issued february 2004 puerto rico - census.gov · 2004-04-15 · donald l. evans, secretary samuel...

39
Census 2000 Topic Report No. 14 TR-14 Issued February 2004 Puerto Rico U.S. Department of Commerce Economics and Statistics Administration U.S. CENSUS BUREAU Census 2000 Testing, Experimentation, and Evaluation Program

Upload: others

Post on 28-Jul-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

Census 2000 Topic Report No. 14TR-14

Issued February 2004

Puerto Rico

U.S.Department of CommerceEconomics and Statistics Administration

U.S. CENSUS BUREAU

Census 2000 Testing, Experimentation, and Evaluation Program

Page 2: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

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 Cassandra H.Thomas provided coordination support. Florence H.Abramson provided editorial review.

This report was prepared by Julie Buckley-Ess andIdabelle Hovland of the Decennial ManagementDivision. The following authors and project managersprepared Census 2000 experiments and evaluationsthat contributed to this report:

Decennial Statistical Studies Division:Sarah E. BradyNathan A. CarterJerry D. ImelKimball T. JonasTracey A. McNallyDarlene A. MoulRobin A. PenningtonMichael C. TenebaumErin Whitworth

Planning, Research and Evaluation DivisionSherri J. NorrisKaren L. OwensMegan C. Ruhnke

Population Division:Matthew Christenson

Independent Contractor:Susan Berkowitz, Westat

The authors would like to recognize the following indi-viduals for their assistance and support in the reviewof this report: Matthew Christenson, Louisa F.Miller, and Joel Sobel.

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

Page 3: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

U.S. Department of CommerceDonald L. Evans,

Secretary

Samuel W. Bodman,Deputy Secretary

Economics and Statistics AdministrationKathleen B. Cooper,

Under Secretary for Economic Affairs

U.S. CENSUS BUREAUCharles Louis Kincannon,

Director

Census 2000 Topic Report No. 14Census 2000 Testing, Experimentation,

and Evaluation Program

PUERTO RICO

TR-14

Issued February 2004

Page 4: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

Suggested Citation

Julie Buckley-Essand Idabelle HovlandCensus 2000 Testing,

Experimentation, and EvaluationProgram Topic Report No. 14, TR-14,

Puerto Rico, U. S. Census Bureau,

Washington, DC 20233ECONOMICS

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

Page 5: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

U.S. Census Bureau Puerto Rico iii

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

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

2. Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

2.1 Historical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

2.2 Questionnaire content . . . . . . . . . . . . . . . . . . . . . . . . . . .3

2.3 Data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

3. Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5

4. Results of Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7

4.1 Address list development activities . . . . . . . . . . . . . . . . . .7

4.2 Mail response and mail return rates . . . . . . . . . . . . . . . .10

4.3 Completeness and quality of data . . . . . . . . . . . . . . . . . .13

4.4 Field data collection . . . . . . . . . . . . . . . . . . . . . . . . . . . .16

4.5 Special places/group quarters . . . . . . . . . . . . . . . . . . . .18

5. Evaluation Authors’ Recommendations . . . . . . . . . . . . . . . . .21

6. Topic Report Authors’ Recommendations . . . . . . . . . . . . . . .23

7. Actions to Date . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25

8. Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29

Contents

Page 6: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 7: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

U.S. Census Bureau Puerto Rico 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

Page 8: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 9: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

The Puerto Rico Topic Reportlooked at 15 evaluations. FourPuerto Rico specific evaluationswere conducted: two involvedfocus groups, one an analysis ofcensus data files, and one an oper-ational analysis. Nonresponse andreactions to the Hispanic originand race questions were evaluatedusing focus groups. Frequencyand cross-tabulation files tables forboth Puerto Rico and the U.S. wereprepared from the census data files

on Hispanic origin and race.Puerto Rico enumeration was thetopic for the operational analysis.

While not specifically Puerto Ricoevaluations, evaluations in theAddress List Development categoryprovide a considerable amount ofdata on Puerto Rico. These evalua-tions covered the following opera-tions: address listing, Local Updateof Census Addresses 99 (LUCA 99),and update/leave.

Many evaluations included in thistopic report include Puerto Rico inthe stateside analysis and providefew, if any, breakouts of PuertoRico data. A number of evalua-tions that could have providedinformation on operations thattook place in Puerto Rico, did notprovide any analysis for PuertoRico. Examples include evalua-tions which looked at the successof the advertising and promotionprograms, enumeration of specialplaces and coverage issues.

U.S. Census Bureau Puerto Rico 1

1. Introduction

Page 10: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 11: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

U.S. Census Bureau Puerto Rico 3

2. Background

Census 2000 in Puerto Rico wasmarked with two historic “firsts:” theuse of the same questionnaire con-tent as stateside and the require-ment that respondents mail backtheir questionnaires. Both of thesesignificant events, one requested bythe Government of Puerto Rico dur-ing the content determinationprocess and the other a CensusBureau methodological decision,affected the results of many of theevaluations included in the PuertoRico Topic Report.

2.1 Historical

Spain ceded the island of Puerto Ricoto the United States in 1898. Priorto then, Spain had taken censuses atirregular intervals between 1765 and1887. The U.S. War Departmenttook a special census of Puerto Ricoin 1899. Puerto Rico, which becamea commonwealth in 1952, has beenincluded in every U.S. decennial cen-sus since 1910. Beginning in 1960,the census of population and hous-ing was conducted as a joint projectof the U.S. Census Bureau and theGovernment of Puerto Rico. ThePuerto Rico Planning Board (PRPB)has been identified by the Office ofthe Governor as the liaison agencyfor census activities on the island.The Census Bureau was responsiblefor the data collection, and the PRPBprovided input on content and dataneeds.

This partnership between the gov-ernment of Puerto Rico and theCensus Bureau is the result of anOctober 1958 special agreementconcerning the censuses in PuertoRico. The basic purposes of theagreement were to assure the effi-

cient operation of the census pro-gram, to provide the Commonwealthwith a large share of the responsibili-ty for planning the census, and toassure full consideration of itsunique statistical needs.1 Each cen-sus thereafter conformed to thebasic 1958 agreement with subse-quent amendments, includingCensus 2000. Governors of PuertoRico regularly have directed the PRPBto serve as the coordinating agencyfor the census operations.

