. Choose from

adopted son, natural born son, stepson, adopteddaughter, natural born daughter, and stepdaughter

else

Association: <association>

Origin: <origin> (if blank, say “Non-Hispanic”)

Race: <race>

Is all of this information correct, yes or no? <tone>

if silence, then

Is all of this information correct? Please answer yesor no after the beep. <tone>

<if no, then>

Please tell us which item or items were incorrect andprovide the correct information after each one. <tone>


if silence, then

For each item you wish to correct, please tell us theitem and the new information AFTER you hear thebeep. <tone>


If more people, then go to PERSON 2..n

##CHECK FOR COMPLETENESS

complete = true

For person 1 to roster, do

if roster = 1, then

1.HOME_OWNER –one answer category;

2.Name—first and last name fields are treated as oneitem and together must have a minimum of threealpha characters; middle initial is not considered;

3. Sex–one answer category;

4. (Age or Date of Birth) OR (Age or year of birth) OR(Age or month and day of birth);

5.Hispanic Origin–at least one answer category;

6.Race—at least one answer category..

if 5 out of 6 of the conditions above are false, then

if 5 out of 6 fields above are silent, then

complete = false

if roster > 1, then for each person 2..n,

1.Relationship–one answer category;

2.Name—first and last name fields are treated as oneitem and together must have a minimum of threealpha characters; middle initial is not considered;

3. Sex–one answer category;

4. (Age or Date of Birth) OR (Age or year of birth) OR(Age or month and day of birth);

5.Hispanic Origin–at least one answer category;

6.Race—at least one answer category..

if 5 out of 6 of the conditions above are false, then

if 5 out of 6 fields above are silent, then

complete = false

if not complete, then

[chime]

if panel 5 or 8A, then

We did not receive enough information from you toactivate your calling card.

else

Your Census form is not complete. If you needhelp....

Please call us at 1-877-8-CENSUS for assistance. ACensus worker may contact you later to complete therest of your information.

goto GOODBYE

[chime] We will now take your calling card information.

If panel 5, then

Please remove the calling card from the insert and turnit over to see the calling card number.

If panel 8A, then

Please remove the calling card from the brochure andturn it over to see the calling card number.

<If panel 8A, then>

<look up calling card number>

Our records show that the calling card we sent youhas this ID: xxxxxxxxxxxx.

Is this correct, yes or no? <tone>

if silence, then

Is this correct? Please say yes or no after thebeep. <tone>

If yes, then

This card is valid for one year and you may beginusing it immediately.

<skip to SURVEY QUESTIONS>

If no, then

If buttons, then

Please enter all ten digits of the calling card numberwith the pushbutton keys on your telephonenow.<tone2>

else

Please say all ten digits of the calling cardnumber now. <tone>

<lookup CC #>

go to NO MATCH

<If panel 5, then>

<If buttons, then>

Please enter all ten digits of your calling card numberwith the pushbutton keys on your telephone now.<tone>

<else>

Please say the all ten digits of your calling card num-ber now. <tone>

<look up CC #>

##NO MATCH

<If no match, then>

If buttons, then

The number you entered is not in our records, pleaseenter all ten digits of your calling card number again.<tone>

else

The number you entered is not in our records, pleasesay all ten digits of your calling card number again.<tone>

<else>


<skip to SURVEY QUESTIONS>

<look up CC #>

<If no match, then>

See bailout specification

<else>


#####SURVEY QUESTIONS

[chime] We will now ask you some questions aboutthis system.

On a scale of 1 to 5, where 5 means Very Satisfied, 1means Very Dissatisfied, and 3 means neither Satisfiednor Dissatisfied, how Satisfied are you overall with thecomputerized questionnaire? <tone>

if silence, then

On a scale of 1 to 5, where 5 means Very Satisfied, 1means Very Dissatisfied, and 3 means neither Satisfiednor Dissatisfied, how Satisfied are you overall with thecomputerized questionnaire? Please answer AFTER youhear the beep. <tone>


<score = 1..5>

<if score < 3, then>

Please tell us what you disliked about the computer-ized questionnaire after the beep. <tone>


if silence, then

Please tell us what you disliked about the question-naire AFTER you hear the beep. <tone>


<if score > 3, then>

Please tell us what you liked about the computer-ized questionnaire after the beep. <tone>


if silence, then

Please tell us what you liked about the questionnaireAFTER you hear the beep. <tone>


Were you able to fully understand the computer, yes orno? <tone>

if silence, then

Were you fully able to understand the computer?Please answer yes or no after the beep. <tone>

<if no, then>

Please tell us what you did not understand after thebeep. <tone>


if silence, then

Please tell us what you did not understandAFTER you hear the beep. <tone>


Was the computer able to fully understand you, yes orno? <tone>

if silence, then

Was the computer fully able to understand you? Pleasesay yes or no after the beep. <tone>

<if no, then>

Please tell us what the computer did not understandafter the beep. <tone>


if silence, then

Please tell us what the computer did not understandAFTER you hear the beep. <tone>


Was there anything about the questionnaire that wasconfusing or frustrating, yes or no? <tone>

if silence, then

Was there anything about the questionnaire that wasconfusing or frustrating? Please answer yes or no afterthe beep. <tone>

<if yes, then>

Please tell us what was confusing or frustrating afterthe beep. <tone>


if silence, then

Please tell us what was confusing or frustrating AFTERyou hear the beep. <tone>


Did you have too much time, too little time, or just theright amount of time to answer the questions? <tone>


if silence, then

Did you have too much time, too little time, or just theright amount of time to answer the questions? Pleaseanswer AFTER you hear the beep. <tone>


Please tell us your suggestions about improving thecomputerized questionnaire after the beep. <tone>


if silence, then

Please tell us your suggestions about improving thequestionnaire AFTER you hear the beep. <tone>


#####GOODBYE

[chime]

Thank you for your help with the 2000 Census. You donot need to mail in your paper questionnaire.Goodbye.