2.2 Questionnaire content

From 1960 - 1990, the CensusBureau worked with the PRPB todevelop questionnaire content thatmet Puerto Rico’s unique needs. Forexample, in 1990 the Puerto Ricoquestionnaire had unique topicssuch as parents’ place of birth, voca-tional training, and condition ofhousing unit; but did not includestateside topics such as race,Hispanic origin, and home heatingfuel.

During the questionnaire contentdevelopment phase for Census2000, however, the Government ofPuerto Rico informed the CensusBureau in a letter to Dr. MarthaFarnsworth Riche, Census BureauDirector, from Norma Burgos, PRPBChairwoman, dated October 31,1997, that Puerto Rico was request-ing the same decennial questionnairecontent as stateside. The reasonsgiven for the same content requestincluded quicker processing and

release of Puerto Rico census dataand the inclusion of Puerto Rico instateside summary statistics as wellas comparability with stateside data.

2.3 Data collection

Prior to Census 2000, Puerto Ricohad always been enumerated usingthe list/enumerate (L/E) methodolo-gy. With L/E data collection, enumer-ators visit each housing unit to com-plete a census questionnaire, recordaddresses and update the L/E maps.

For Census 2000, a decision wasmade by the Census Bureau to usethe update/leave (U/L) methodologyfor the entire Island.2 In order toconduct an U/L operation, theCensus Bureau first had to conductan islandwide address listing (AL)operation. The AL operation wasconducted in Puerto Rico fromOctober, 1998 through January,1999.

During U/L, enumerators canvassassignment areas to deliver a censusquestionnaire to each housing unit,update the address listing pages andCensus Bureau maps. The householdthen completes and returns thequestionnaire by mail. The move toan U/L strategy responded tochanges taking place on the Island,including an increase in limitedaccess communities and a popula-tion that mirrored the United Statesin the prevalence of two incomefamilies. These two trends indicated

1 U.S. Census Bureau, 1990 Census ofPopulation and Housing, History, Part D,Chapter 13, Puerto Rico, Virgin Islands, andthe Pacific Island Territories, March 1996

2 Thompson, John H., 2000 DecennialCensus, Decision Memorandum No. 6,Recommendation that the Census Bureau Usethe Update/Leave Methodology for DataCollection During the Year 2000 Census ofPuerto Rico, signed by Robert W. Marx onAugust 12, 1996

Page 12: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

4 Puerto Rico U.S. Census Bureau

that it would be more difficult toimplement a L/E methodology in2000. The U/L methodology alsooffered other benefits, available

stateside for many decennial census-es: self-response resulting in higherdata quality and the development of

an address list that could be used infuture decennial operations.

Page 13: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

The purpose of this report is tosummarize the findings and recom-mendations from the evaluationstudies for those operations whichtook place in Puerto Rico. Not allevaluations provide information ordata specific to Puerto Rico.Sometimes Puerto Rico data areincluded within the stateside analy-sis of an operation. The followingis a list of the evaluations whichprovided Puerto Rico specific data:

Response Rates andBehavior Analysis

A.2.b Internet Data Collection

A.3 Be Counted Campaign forCensus 2000

A.8 Puerto Rico Focus Groupson Why Households DidNot Mail Back the Census2000 Questionnaire

Content/Data Quality

B.1.b Analysis of ItemNonresponse Rates for the100 Percent Housing andPopulation Items fromCensus 2000

B.12 Puerto Rico Census 2000Responses to the Race &Ethnicity Questions

B.13 Puerto Rico Focus Groupson the Census 2000 Raceand Ethnicity Questions

Special Places and GroupQuarters

E.5 Group QuartersEnumeration

Address List Development

F.2 The Address ListingOperation and Its Impacton the Master Address File

F.6 Evaluation of the LocalUpdate of CensusAddresses 99 (LUCA 99)

F.10 Evaluation of theUpdate/Leave Operation

Field Operations

H.2 Assessment of FieldVerification

H.5 Nonresponse Followup forCensus 2000

H.8 Operational Analysis ofEnumeration of Puerto Rico

H.9 Local Census Office Profilefor Census 2000

H.10 Date of Reference for Ageand Birth Date used byRespondents of Census2000

U.S. Census Bureau Puerto Rico 5

3. Scope

Page 14: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 15: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

4. Results of Analysis

4.1 Address listdevelopment activities

The research question posed was:How successful were the addresslist development activities? Thereare three relevant evaluations inthis category: F.2 The AddressListing Operation and Its Impact onthe Master Address File, F.6Evaluation of the Local Update ofCensus Addresses 99 (LUCA 99),and F.10 Evaluation of theUpdate/Leave Operation. All ofthese evaluations provide a signifi-cant amount of data that arePuerto Rico specific.

In 1990, all of Puerto Rico wasenumerated using the list/enumer-ate methodology. A decision wasmade to enumerate all of PuertoRico using the update/leavemethodology in Census 2000 -basically a rural procedure in theU.S. While it represented a stepforward, it was difficult to use arural listing procedure, whichincluded map spotting, in urbanareas.

In addition to methodologicalissues, there was concern aboutPuerto Rico’s unique addressingconventions and the use ofSpanish. Most notable is the fourline address where the urbaniza-tion name (neighborhood equiva-lent/connotation) is used to elimi-nate the tie between repeatedstreet names in different urbaniza-tions. In some instances, theurbanization, condominium, orcommunity/district name is usedin lieu of a street name.

Puerto Rico’s unique addressingnorms and systems do not allow

the Census Bureau to apply thesame business rules for automatedprocessing that are followed state-side. The address landscapeacross Puerto Rico is a mix ofstyles and standards. These situa-tions that complicate parsing andstandardization do occur stateside,but not to the degree and varietyfound in Puerto Rico.

In addition to the mix of styles andstandards related to Puerto Ricoaddresses, there were several cap-ture and processing problems.The problems stem from not suffi-ciently modifying the Census 2000procedures and instrumentsdesigned for stateside opera-tions to capture, process, transfer,and store address information.From field listing forms, to ques-tionnaire design, to data entryinstruments, to file exchange lay-outs, to the basic layout of theMaster Address File (MAF), thefields and field lengths providedwere not always adequate to han-dle Puerto Rico addresses and insome instances not consistentfrom one medium to the next.