ASQ Usability Questions

The ASQ concluded with a set of questions to assessthe respondents’ satisfaction with the data collectionmethod:

• On a scale of one to five, where five means verysatisfied, one means very dissatisfied, and threemeans neither satisfied nor dissatisfied, how satis-fied are you overall with the computerized ques-tionnaire?

If response to the above question was less thanthree:Please tell us what you disliked about the computer-ized questionnaire after the beep.

If response to the above question was greater thanthree:Please tell us what you liked about the computer-ized questionnaire after the beep.

• Were you able to fully understand the computer, yesor no?

If response to above question was “no”:Please tell us what you did not understand after thebeep.

• Was the computer able to fully understand you, yesor no?

If response to above question was “no”:Please tell us what the computer did not understandafter the beep.

• Was there anything about the questionnaire thatwas confusing or frustrating, yes or no?

If response to above question was “yes”:Please tell us what was confusing or frustratingafter the beep.

• Did you have too much time, too little time, or justthe right amount of time to answer the questions?

• Please tell us your suggestions about improving thecomputerized questionnaire after the beep.

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 D-1

Appendix D

Census 2000

Telephone Interviewing Specifications forShort Form

SPECS: All numeric fields must be stored as right justified. All character fields must be stored as leftjustified.

Refer to the last page for a definition of completedinterviews.

D-1 = Interview with a 22 character census ID number.

D-10 = Interview without a 22 character census IDNumber, assign processing ID as noted below andstart at >address_a<

Assignment of processing id:

Character 1-2 66 = TQA generated BCF interviews

Character 3-5 000 = English 001 = English (in sequence for 000

is filled) 002 = Spanish003 = Chinese004 = Korean005 = Tagalog006 = Vietnamese007 = English (PR)008 = Spanish (PR)

Character 6-12 sequence number 0000001 9999999

Character 13-14MAD97 check digits

________________________________________________________

SPECS: If census identification number was forwardedfrom TQA, skip to >POP_count<using formtype to indi-cate which form to complete. If NO census identifica-tion number was provided, skip to >ID<. The censusidentification number must be part of the output infor-mation for the Bureau of the Census with the shortform data from the interviews.

>ID<

If you have your census form available, please refer tothe census identification number located on the backpage underneath the bar code. What is the ID numberon your questionnaire?

_ _ _ _ _ - _ _ _ _ _ _ _ - _ _ - _ _ _ - _ _ _ - _ _ (allow 22characters)—Skip to >POP_count<<

(N) Not available,

If no ID number, set formtype=D10 and go to>address_a<;

If valid ID, and Phone Num. available, go to >ANIchk<,

If valid ID, and No Phone Num., go to >GetPhone<

OUTPUT SPECS: Store in Record 1, at location 2.

______________________________________________________

Programming note: If uheflag=1, use second fill. Else,use first fill.

>address_a<

We need to be sure that everyone is counted correctlyin the census. I’d like to take some information aboutyour household, starting with your home address.

What is the mailing address where you (lived onSaturday, April 1, 2000/ live or stay MOST OF THETIME)?

Mailing Address: _____________________ >houses-treet1< (allow 63 characters—If address is a P.O. Boxaddress, Rural Route/Box address or No address, storeaddress collected by agent starting at character 9 sothat 8-letter string may be stored in characters 1-8.See specs below. For house number and street/roadname style address, start address at character 1.)

SPECS: P.O. Box address, Rural Route/Box address orNo address are stored in >housestreet1<.

AND CHECK ONE BOX IF APPROPRIATE:

[ ] P.O. Box address — Skip to >aptno1<<

SPECS: For output, fill >housestreet1< to position 62.At position 63, fill with a “P.” Set nohouse = 1.

[ ] Rural Route/Box address – Skip to >aptno1<

SPECS: For output, fill >housestreet1< to position 62.At position 63, fill with an “R.”. Set nohouse = 1

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 E-1

Appendix E

[ ] No address on April 1, or address is a locationdescription such as a park, or street names – Skipto >zip1<

SPECS: For output, fill >housestreet1< to position 62.At position 63, fill with an “O.” Also, if “No address”box is marked, store “1” in variable bcmailno at loca-tion 81. Set nohouse = 1.

OUTPUT SPECS: Store in Record 1 for D-10 at the fol-lowing locations:

bcmailno: location 81housestreet1: location 82

Programming note: Do NOT advance to the nextscreen when write-in field is blank unless the “Noaddress” box is marked.

_____________________________________________________

>aptno1<

Do you have an apartment number?

Yes — ____________ Apartment number (allow 16characters)

No

OUTPUT SPECS: Store apartment number in Record 1for D-10 at location 145.

_____________________________________________________

>zip1<

What is the ZIP code?

___________ (allow 5 characters)

OUTPUT SPECS: Store in Record 1 for D-10 atlocation 195.

_____________________________________________________

>city1<

What is the name of your city or town?


OUTPUT SPECS: Store in Record 1 for D-10 at location 161.