All three authors acknowledge thatsome parts of their analysis forPuerto Rico are limited due to anaddress listing processing error ofthe keyed listing pages. This erroraffected the Puerto Rico addresslist and subsequent operationswhich used and/or built upon thislist. The keyed listing pages had a60 character address field thatcould contain a city-style addressor a location description. Thestateside files used a flag, “A/D,”set by the lister to indicatewhether it was “A” for a city-style

address or “D” for a locationdescription. In Puerto Rico, theaddress listing pages were inSpanish and the flag set by the lis-ter was “D/L.” The “D” was usedfor city-style addresses and stoodfor the Spanish word dirección.The “L” was used for a physicallocation description and stood forthe Spanish word localización.Another difference between thestateside and Puerto Rico listingpages was the addition of a fourthline for urbanization or condomini-um name.

When the Decennial Systems andContracts Management Office(DSCMO) processed the files forPuerto Rico, the “D” flag wasprocessed as a location descrip-tion, as it was in the U.S., and thePuerto Rico keyed data were runthrough the stateside standardizer.However, the Census Bureau didnot have a standardizer that couldaccommodate the Spanish, fourline addresses from Puerto Rico.As a result, all of the address infor-mation collected in Puerto Rico hadto be moved to the locationdescription field. The locationdescription field then had a flagindicating whether the address wasa city-style or non-city styleaddress. Processing was compli-cated by the fact that often the 60character field frequently con-tained a field listing entry thatincorrectly blended both city styleand location description together.Any lister error in setting the flagsor recording the information in theassigned fields, further complicat-ed this solution. While this solu-tion was less than perfect, it

U.S. Census Bureau Puerto Rico 7

Page 16: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

8 Puerto Rico U.S. Census Bureau

allowed the Census Bureau to usethe addresses that had been col-lected during address listing forsubsequent census operations.

F.2 The Address Listing andOperation and Its Impact on theMaster Address File

This evaluation provides sometables where information from theU.S. and Puerto Rico are combined,as well as some specific tables foreither the U.S. or Puerto Rico.

Since all of Puerto Rico was to beupdate/leave, the entire Islandneeded to be listed, about 1.4 mil-lion addresses. Of those addresseslisted, 99.5 percent were DecennialMaster Address File (DMAF) deliver-able and over 94 percent were inthe final census count.

The processing error described inSection 4.1 affected the results ofthis analysis, notably the absenceof any information on multi-unitand city-style addresses in PuertoRico. Both items require the use ofthe city-style address field on theMaster Address File (MAF) extractas the starting point and all ofPuerto Rico’s addresses were in thelocation description field. Thus inthis analysis, all of Puerto Ricohousing units were treated as sin-gle units and there were no city-style addresses in Puerto Rico.

F.6 Evaluation of the Local Updateof Census Addresses 99 (LUCA 99)

The LUCA 99 program invited localgovernments to participate in thereview of the addresses collectedduring the address listing opera-tion. In Puerto Rico, 50 of the 78muncipios signed up to participate.This was a higher participationrate, by eligible governmentalunits, than any of the four regionsof the U.S. The Boston RegionalCensus Center (RCC) consideredthe Puerto Rico LUCA 99 program asuccess because the high participa-

tion rate indicated an awareness ofthe importance of Census 2000 tothe local governments. About 20percent of the Puerto Rico partici-pants challenged blocks.

In the U.S., program participantswere provided map spotted maps,a census block housing unit sum-mary list, and a list of addressesfor their area to be used as a refer-ence. A LUCA 99 material modifi-cation was required for Puerto Ricobecause of map quality concerns.Due to map spot crowding, pro-gram participants received blockmaps with the map spots removed.

Using the materials provided bythe Census Bureau, municipiosidentified any block counts theydeemed inaccurate and the CensusBureau recanvassed those blocks.In Puerto Rico 35,563 addresseswere sent out for review.Enumerators could verify, delete,declare non-residential, correct oradd addresses. In Puerto Rico:33,029 addresses were verified;2,513 were deleted; 21 were deter-mined to be non-residential; andzero were corrected. In areas thatwere recanvassed, enumeratorsadded a total of 9,874 addresses.

The zero corrections requires fur-ther explanation. The LUCA 99field verification listings for PuertoRico were erroneously run usingthe stateside listing page formatinstead of the Puerto Rico format.This error was discovered whenthe enumerators were in the fieldwith the binders. The statesidelisting page did not display theadditional line for the urbanizationor condominium name and used“A/D” (address/description) ratherthan the Puerto Rico “D/L” (direc-ción/localización) for capturingaddresses or location descriptions.The complication was that enumer-ators made corrections to theurbanization/condominium name

anywhere on the listing page theyfound space. The correct PuertoRico Spanish blank add pages,however, were inserted into thebinders.

As a result, all added addressesand existing addresses with actioncodes of “does not exist” and “non-residential” could be keyed.Corrections to address listingswere not keyed since the correc-tions made on the statesideaddress listing pages for itemssuch as urbanization/condominiumname could have been placed any-where on the line or page. TheNational Processing Center (NPC)did not have a sufficient number ofbilingual clerks that could haveinterpreted these corrections.3

As in Evaluation F.2, this evaluationis limited by the processing errorthat occurred during address list-ing with respect to analysis bycity-style address or multi-unit status.

F.10 Evaluation of theUpdate/Leave Operation

All of Puerto Rico was Type ofEnumeration Area (TEA) 2 and enu-merated using update/leave (U/L).The pre-printed U/L listing pagefor Puerto Rico was in English andthe U/L add page was in Spanish.Both pages carried an additionalline for the urbanization or condo-minium name.

In Puerto Rico, a total of 1,471,225U/L actions were taken. This con-sisted of:

adds 111,787corrections 751,156deletes 122,815verifies 485,467

3 Hovland, Idabelle B., Memorandum forthe Record, Keying LUCA 99 FieldVerification Results for Puerto Rico, August24, 1999.

Page 17: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

The author notes that Puerto RicoU/L had a higher percentage ofadds, deletes and corrections thanstateside operations, but acknowl-edges that the address listing pro-cessing error may account for theadditional work that was required.

Of the 111,787 added addresses,nearly 90 percent of the adds perblock occurred in blocks withfewer than ten adds. The numbersof corrections and deletes perblock were more widely distrib-uted, especially corrections. Thewide distribution of correctionsmay be attributable to the initialprocessing error and that LUCA 99field verification corrections werenot keyed. Of the 111,787addresses added in the U/L opera-tion, 93,607 (83.7 percent) were inthe final count.

As with evaluations F.2 and F.6,this evaluation is limited by theprocessing error that occurred dur-ing address listing with respect toanalysis by city-style address ormulti-unit status. In addition, theDelivery Sequence File analysisthat the author provides for state-side, as an indicator of possiblefuture mail-out/mail-back areas,was not done for Puerto Ricobecause the addresses were notmatchable city-style addresses.

Independent Analysis of AddressList Development Activities inPuerto Rico

In 1996, a Puerto Rico 2000Working Group was formed withrepresentatives from the followingdivisions: Decennial Management(DMD), Population (POP), Field(FLD), Decennial Statistical Studies(DSSD), and Geography (GEO). (Atthat time DMD also encompassedthe functions of the currentDecennial Systems and ContractsManagement Office). In retrospect,a critical omission from the teamwas a representative from the

Technologies Management Officesince many of the address listingswere of the automated varietywhose templates could not bemodified in time to accommodateSpanish.

The Puerto Rico 2000 WorkingGroup was responsible for theoverall planning of census activi-ties. In addition, they providedbackground support to other divi-sional teams that were responsiblefor specific operations (e.g. GroupQuarters enumeration) that wouldtake place in Puerto Rico.

The Puerto Rico 2000 WorkingGroup supported the move fromthe 1990 List/Enumerate method-ology to Update/Leave. The 1990census data showed 71 percent ofthe population in Puerto Rico asurban. There had even been a1990 Puerto Rico Multi-unitCoverage Improvement OperationEvaluation that encompassed thefour San Juan area District Offices.That evaluation found 262 eligiblemulti-units with over 34,000 units.Given this information, the work-ing group even explored the possi-bility of having a Mailout/Mailbacktest site in Bayamón but concludedthat it would not be feasible with-out a 1990 census address list thatcould be used in conjunction withthe Puerto Rico United StatesPostal Service (USPS) DeliverySequence File that used four-lineaddresses. While the workinggroup recognized thatUpdate/Leave represented a stepforward along the enumerationcontinuum, it also recognized thedrawback of its being a basicallyrural procedure.

The working group used the state-side address listing page as thebase for developing an addresslisting page for Puerto Rico.Concern over processing issues,kept modifications to a minimum.

It was agreed that, at a minimum,the page would have to be inSpanish and a fourth line would berequired for urbanization or condo-minium name. The working groupused the Postal AddressingDirectory, produced by the USPSCaribbean District, as a guide indeveloping address collectionrequirements for Puerto Rico.Research in Puerto Rico revealedthat E-911 addresses were notestablished. The working groupagreed to remove that field fromthe Puerto Rico listing page andallow larger spaces for the existingfields.

The working group discussed pro-cessing concerns such as lengthsof fields, capturing diacriticals, andthe transfer of files between GEOand DMD. There was generalagreement that modifications forPuerto Rico would be necessaryand modifications were kept to aminimum. The working group wasaware of the changes that weremade to the address fields inPuerto Rico, and communicatedthem via memoranda to affecteddivisions (Puerto Rico 2000Memoranda Series, NO. 97-01;Memorandum to Distribution Listfrom Idabelle B. Hovland, January28, 1998). While the memorandawere widely distributed, the infor-mation was not communicated toeveryone within each division whoneeded to know. The most glaringresult of this lack of communica-tion between all affected parties isevidenced in the initial processingof the address listing.

From that point on, salvaging theaddress listings from Puerto Ricobecame the goal for each succes-sive census operation - some ofwhich introduced new problems.These unplanned for operationalchallenges, plus the use of a basi-cally rural procedure in a pre-dominantly urban environment,

U.S. Census Bureau Puerto Rico 9

Page 18: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

complicated enumeration andaffected the address list develop-ment evaluations. Loading alladdresses into the locationdescription field allowed us to pro-ceed with enumeration operations,but could not provide any evalua-tion statistics based on city styleor multi-unit status.

The compromised state of thePuerto Rico Master Address File(MAF) at the end of census opera-tions was the reason the CensusBureau entered into a contract withSeek Data to clean up the PuertoRico MAF. Seek Data took thePuerto Rico MAF records and stan-dardized and parsed them intoindividual address components andadded components as appropriate.Seek Data is currently workingwith GEO to develop a new datamodel for Puerto Rico MAFaddresses that will better enablethe Census Bureau to implementautomated address processingactivities for the 2010 census.

The need to improve communica-tion with, and involvement of, allaffected parties with interest in thePuerto Rico address list develop-ment activities is the basis for ourrecommendation to improve theprocess for including and inform-ing all parties when customizationis required. Among other things,we suggest an increase in the useof Joint Application Developmentsand the inclusion of in-house per-sonnel from all relevant subjectmatter and operational divisions inthe planning, specification, devel-opment, and testing of materialsfor Puerto Rico.

4.2 Mail response and mailreturn rates

The research questions posed are:What were the mail response andmail return rates? and How didthey differ by long and short form?

In Puerto Rico, the response rate asof April 18, 2000 was 48.4 per-cent. Stateside, the response rateas of April 18, 2000 was 59.3 per-cent. The Evaluation H.8 authorexplains that the mail responserate is defined as the number ofmail returns received prior to thecut date for the NRFU universedivided by the total number ofhousing units in mailback areasthat were eligible for NRFU. It is ameasure that represents the per-centage of addresses eligible forNRFU that returned questionnairesprior to the designation of theNRFU universe. In Puerto Rico, forshort form questionnaires, theresponse rate as of April 18, 2000was 50.5 percent; for long formquestionnaires the response ratewas 37.6 percent. Stateside, as ofApril 18, 2000, for short formquestionnaires the response ratewas 61.9 percent; for long formquestionnaires it was 51.9 percent.