_____________________________________________________

>state1<

What state? SELECT THE STATE USING THE ARROWKEYS IF NECESSARY AND PRESS THE ENTER KEY.

_________ (allow 2 characters)

_____________________________________________________

>county1<

What county is that city or town in?

____________________________(allow 16 characters)

D Don’t knowR Refused

OUTPUT SPECS: Store in Record 1 for D-10 at location177.

Programming note: From the Mailing Address screen,do not advance to the next screen without the city andstate.

_____________________________________________________

SPECS: Ask if nohouse=1. Else, skip to >bcallres<

>bchsnnno<

Do you have a street address with a house number?

(1) Yes — Skip to >housestreet2<<

(2) No – Skip to >bcallres<

OUTPUT SPECS: If 2, store “1” in Record 1 for D-10 atlocation 200, else leave this location blank.

Programming note: Do NOT advance to the nextscreen without a Yes or No answer.

_____________________________________________________

>housestreet2<

E-2 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

(1) AL--Alabama (19) LA--Louisiana (37) OK--Oklahoma

(2) AK--Alaska (20) ME--Maine (38) OR--Oregon

(3) AZ--Arizona (21) MD--Maryland (39) PA--Pennsylvania

(4) AR--Arkansas (22) MA--Massachusetts 40) RI--Rhode Island

(5) CA--California (23) MI--Michigan (41) SC--South Carolina

(6) CO--Colorado (24) MN--Minnesota (42) SD--South Dakota

(7) CT--Connecticut (25) MS--Mississippi (43) TN--Tennessee

(8) DE--Delaware (26) MO--Missouri (44) TX--Texas

(9) DC--District of Columbia (27) MT--Montana (45) UT--Utah

(10) FL--Florida (28) NE--Nebraska (46) VT--Vermont

(11) GA--Georgia (29) NV--Nevada (47) VA--Virginia

(12) HI--Hawaii (30) NH--New Hampshire (48) WA--Washington

(13) ID--Idaho (31) NJ--New Jersey (49) WV--West Virginia

(14) IL--Illinois (32) NM--New Mexico (50) WI--Wisconsin

(15) IN--Indiana (33) NY--New York (51) WY--Wyoming

(16) IA--Iowa (34) NC--North Carolina

(17) KS--Kansas (35) ND--North Dakota

(18) KY--Kentucky (36) OH--Ohio

SPECS: Do NOT output codes. Output 2-letter abbreviation associated with codes.


House number and street/road name_____________________________ (allow 34 characters)


_____________________________________________________

>aptno2<

Do you have an apartment number?

Yes — ______ Apartment number (allow 16 characters))No

OUTPUT SPECS: Store apartment number in Record 1for D-10 at location 235.

_____________________________________________________

>zip2<

What is the ZIP code?



_____________________________________________________

>city2<

What is the name of your city or town?



_____________________________________________________

>state2<

What state? SELECT THE STATE USING THE ARROWKEYS IF NECESSARY AND PRESS THE ENTER KEY.

_________ (allow 2 characters)

>county2<

What county is that city or town in?

____________________________ (allow 16 characters)

D Don’t know

R Refused

SPECS: Go to >bcallres<


Programming note: From the Second Address screen,do NOT advance to the next screen without the cityand state.

_____________________________________________________

>bcallres<

Programming Note: Ask if formtype=D10; if uhe-flag=1 use second fill, else use first fill.

(1) Yes

(2) No

SPEC: If Phone Num. available, go to >ANIchk<

_____________________________________________________

SPECS: Ask if Phone Num. available, else go to>GetPhone<

>ANIchk<

We used our caller ID system to capture the phonenumber you are calling from. Is (FILL with verified/corrected ANI) your correct home phone number?

(1)Yes — Store as >phonarea<, >phonepre<,>phonesfx<<

(2)No — read::

What is your home phone number startingwith your area code?

____________________ ** (area (prefix) (suffix)code)

**Note: Phone number is captured as one field, butoutput as three fields.

SPECS: If formtype=D-1, go to >POP_count<If formtype=D-10 and:

>bcallres< =1 then go to >POP_count<>bcallres< =2 then go to >tenure<

OUTPUT SPECS: Store the area code as >phonarea<;store the prefix as >phonepre<; store the suffix as>phonesfx<.

Store in Record 1 as follows:


(1) AL--Alabama (19) LA--Louisiana (37) OK--Oklahoma

(2) AK--Alaska (20) ME--Maine (38) OR--Oregon

(3) AZ--Arizona (21) MD--Maryland (39) PA--Pennsylvania

(4) AR--Arkansas (22) MA--Massachusetts 40) RI--Rhode Island

(5) CA--California (23) MI--Michigan (41) SC--South Carolina

(6) CO--Colorado (24) MN--Minnesota (42) SD--South Dakota

(7) CT--Connecticut (25) MS--Mississippi (43) TN--Tennessee

(8) DE--Delaware (26) MO--Missouri (44) TX--Texas

(9) DC--District of Columbia (27) MT--Montana (45) UT--Utah

(10) FL--Florida (28) NE--Nebraska (46) VT--Vermont

(11) GA--Georgia (29) NV--Nevada (47) VA--Virginia

(12) HI--Hawaii (30) NH--New Hampshire (48) WA--Washington

(13) ID--Idaho (31) NJ--New Jersey (49) WV--West Virginia

(14) IL--Illinois (32) NM--New Mexico (50) WI--Wisconsin

(15) IN--Indiana (33) NY--New York (51) WY--Wyoming

(16) IA--Iowa (34) NC--North Carolina

(17) KS--Kansas (35) ND--North Dakota

(18) KY--Kentucky (36) OH--Ohio

SPECS: Do NOT output codes. Output 2-letter abbreviation associated with codes.