In Puerto Rico, the mail return rateas of April 18, 2000 was 55.0 per-cent. Stateside, the return rate asof April 18, 2000 was 69.6 per-cent. The Evaluation H.8 authorexplains that the mail return rate isdefined as the number of mailreturns received prior to the cutdate for the NRFU universe dividedby the total number of occupiedhousing units in mailback areasthat were on the DMAF prior toNRFU. In Puerto Rico, as of April18, 2000, for short form question-naires the return rate was 57.2percent and for long form ques-tionnaires it was 43.7 percent.Stateside, as of April 18, 2000, thereturn rate for short form question-naires was 72.3 percent; for longform questionnaires it was 61.9 percent.

Two evaluations in this categoryaddress additional ways people inPuerto Rico could answer the cen-sus - by responding via the

Internet or through the use of a BeCounted questionnaire (A.2.b.Internet Data Collection and A.3 BeCounted Campaign for Census2000). The third evaluation in thiscategory is devoted solely to theissue of why Puerto Rico house-holds chose not to mail back thequestionnaire (A.8 Puerto RicoFocus Groups on Why HouseholdsDid Not Mail Back the Census 2000Questionnaire). Evaluation H.8Operational Analysis ofEnumeration in Puerto Rico, dis-cussed more fully in the field datacollection category, is the sourcefor the mail response and mailreturn rates for Puerto Rico.

A.2.b. Internet Data Collection

This evaluation provides opera-tional summaries on the use of theInternet as a response mode.Puerto Rico is included in the over-all data analysis. The author doesnote that there was insufficienttime to create a Spanish-languageversion of the internet form forPuerto Rico. “Thus, respondents inPuerto Rico and other Spanishspeakers could respond on theInternet in English only.”

Only short form mailback house-holds with an ID number from thedelivered questionnaire were eligi-ble for the internet responseoption. Puerto Rico had 1,094,593potentially eligible households and107 households chose to respondto the Puerto Rico English shortform questionnaire via theInternet. Table P19, Age byLanguage Spoken at Home byAbility to Speak English for thePopulation 5 Years and Over, fromthe Census 2000 Summary File 3detailed tables for Puerto Rico indi-cate that only 30 percent of thepopulation 18 to 64 years oldspeak only English or speakEnglish “very well.”

10 Puerto Rico U.S. Census Bureau

Page 19: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

A.3 Be Counted Campaign forCensus 2000

A Be Counted campaign was imple-mented in Puerto Rico for Census2000 with the same goals as state-side:

• count persons who did notreceive a census questionnaire

• count persons who believedthey were not included on anyother census form

• encourage participation of per-sons who are traditionallyundercounted in the census

• provide a means for personswith no usual residence to becounted

In Puerto Rico, Be Counted forms(BCFs) were available in Spanishand English. In addition, BeCounted display boxes were avail-able with Be Counted messages inHaitian-Creole.

The author includes Puerto Rico inthe overall evaluation of the BeCounted campaign. The authorreports that 60,000 English BCFsand 360,000 Spanish BCFs wereprinted for Puerto Rico and areincluded in the numbers of totalEnglish and Spanish BCFs printed.Table T., Cost and ExpenditureCategory, shows that printing theEnglish and Spanish BCFs forPuerto Rico cost $127,181.

A.8 Puerto Rico Focus Groups onWhy Households Did Not Mail Backthe Census 2000 Questionnaire

The author draws conclusions forreasons for mail nonresponse inPuerto Rico based on focus groupsconducted in nine sites acrossPuerto Rico with a combined totalof 41 participants (28 women and13 men). The nine sites were:Mayagüez, Ponce, Cayey, Loíza,Bayamón, Humacao, Añasco, OldSan Juan and Santurce. Focus

group participants were householdheads who had not returned thequestionnaire by mail or householdheads who had later filled out theirquestionnaire with the help of anenumerator.

The author divides her reasons formail non-response into four broadcategories:

• Motivational and process-related

• Practical and logistical

• Cultural and political

• Related to questionnaire contentand design

Motivational and process-relatedreasons. A lack of clarity aboutthe purpose of the census isincluded in this category. Thefocus group participants remem-bered the advertising campaign asemphasizing mailing back thequestionnaire and how the numberof responses was important indetermining federal aid. They feltthat the advertising campaign didnot convey a broad sense of pur-pose and was reminiscent of apolitical campaign.

Also included in this broad catego-ry was focus group participants’confusion over the distribution ofquestionnaires and the role of enu-merators in Census 2000 versus1990. Participants may have seenan U/L enumerator updatingaddress lists and dropping offquestionnaires, an Accuracy andCoverage Evaluation (A.C.E.) enu-merator with A.C.E. questionnaires,and/or a Nonresponse Followup(NRFU) enumerator with anEnumerator Friendly Questionnaire(EFQ).

Practical and logistical. This broadcategory includes focus group par-ticipants reporting a lack of time tocomplete the questionnaire. Anexample given was that of a work-ing mother returning home from

work and having to balance com-peting demands on her time - mealpreparation, child care, and com-pleting a census questionnaire.

Another example included in thiscategory is the difficulty somefocus group participants had, espe-cially in rural areas, in returningtheir form by mail.

Cultural and political. Some focusgroup participants voiced theirfears about the perceived lack ofconfidentiality with questionnaireresponses and the attendant con-sequences if the information fellinto the wrong hands - e.g. the taxauthority, welfare officials. Thefocus group participants voiced ageneral mistrust of governmentand politicians.

Nonresponse focus group partici-pants “in all communities andacross all demographic and eco-nomic groups expressed a strong,unqualified support for in-persondata collection as the means ofgathering the necessary informa-tion.” The 1990 L/E data collectionmethod offered this approach. InL/E, an enumerator came by eachhousehold to pick up a completedAdvance Census Report or fill out aquestionnaire with the respondentand took the completed question-naire with him/her. Anecdoteswhich elaborated on the perceivedcultural preference for a personalapproach were supplied by the A.8author.

Related to questionnaire contentand design. As discussed inSection 2.2, Questionnaire Content,Census 2000 was the first timethat stateside questionnaire con-tent was used in Puerto Rico. TheCensus Bureau used the statesidequestionnaire content in PuertoRico at the request of the PuertoRico Planning Board, the CensusBureau’s officially designated liaison.

U.S. Census Bureau Puerto Rico 11

Page 20: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

12 Puerto Rico U.S. Census Bureau

As a result of this request, severalnew topics were introduced (suchas race and Hispanic origin) whileother topics from previous census-es were dropped. Many nonre-sponse focus group participantsvoiced their objections to the useof the race and Hispanic originquestions in Puerto Rico. Theyviewed the questions as divisiveand insensitive to the ‘mixed’ reali-ties of Puerto Rico.

Various nonresponse focus groupparticipants also voiced objectionsto questions on income, maritalstatus and household composition.However, these were not newquestions for Census 2000 inPuerto Rico.

Finally, nonresponse focus groupparticipants voiced concerns aboutthe overall length and complexityof the long form questionnaire.Some nonresponse focus groupparticipants also felt that the lan-guage used on the questionnairewas hard to understand. The A.8author suggests that in some casesthe basic issue was literacy andreported that respondent strategiesincluded waiting for an enumeratoror neighbor to help them completethe form correctly.

Independent Analysis of MailResponse and Mail Return Rates

According to Evaluation H.8,Operational Analysis ofEnumeration in Puerto Rico, themail response rate in Puerto Ricoas of April 18, 2000 (NonresponseFollowup [NRFU] cut date) was48.4 percent and the mail returnrate was 55.0 percent. Cost modelassumptions for the Puerto RicoNRFU operation assumed a 50 per-cent response rate. We used thisrate based on the fact that PuertoRico households had no priorexperience with mailback censusesor surveys. (At that time PuertoRico was not included in some of

the larger U.S. surveys conductedby the Census Bureau, such as theAmerican Community, that requirerespondents to mail back a ques-tionnaire.) Stateside respondentshave been responding by mailsince 1970. We would expect theresponse rate in Puerto Rico toincrease in the 2010 Census, giventhe mailback experience in Census2000, and a clear media messagein 2010 to continue mailing backthe questionnaires.

Be Counted forms (BCFs) wereanother way respondents couldanswer the census. Past experi-ence has shown that the mostused language in Puerto Rico isSpanish, followed by English. Thusthe BCFs and display boxes inPuerto Rico were available in bothSpanish and English. A June 28,2000 check-in record of PuertoRico BCFs shows that there were13,300 Spanish and 1,069 EnglishBCFs checked in to the DataCapture Center in Jeffersonville, IN.

In an effort to count what wasthought to be an emerging Haitianimmigrant population, Haitian-Creole Language Assistance Guideswere available for Puerto Rico aswell as a Be Counted display boxwith Haitian-Creole stickers.Results from Census 2000 showthat 328 residents claimed Haiti astheir place of birth. However, wehave no data on the use of theHaitian-Creole language assistanceguides in Puerto Rico.

The internet was another responseoption. In Puerto Rico, this optionwas available in English to shortform mailback household respon-dents with an ID number from thedelivered U/L questionnaire. InPuerto Rico, only 107 respondentschose the Internet as a responseoption. If the Internet responseoption is expanded to includeSpanish, we would expect that the

number of Internet respondents inPuerto Rico to increase for 2010.We base this recommendation onthe increase in the number ofInternet users and services avail-able to them in Spanish.

A December 2002 study conductedin Puerto Rico by Research &Research in collaboration withNobox Marketing Group for thePuerto Rico chapter of the InternetSociety, points to 970,000 Internetusers. This is up from an estimat-ed 551,000 Internet users in2000.4 In addition, several of themost popular Internet ServiceProviders (ISPs) like AOL and ATT,as well as local ISPs, provide serv-ice to the island and have Spanishspeaking support.

There are many websites inSpanish, including many of thepopular stateside websites such asYahoo and Amazon, which havemirror pages in Spanish. On islandthere are many websites, such asthe El Nuevo Día newspaper, whichprovide content solely in Spanish.Newer versions of web browserssuch as Internet Explorer andNetscape Navigator can be set upin Spanish to view all browser but-tons and menus in Spanish.

Some potential respondents chosenot to complete a questionnaire.Many of the reasons are familiar -competing demands for time, thequestionnaire is too long, thequestions are offensive - and areapplicable across cultures. Asvoiced by the focus group partici-pants in Evaluation A.8, there doesappear to be at least some culturalpreference in Puerto Rico for themore personal approach thatList/Enumerate offered and focusgroup participants were willing to

4 Caribbean Business, Internet Users inPuerto Rico Near One Million, Volume 31, No.15, April 10, 2003.

Page 21: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

wait for the NRFU enumerator tocome around to complete the inter-view. Whether this would be truefor a larger segment of the popula-tion is unclear.

4.3 Completeness andquality of data

The research questions posed are:How complete are census data for Puerto Rico and What is theirquality as measured by item non-response and rate of proxyresponses? Two of the three evalu-ations in this category look atPuerto Rico responses to theHispanic origin and race questions:B.12 Puerto Rico Census 2000Responses to the Race andEthnicity Questions and B.13Puerto Rico Focus Groups on theCensus 2000 Race and EthnicityQuestions. Since the CensusBureau has never asked theHispanic origin question in PuertoRico, and race was last collected inthe 1950 census, there has beenconsiderable interest in analyzingthe results of these questions.

One evaluation, B.1.b Analysis ofItem Nonresponse Rates for the100 Percent Housing andPopulation Items from Census2000, looks at overall item nonre-sponse.

Past research, cited by the B.12and B.13 authors, suggest thatrespondents of Hispanic origintend not to differentiate betweenrace and Hispanic origin. This ideais more fully explored with theEvaluation B.13 focus group partic-ipants who corroborate the find-ings from this earlier research.Results from the focus groups eval-uation discuss how participantsfelt that the existing race cate-gories did not match their PuertoRican identity. Focus group partici-pants discussed whether race wasmeasured by color, features, origin,etc. and that members of the same

family could be classified different-ly, depending on how race wasdefined.