D-10 D-1>phonarea<: location 291 location 272>phonepre<: location 294 location 275>phonesfx<: location 297 location 278

_____________________________________________________

>GetPhone<

SPEC: Ask if No Phone Num.

In case we need to contact you later, please give meyour home phone number starting with your areacode.

___________________ **(area (prefix) (suffix)code)

_ Don’t know/Refused

**Note: Phone number is captured as one field, butoutput as three fields.

SPECS: If formtype=D-1, go to >POP_count<If formtype=D-10 and:

>bcallres< =1 then go to >POP_count<>bcallres< =2 then go to >tenure<

OUTPUT SPECS: Store the area code as >phonarea<;store the prefix as >phonepre<; store the suffix as>phonesfx<.

Store in Record 1 as follows:

D-10 D-1>phonarea<: location 291 location 272>phonepre<: location 294 location 275>phonesfx<: location 297 location 278

_____________________________________________________

SPECS: Include the residence rules job aid from knowl-edge data base as HELP.

>POP_count<

(How many people were living or staying in thishouse, apartment, or mobile home on April 1, 2000/How many people were living or staying in the house,apartment, or mobile home where you live or stayMOST OF THE TIME)?

___ (allow 2 characters)

(H) HELP

OUTPUT SPECS: Store in Record 1 for D-1 or D-10, atlocation 83.

_____________________________________________________

>tenure<

(Is this house, apartment, or mobile home—/ Is thehouse, apartment, or mobile home where you live orstay MOST OF THE TIME—-))

(1) Owned by you or someone in this household witha mortgage or loan?

(2) Owned by you or someone in this household freeand clear without a mortgage or loan?

(3) Rented for cash rent?

(4) Occupied without payment of cash rent?

D Don’t know

R Refused

OUTPUT SPECS: Store for D-1 in Record 1, at location282. Store for D-10, Record 1, at location 301.

_____________________________________________________

>partial_roster<

This screen does not exactly reflect the OSS format.

Programming note: Ask if bcallres=2, else go to >ros-ter<; If uheflag=1 use second fill, else use first fill.

(What are the names of the persons who were living orstaying in this house, apartment, or mobile home onApril 1, 2000/ What are the names of the persons whowere living or staying in the house, apartment, ormobile home where you live or stay MOST OF THETIME)? Start with yourself or a person living with youwho was not counted.

ENTER NAMESMIDDLE

FIRST NAME INITIAL LAST NAME CODE

(ALLOW ENOUGH SPACE FOR PEOPLE)Allow 15 characters for last name. Allow 13 characters for first name.Allow 1 characters for middle initial.Allow 1 character for code.

Add boxes for indicating “respondent” and “proxy.”

OUTPUT SPECS: Store a “2” in location 1 for recordtype 2. Store in Record 2, for D-1 or

D-10 at locations:Last Name: location 47First Name: location 63Middle Initial: location 76

Store a “1” in PNUM at location 42 for person on line 1of the roster; store a “2” in PNUM at location 42 forperson on line 2 of the roster; etc._____________________________________________________


>roster<This screen does not exactly reflect the OSS format.

What are the names of all persons who were living orstaying (in this house, apartment, or mobile home onApril 1, 2000/ at the house, apartment, or mobilehome where you live or stay MOST OF THE TIME)?Start with the name of one of the people living herewho owns, is buying, or rents this house, apartment,or mobile home.

ENTER NAMESMIDDLE

FIRST NAME INITIAL LAST NAME CODE(ALLOW ENOUGH SPACE FOR PEOPLE)

Allow 15 characters for last name. Allow 13 characters for first name.Allow 1 characters for middle initial.Allow 1 character for code.

Add boxes for indicating “respondent” and “proxy.”

OUTPUT SPECS: Store a “2” in location 1 for recordtype 2. Store in Record 2, for D-1 or

D-10 at locations:Last Name: location 47First Name: location 63Middle Initial: location 76

Store a “1” in PNUM at location 42 for person on line 1of the roster; store a “2” in PNUM at location 42 forperson on line 2 of the roster; etc.

_____________________________________________________

>coverage1<

Did anyone else such as housemates, roommates, live-in employees, boarders, foster children or anyone tem-porarily away on business or vacation live (at thisaddress on April 1, 2000/ at the place where you liveMOST OF THE TIME)?

(1) Yes—Ask: What is/are their name(s)??

(2) No

SPECS: If “1” go to >roster<. Add the name(s) to thelist and enter an “A” in the “Code” column for eachname given. If “2” go to >coverage2<. Increase thenumber in >POP_count< if names are added.

_____________________________________________________

>coverage2<

Did you include any people who were living away atcollege, in the Armed Forces and living somewhereelse, in a correctional facility, in a mental hospital, in a

nursing home, hospice or ward for the chronically ill,or staying at another residence most of the weekwhile working?

(1) Yes—Ask: What is/are their name(s)??

(2) No

SPECS: If “1” go to >roster<. Enter a “D” in the “Code”column beside each name given.

Decrease the number in >POP_count< if names aredeleted.