The B.13 evaluation author reportsthat past research has also foundthat Hispanics may feel pressure touse the “White” category eventhough they do not consider thisresponse accurate nor feel theywould be considered “White” intheir country of origin. Findingsfrom the response tally and thefocus group participants give fur-ther credence to this theory.

While Evaluation B.13 indicatesthere was controversy over theHispanic origin and race questionsin Puerto Rico, Evaluation B.12demonstrates that Puerto Ricorespondents of Hispanic originseemed to have less trouble withthe questions than their statesidecounterparts of Hispanic origin, asmeasured by lower item nonre-sponse rates.

B.1.b. Analysis of ItemNonresponse Rates for the 100Percent Housing and PopulationItems from Census 2000

Item nonresponse occurs when noanswer is provided to an item onthe questionnaire. Examining itemnonresponse provides informationabout data quality. The item non-response evaluation universe is thehousing unit population.

While the analysis in this reportdoes not include Puerto Rico, sta-tistics for Puerto Rico are providedin Appendix G. We looked at someof the major stateside findings tosee if they held true for PuertoRico.

In the U.S., item nonresponse forthe Census 2000 100 percentitems ranged from 1.13 percent forthe sex item to 4.14 percent forthe tenure item. In Puerto Rico,the overall item nonresponse ratefor the Census 2000 100 percent

items ranged from 1.01 percent forthe Hispanic origin item to 6.15percent for the tenure item. Thelow nonresponse rate to theHispanic origin question is not sur-prising given that Puerto Rican wasa checkbox choice.

In the U.S., item nonresponse wasgenerally higher for enumeratorreturns than for self-response andhigher for long forms than shortforms. For both short and longforms, item nonresponse for allquestions except Hispanic originwas higher for enumerator returns.The tenure and age items had thelargest absolute differences in itemnonresponse rates betweenresponse modes. The absolute dif-ferences were 6.22 and 6.91 per-centage points, respectively. InPuerto Rico, item nonresponse wasgenerally higher for enumeratorreturns than for self-response andhigher for long forms than shortforms. For both short and longforms, item nonresponse for allquestions except race was higherfor enumerator returns. The raceand tenure items had the largestabsolute differences in item nonre-sponse rates between responsemodes. The absolute differenceswere 3.56 and 3.15 percentagepoints, respectively.

It is not surprising that theHispanic origin (stateside) and race(Puerto Rico) swapped places, withregards to the one item where theenumerator nonresponse rate waslower than self response. PuertoRican was an obvious choice forPuerto Rico, but Puerto Ricans arenot accustomed to being askedtheir race on official Puerto Ricogovernment documents (see subse-quent discussions of EvaluationsB.12 and B.13). Enumerators weretrained on the differences betweenrace and Hispanic origin.

U.S. Census Bureau Puerto Rico 13

Page 22: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

In the U.S., tenure had a relativelyhigher overall nonresponse ratecompared to other items.Specifically, long form tenure non-response rates are higher perhapsdue to the placement of tenureafter all long form items for Person1. In Puerto Rico, tenure had a rel-atively higher overall nonresponserate (6.15 percent) compared toother items. In Puerto Rico, thetenure nonresponse rate for theshort form was 5.35 percent and10.26 percent for the long formtenure. As in the U.S., the place-ment of the tenure question afterall of the Person 1 population char-acteristics may be the reason forthis.

In the U.S., age had differentiallyhigher nonresponse for enumera-tors; which may be due to the useof proxy in personal interviews. InPuerto Rico, tenure had differential-ly higher nonresponse for enumer-ators.

In the U.S., for the Hispanic originitem, nonresponse was higher forself-response. The lowest nonre-sponse for Hispanic origin is asso-ciated with enumerator shortforms. It appears that self respon-dents do not understand the differ-ence between race and Hispanicorigin, so they answer the racequestion and leave Hispanic originquestion blank. In Puerto Rico, forthe race item, nonresponse washigher for self-response. The low-est nonresponse for race is associ-ated with enumerator short forms.It appears that self respondents inPuerto Rico do not understand thedifference between race andHispanic origin, so they answer theHispanic origin question and leavethe race question blank.

B.12 Puerto Rico Census 2000Responses to the Race andEthnicity Questions

Evaluation B.12 for Puerto Ricolooks at the Hispanic origin andrace responses, item non-responserates for the Hispanic origin andrace questions, and any differencesin Hispanic origin and raceresponses by response mode(respondent or enumerator sup-plied). Many results are comparedwith those of respondents ofHispanic origin in the 50 statesand DC.

Hispanic origin responses. Theanalysis shows that the residentsof Puerto Rico identified them-selves as overwhelmingly ofHispanic origin (98.8 percent).About 95.1 percent of the popula-tion identified themselves as ofPuerto Rican origin; 1.5 percent asof Dominican origin and less than1.0 percent as of either Cuban orMexican origin. Of these groups,only Dominican required a write-inresponse as there was no checkbox for Dominican.

Race responses. The analysisshows that more than 95 percentof people who identified them-selves as of Hispanic origin inPuerto Rico also identified them-selves as of one race. The greatmajority of those who identifiedthemselves as Hispanic also report-ed themselves as White alone (80.7percent), while 7.9 percent identi-fied themselves as Black or AfricanAmerican alone, and just under 7percent reported themselves as ofSome Other Race alone. This dif-fers from stateside Hispanicswhere only 47.9 percent identifythemselves as White alone, 2.0percent as Black or AfricanAmerican alone, and 42.2 percentidentify themselves as Some OtherRace alone.

More than 90 percent of the PuertoRico respondents who identifiedthemselves as of Hispanic originused the race check boxes, while

9.2 percent provided write-inresponses. Of the write-inresponses, 82.8 percent were inthe space provided to specify theSome Other Race response.Almost two-thirds of theseresponses provided Hispanic originidentifications and not their race,while slightly less than one-thirdprovided a color response such as“moreno” or “brown.”

Item non-response for Hispanic origin. Item non-response, includ-ing invalid response, can be anindicator of how well a questionhas been received and handled bythe respondent. This can be meas-ured by the number of responsesthat need to be generated by theedit and allocation process duringCensus processing.