_____________________________________________________

>adc_names<

I’m going to read you the list of people to verify thatall names are listed correctly. (READ NAMES AND VERI-FY SPELLING)

MAKE SURE [fill with respondent’s name] IS LISTEDON ROSTER EXCEPT FOR A PROXY

(P) All correct

(A) Add person not listed

(D) Delete person listed

(C) Spelling Change

(U) Undelete person listed

ENTER LINE NUMBER OF PERSON:____(To advance to>Spelling< and take the appropriate action above)

SPECS: If “A”, increase >POP_Count<, If “D”, decrease>POP_count<.

_____________________________________________________

INTERVIEWER SCREEN

>resp@1<(Ask if necessary.)Enter line number of person you are talking to ____

SPECS: We need to allow for proxy interviews (code99). A “proxy interviewer” is someone providing theinterview that is not on the >roster<.

_____________________________________________________

>Spelling<

MAKE THE CORRECTIONS NEEDED:

First _____________________

MI ______

Last ________________________

_____________________________________________________


SPECS: Ask the >relation< question of EVERYONE listedon the Roster except the person on line one. For per-son one, store 0 in >relation<. Ask the >otherrel<question only if the answer is “10”. Then continuewith >sex1< starting with the person on line one.

>relation<

How (are you /is ...) related to (fill with the name online 1)*?

(1) Husband/wife

(2) Natural-born son/daughter

(3) Adopted son/daughter

(4) Stepson/stepdaughter

(5) Brother/sister

(6) Father/mother

(7) Grandchild

(8) Parent-in-law

(9) Son-in-law/daughter-in-law

(10) Other relative—Skip to >otherrel<<

(11) Roomer, boarder

(12) Housemate, roommate

(13) Unmarried partner

(14) Foster child

(15) Other nonrelative


Skip to >sex1< except for category (10).

* If respondent is person on line 1, use “you” insteadof name of person on line 1.

SPECS: If “D” or “R”, store 0.

OUTPUT SPECS: Store in Record 2 for D-1 or D-10, atlocation 77.

_____________________________________________________

>otherrel<

SPECS: Ask only if needed or fill with information therespondent provided when asked >relation<.

How (are you/is...) related?

___________________________ (allow for 12 characters)


SPECS: If “D” or “R”, store a “Blank”.

OUTPUT SPECS: Store in Record 2, for D-1 or D-10, atlocation 79.

_____________________________________________________

>sex1<

SPECS: FOR THE RESPONDENT SCREEN ONLY, ADD:ASK ONLY IF NECESSARY.

SPECS: Ask question of everyone listed on the Rosterbefore continuing to >dob<.

(Are you/Is...) male or female?

(1) Male

(2) Female


SPECS: If “D” or “R”, store 0.


_____________________________________________________

SPECS: Ask questions of EACH person on the rosterbefore continuing to >hisp_origin1<.

>dob<

What is (your/...’s) date of birth?

Month Day Year of birth_____ _____ __/__/__/__ (allow 1884-2000)(01-12) (01-31)dob@mth dob@dy dob@yr

D Don’t know—Skip to >age<<R Refused—Skip to >age<<

SPECS: If any part (month, day, or year of birth) isDon’t know or Refused, skip to >age<. For year ofbirth, output full year such as “1985” and not “985.” If“D” or “R” in any field, store a “Blank”. Output>dob@mth< as >DOBMONTH_4<; >dob@dy< as >DOB-DAYXX_4<; and >dob@yr< as >DOBYEARX_4<.

OUTPUT SPECS: Store in Record 2, for D-1 or D-10 andthe following locations:

Month: location 95

Day: location 97

Year of Birth: location 99

_____________________________________________________

SPECS: If computed age is less than 1 year, substitutethe “computed age” with appropriate “months/weeks”.This screen does not exactly reflect the OSS format.

>ver_age<

So (were you/was...) (computed age) years old onApril 1, 2000?


(1) Yes—Skip to >hisp_origin1<<

(2) No

D Don’t know—Skip to >hisp_origin1<<

R Refused—Skip to >hisp_origin1<<

SPECS: If 1, store computed age in >age<. If the com-puted age is less than 1 year, store “0”. If “D” or “R”,store a “Blank”.


_____________________________________________________

>age<

What was (your/...’s) age on April 1, 2000? IF CALLERDOES NOT KNOW THE EXACT AGE – Please estimate(your/…’s) age on April 1, 2000?

___ (allow 0-116)

D Don’t know

R Refused

SPECS: Store age in >age<. If the computed age is lessthan 1 year, store “0”. If “D” or “R”, store a “Blank”.


_____________________________________________________

SPECS: Ask questions of EACH person on roster beforecontinuing to >race<.

>hisp_origin1<

(Are you/Is ...) Spanish, Hispanic, or Latino? READ IFNECESSARY: FOR EXAMPLE, MEXICAN, MEXICAN AMERI-CAN, CHICANO, PUERTO RICAN, CUBAN, OR ANOTHERSPANISH, HISPANIC, OR LATINO GROUP.

(1)Yes – continue to >hisp_origin2<

(2)No—Skip to >race<<

D Don’t know—Skip to >race<<

R Refused—Skip to >race<<

SPECS: If 2, store 1 in >HISPCB01_5<. If “D” or “R”,store 0.


_____________________________________________________

>hisp_origin2<

SPECS: Accept only ONE response.

Which one of the following Spanish, Hispanic, orLatino groups (do you/does …) identify with?Mexican, Mexican American, Chicano, Puerto Rican,Cuban, or another Spanish, Hispanic or Latino group.

(1)Mexican, Mexican American, Chicano—Skip to>race<<

(2)Puerto Rican—Skip to >race<<

(3)Cuban—Skip to >race<<

(4)Other Spanish/Hispanic/Latino—continue to>othr_sp1<<



SPECS: Store answers of 1, 2, 3, 4 as follows:

Current answer Stored value Variable1 1 >HISPCB02_5<2 1 >HISPCB03_5<

3 1 >HISPCB04_5<4 1 >HISPCB05_5<

OUTPUT SPECS: Store in Record 2, for D-1 or D-10, atthe following locations:

HISPCB02_5: location 104HISPCB03_5: location 105HISPCB04_5: location 106HISPCB05_5: location 107

_____________________________________________________

>othr_sp1<

What is this group?

(1) Argentinean—Skip to >race<<

(2) Colombian—Skip to >race<<(3) Dominican—Skip to >race<<(4) Nicaraguan—Skip to >race<<

(5) Salvadoran—Skip to >race<<(6) Spaniard—Skip to >race<<(7) Other—Skip to >othr_sp<<



SPECS: Store words corresponding to categories 1through 6 in >HISPANWI_5<.


_____________________________________________________

>othr_sp<

What is the name of the other Hispanic group?

__________(allow for 19 characters)


D Don’t know

R Refused

SPECS: Store >othr_sp< in >HISPANWI_5<. If “D” or “R”,store “Blank.”


_____________________________________________________

SPECS: Respondent may choose one or more cate-gories.

SPECS: Ask questions of each person listed on theRoster in the following order:

>race<, >othr_race<, >amer_ind<, >asian_group<,>othr_asian<, >pacific_group<, and >othr_pacific<(when appropriate) before continuing to the next per-son.

>race<

I’m going to read a list of race categories. Pleasechoose one or more categories that best indicate(your/...’s) race. (Are you/Is...) White? Black, AfricanAmerican or Negro? American Indian or Alaska Native?Asian? Native Hawaiian? Other Pacific Islander? orSome other race?

(1) White

(2)Black, African American, or Negro

(3)American Indian or Alaska Native—Skip to>amer_ind<<

(4)Asian—Skip to >asian_group<<

(5)Native Hawaiian

(6)Other Pacific Islander—Skip to >pacific_group<<

(7)Some other race—Skip to >othr_race<<

D Don’t know

R Refused

SPECS: If “D” or “R”, store 0 in each variable. Storeanswers of (1) - (7) as follows:

Current answer Stored value Variable

1 1 >RACECB01_6<

2 1 >RACECB02_6<

3 1 >RACECB03_6<

5 1 >RACECB11_6<

7 1 >RACECB15_6<

Store 0 in all variables without a value of 1.


RACECB01_6: location 127

RACECB02_6: location 128

RACECB03_6: location 129RACECB11_6: location 137

RACECB15_6: location 141_____________________________________________________

>othr_race<

What is the name of (your/...’s) race?

______________________________ >othr_race1<(allow for 19 characters)

D Don’t know

R Refused

SPECS: If “D” or “R”, store a “Blank.”


>amer_ind<

What is the name of (your/...’s) enrolled or principaltribe?

(H) HELP

_______________________________ >amer_ind1< (allow for 19 characters)

D Don’t know

R Refused


OUTPUT SPECS: Store in Record 2, for D-1 or D-10, atlocation 142._____________________________________________________

>H_AMERIND<

Add a pop-up or help screen for agents to select thefollowing for the >amer_ind< screen.

Cherokee BlackfootNavajo ChickasawChippewa PotawatomiSioux Tohono O’OdhamChoctaw PimaPueblo TlingitApache SeminoleIroquois Alaskan AthabaskansLumbee Cheyenne

Creek Comanche

SPECS: More than one category is acceptable. Whenstoring more than one category selection, use whitespace delimiter between the selections.

_____________________________________________________


>asian_group<

To what Asian group (do you/does...) belong? (READCATEGORIES.)

(1) Asian Indian

(2) Chinese

(3) Filipino

(4) Japanese

(5) Korean

(6) Vietnamese

(7) Other Asian—Skip to >othr_asian<<

D Don’t know

R Refused

SPECS: If “D” or “R”, store 0. Store answers of (1) - (7)as follows:

Current answer Stored value Variable1 1 >RACECB04_6<2 1 >RACECB05_6<

3 1 >RACECB06_6<4 1 >RACECB07_6<5 1 >RACECB08_6<

6 1 >RACECB09_6<7 1 >RACECB10_6<


RACECB04_6: location 130RACECB05_6: location 131

RACECB06_6: location 132RACECB07_6: location 133RACECB08_6: location 134

RACECB09_6: location 135RACECB10_6: location 136_____________________________________________________

>othr_asian<

What other Asian group (do you/does...) belong?

(H) HELP

_____________________ >othr_asian1<(allow 9 characters)

D Don’t know

R Refused


NOTE: If answers reported for both >othr_asian1< and>othr_pacific1<, combine into one output variable andstore in >AISPIWIN_6<. When storing , use white space

delimiter between the two answers. Otherwise, storesingle answer in >AISPIWIN_6<.


_____________________________________________________

>H_OTHRASIAN<

Add a pop-up or help screen for agents to select thefollowing for the >othr_asian< screen:

Cambodian

Hmong

Indonesian

Pakistani

Laotian

Thai

SPECS: More than one category is acceptable.

_____________________________________________________

>pacific_group<


To what Pacific Islander group (do you/does ...)belong? READ CATEGORIES.

(1) Guamanian or Chamorro

(2)Samoan

(3)Other Pacific Islander—Skip to >othr_pacific<<

D Don’t know

R Refused

SPECS: If “1”, store 1 in >RACECB12_6<. If “2”, store 1in >RACECB13_6<. If “3”, store 1 in >RACECB14_6<. If“D” or “R”, store a “Blank” in >RACECB14_6<.

OUTPUT SPECS: Store in Record 2 for D-1 or D-10, atthe following locations:

RACECB12_6: location 138RACECB13_6: location 139RACECB14_6: location 140_____________________________________________________

>othr_pacific<

What other Pacific Islander group (do you/does...)belong?

(H) HELP

______________________________>othr_pacific1<(allow 9 characters)

D Don’t know

R Refused



NOTE: If answers reported for both >othr_asian1< and>othr_pacific1<, combine into one output variable andstore in >AISPIWIN_6<. When storing, use white spacedelimiter between the two answers. Otherwise, storeone answer in >AISPIWIN_6<.


_____________________________________________________

>H_OTHPACIF<

Add a pop-up or help screen for agents to select thefollowing for the >othr_pacific< screen:

Fijian

Palauan

Tahitian

Tongan


_____________________________________________________

If CEFU case, skip to >THE END< of CEFU script.Otherwise, go to >closing<.

_____________________________________________________

>closing<

SPECS: If custsat=missing, follow path A. If custsat=1,follow path B.

A:

This completes all the questions. Thank you for tak-ing part in Census 2000.

B:

This completes all the questions. Thank you for tak-ing part in Census 2000.

Before you hang-up, we would appreciate feedbackregarding the service you received today. I’m going to

transfer you to our automated customer satisfactionsurvey, which on average takes less than 3 minutes tocomplete.

Programming note: Include the following on this

screen.

IF ASKED WHAT TO DO ABOUT THE CENSUS FORM,PLEASE RESPOND—

Since you have given me your census information, youmay discard the form you received in the mail.

_____________________________________________________

SPECS: If all of the following fields in >mail_add< are

not blank: housestreet1, city1, state1, zip1 and acomplete interview is collected, then set Nomail=1.

END INTERVIEW

TO QUALIFY AS A COMPLETED INTERVIEW

There must be complete answers* for any two of

the following questions for each person on the

roster:

>relation< (except person on line 1)

>sex1<

>age<

if >hisp_origin1< = 2 OR if >hisp_origin1< = 1, then

>hisp_origin2< must be answered

>race<

* “Don’t know” or “Refused” do not qualify as an“answer.”

_____________________________________________________


U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 F-1

Appendix FInternet Census Form

F-2 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 G-1

Internet Usage Survey Questionnaire

Screen name: I_INTRO1

———————————————————————————

1.01 I_INTRO1

Hello, my name is {DISPLAY INTERVIEW NAME} and Iam calling for the Census Bureau to follow-up onsome of the information that you provided to theCensus earlier this year.

I’d like to confirm that I am calling the correcthousehold. According to our records, the addressfor this household is{DISPLAY ADDRESS}. Is this correct?

[IF R REFUSES, GO TO RESULT CODE AND CODECASE REFUSAL. IF R ANSWERS DON’T

KNOW ADDRESS, ASK TO SPEAK TO SOMEONE WHODOES KNOW. IF NO ONE AVAILABLE, CODE A CALLBACK.]

1. YES

2. NO

3. TELEPHONE COMPANY RECORDING

AM. ANSWERING MACHINE

RT. RETRY DIALING

GT. GO TO RESULT

_____________________________________________________

Screen name: I_INTRO2

_____________________________________________________

1.02 I_INTRO2

May I speak to {the person who filled out theCensus Form}?

[IF R REFUSES, GO TO RESULTS AND CODE CASEREFUSAL.

IF R ANSWERS DON’T KNOW PERSON, CODE 4 -NEVER HEARD OF SUBJECT.]

[VERIFY R IS 18 OR OLDER, IF NOT CODE 4 AND GOTO NEXT PERSON.]

1. SUBJECT SPEAKING/COMING TO PHONE

2. SUBJECT LIVES HERE - NEEDS APPOINTMENT

3 SUBJECT KNOWN LIVES AT ANOTHER NUMBER

4. NEVER HEARD OF SUBJECT

GT. GO TO RESULT

_____________________________________________________

Screen name: BEGIN

_____________________________________________________

2.00 BEGIN

[Hello, my name is {DISPLAY INTERVIEWER NAME} and Iam calling for the Census Bureau to follow-up on someof the information that you provided to the Censusearlier this year.]

Thank you for mailing in your Census Form. We areinterested in learning your opinion about using theInternet to fill out your Census Form. Your participa-tion is voluntary and will only take 5 minutes. We willnot use your name in any of our reports and youranswers will not be shared with with anyone who isnot part of this project.

( )

1. CONTINUE

GT. GO TO RESULT

_____________________________________________________

Screen name: INTERNET

2.01 INTERNET

Do you have access to the Internet at...

[0=DON’T UNDERSTAND OR KNOW INTERNET, 1=YES, 2=NO]

A. At Home ( )

B. At Work ( )

C. At School ( )

D. At the Library ( )

E. Or anywhere else [SPECIFY] ( )_____________________________________________________

Appendix G

Screen name: Q4

_____________________________________________________

2.04 Q4

Did you know that you could have filled out yourCensus Form on the Internet?

( )

1. YES

2. NO_____________________________________________________

Screen name: Q5

_____________________________________________________

2.05 Q5

Why didn’t you use the Internet to fill out your CensusForm?

[CODE ALL THAT APPLY, CTRL/P TO EXIT]

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )

1. DON’T HAVE ACCESS TO A COMPUTER

2. HAVE ACCESS TO A COMPUTER, BUT IT DOESN’THAVE INTERNET ACCESS

3. CONCERNED ABOUT PRIVACY OF MY ANSWERS

4. NOT ALLOWED TO USE INTERNET AT WORK FORTHIS PURPOSE

5. DON’T LIKE THE INTERNET

6. DON’T HAVE ENOUGH INTERNET EXPERIENCE TOUSE IT COMFORTABLY

7. DON’T THINK THE INTERNET WOULD COLLECTDATA IN ACCURATE WAY

8. BLINDNESS, OTHER DISABILITY PREVENTS USE

9. OTHER (SPECIFY)_____________________________________________________

Screen name: Q6_____________________________________________________

2.06 Q6

The Census Bureau {would like to} offer people theopportunity to fill out their Census Form on theInternet. If the Census Bureau gave you that option,would you use the Internet to fill out your CensusForm?

( )

1. YES

2. NO_____________________________________________________

Screen name: Q7

_____________________________________________________

2.07 Q7

Why would you use the Internet to fill out your CensusForm?

[CODE ALL THAT APPLY, CTRL/P TO EXIT]

( ) ( ) ( ) ( )

1. IT WOULD BE EASIER THAN FILLING OUT THE FORM

2. IT WOULDN’T TAKE AS LONG

3. IT WOULD BE FUN, INTERESTING OR A NEW EXPERI-ENCE

4. OTHER [SPECIFY]_____________________________________________________

Screen name: Q8

_____________________________________________________

2.08 Q8

Why wouldn’t you use the Internet to fill out yourCensus Form?

[CODE ALL THAT APPLY. CTRL/P TO EXIT]

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )

1. DON’T HAVE ACCESS TO A COMPUTER

2. HAVE ACCESS TO A COMPUTER, BUT ITDOESN’T HAVE INTERNET ACCESS

3. CONCERNED ABOUT PRIVACY OF MY ANSWERS

4. NOT ALLOWED TO USE INTERNET AT WORK FORTHIS PURPOSE

5. DON’T LIKE THE INTERNET

6. DON’T HAVE ENOUGH INTERNET EXPERIENCE TO USEIT COMFORTABLY

7. DON’T THINK THE INTERNET WOULD COLLECT DATAIN ACCURATE WAY

8. BLINDNESS, OTHER DISABILITY PREVENTS USE

9. OTHER [SPECIFY]_____________________________________________________

Screen name: Q9_____________________________________________________

2.09 Q9

When you filled out your Census Form, did you knowabout the offer of a free calling card worth 30 minutesof long distance calls?

( )

1. YES

2. NO

_____________________________________________________

G-2 Results From the Response Mode and Incentive Experiment in 2000 U.S. Census Bureau

Screen name: Q10_____________________________________________________2.10 Q10

What do you remember about this offer?

( )

1. HAD TO USE THE INTERNET TO GET THE FREE MINUTES

2. NO CONDITIONS APPLIED (I SHOULD HAVE GOTTENTHE FREE MINUTES)

3. OTHER CONDITIONS APPLIED_____________________________________________________

Screen name: Q11A

_____________________________________________________

2.1101 Q11A

Would you have used the Internet to respond to theCensus if we rewarded you with a calling cardworth.60 free minutes of long distance calls?

( )

1. YES

2. NO_____________________________________________________

Screen name: Q11B

_____________________________________________________

2.1102 Q11B

[Would you have used the Internet to respond to theCensus if we rewarded you with a calling cardworth...]

90 free minutes of long distance calls?

( )

1. YES

2. NO_____________________________________________________

Screen name: Q12

_____________________________________________________

2.12 Q12

Suppose we gave you a calling card that allowed youto make 30 minutes of free long distance phone calls,but only if you used the Internet to fill out yourCensus Form. Would you use the Internet to fill outyour Census Form then?

( )

1. YES

2. NO_____________________________________________________

Screen name: Q13A_____________________________________________________

2.1301 Q13AWould you use the Internet to respond to the Census ifwe rewarded you with a calling card worth 60 freeminutes of long distance calls?

( )

1. YES

2. NO_____________________________________________________

Screen name: Q13B_____________________________________________________

2.1302 Q13B

[Would you use the Internet to respond to the Censusif we rewarded you with a calling card worth...]

90 free minutes of long distance calls?

( )

1. YES

2. NO_____________________________________________________

Screen name: END2.14 END

Thank you very much for your time.

[PRESS ENTER TO CONTINUE]

_____________________________________________________

Screen name: COLLECT_____________________________________________________

2.15 COLLECT

Can you please give me the name of the person inyour household who filled out the Census Form?

( )

1. YES

FIRST NAME

LAST NAME:

_____________________________________________________

Screen name: COLLECT_____________________________________________________

2.15 COLLECT

Can you please give me the name of the person inyour household who filled out the Census Form?

( )

1. YES

FIRST NAME:

LAST NAME:

_____________________________________________________

U.S. Census Bureau Results From the Response Mode and Incentive Experiment in 2000 G-3

Census 2000 Topic Report No. 7TR-7

Issued March 2004

Data Processing in Census 2000

U.S.Department of CommerceEconomics and Statistics Administration

U.S. CENSUS BUREAU