The percent of responses to theHispanic origin question resultingfrom the edit and allocation byHispanic origin was 4.6 percent inPuerto Rico as compared to thestateside rate of 9.2 percent.

Item non-response for race. Theauthor looks at the percent ofresponses to the race question thatwere the result of the edit and allo-cation process, by Hispanic origin,and finds that only 6.2 percent ofthe Puerto Rico responses by thoseof Hispanic origin were edited andallocated, while 18.2 percent ofthe stateside responses were edit-ed and allocated.

Differences between respondentand enumerator completed ques-tionnaires. The author explainsthat another indicator of how wella question is received andanswered by respondents is ifthere are differences in responsesfrom questionnaires filled out byrespondents as compared withenumerator completed question-naires.

14 Puerto Rico U.S. Census Bureau

Page 23: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

The response to the Hispanic ori-gin question shows very little dif-ference between respondent com-pleted and enumerator completedquestionnaires. More variation isshown for the race question.During enumerator conductedinterviews, those of Hispanic origintended to give a White alone orBlack or African American aloneresponse less frequently, while giv-ing a Some Other Race alone orTwo or More Races response morefrequently.

B.13 Puerto Rico Focus Groups onthe Census 2000 Race andEthnicity Questions

The author draws her conclusionsfrom 86 participants at focusgroup discussions that were con-ducted in 12 sites across theIsland. The 12 sites representedboth urban and rural areas(Arecibo, Isabela, Rincón,Mayagüez, Lares, Yauco, Ponce,Cidra, Bayamón, San Juan, RíoGrande and Yabucoa).

As a context for the discussion onHispanic origin and race in PuertoRico, the author makes two impor-tant observations:

• Puerto Ricans are not accus-tomed to being asked for theirrace in official Puerto Rico gov-ernment documents.

• Issues of race, nationality, andidentity are politically charged.

Hispanic origin. Since Puerto Ricanwas one of the check box options,most participants had no difficultywith this question and it was notsubject to the same scrutiny as therace question. Some focus groupparticipants did report that thephrase “Spanish/Hispanic/Latino”origin was confusing or moreappropriate to a U.S. context.

Race. Having answered theHispanic origin question, many

focus group participants thoughtthe race question, which followedthe Hispanic origin question, wasredundant. They did not see them-selves accurately represented inthe race question answer cate-gories and they would have pre-ferred to see the “Puerto Rican” cat-egory as a pre-printed category onthe race question.

Focus group participants repeated-ly observed that the race questionwas inappropriate to the PuertoRico context and inherently divi-sive. This varied surprisingly littleby place or social class. The par-ticipants pointed out that in schoolthey learn that the Puerto Rican“race” is a distinctive mixture ofSpanish, Indian, and African. This,plus the fact that race is notrequested on official Puerto Ricogovernment documents, did notgive participants a lot of experi-ence defining race using CensusBureau categories. (AlthoughPuerto Ricans who have had expe-rience filling out other forms fromthe U.S. government may be famil-iar with race and ethnicity ques-tions.) Participants therefore tend-ed to think of race in terms ofnationalities or in terms of color.Participants then reported, howev-er, that the problem with usingcolor or phenotypic characteristicswas that family members could bedifferent races. While the focusgroup participants reported diffi-culty with the race question, theresults of the B.12 evaluation showthat only 6.2 percent of theresponses to the race question byrespondents of Hispanic origin inPuerto Rico required edit and allo-cation.

The B.12 evaluation also showsthat 80.7 percent of the respon-dents in Puerto Rico who reportedHispanic origin also reported them-selves in the White alone category.Focus group participants provide

some insights as to why this maybe, including: White was the bestanswer among inappropriate alter-natives, Black was only for thosewho were pure Black, and thatthere is still a stigma to beingidentified as Black in Puerto Rico.

Given that the focus group partici-pants did not see the individualpre-printed race categories asapplicable to their situation, thequestion remains as to why theydid not check multiple boxes toindicate their mixed heritage. TheB.12 evaluation reports that only4.1 percent of the Hispanic originrespondents in Puerto Rico report-ed two or more races. The B.13author observes: “Quite a few ofthe focus group participants hadnot realized they could havechecked off multiple racial cate-gories for each person in theirhousehold.” Other participants,who were aware of the option,thought of the Puerto Rican race asa unitary or continuous conceptwhich would not be captured bychecking multiple boxes. Manyfocus group participants wouldhave preferred a single responseoption that acknowledged theirmixed ancestry and suggested“Creole,” “trigueño,” or “Caribbean.”

Independent Analysis ofCompleteness and Quality of Datain Puerto Rico

There has been considerable inter-est within Puerto Rico and theCensus Bureau as to how the raceand Hispanic origin questions werereceived by respondents and theresulting data. Race was last col-lected by enumerators in the 1950Census of Puerto Rico. Hispanicorigin has never been asked inPuerto Rico. These questions wereasked in Puerto Rico for Census2000 as the result of theGovernment of Puerto Rico’s

U.S. Census Bureau Puerto Rico 15

Page 24: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

request for the same questionnairecontent as the United States.

Because two of the three evalua-tions in this category focus exclu-sively on the race and Hispanic ori-gin questions, we limit ouranalysis to these questions.

All three evaluations suggest thatin Puerto Rico there is morerespondent difficulty with the racequestion than the Hispanic originquestion. Intuitively, that seemsreasonable since Puerto Rican is acheck box response for theHispanic origin question. AsEvaluation B.1.b. demonstrates, therace item was the only item wherenon-response was higher for self-response than enumerator returns.On the other hand, Evaluation B.12finds that for respondents ofHispanic origin only 6.2 percent ofthe Puerto Rico responses to therace question were edited and allocated, while 18.2 percent ofthe stateside responses to the race question were edited and allocated.

Contextually, as the EvaluationB.13 author points out, PuertoRicans are not accustomed tobeing asked for their race in offi-cial Puerto Rico government docu-ments and issues of race, nationali-ty, and identity are politicallycharged. A panel of social scien-tists and professors convened bythe San Juan City Magazine in1995 to analyze the variations onracism in Puerto Rico, would agreewith that assertion. The directorof the Institute of CaribbeanStudies at the University of PuertoRico, Aaron Gamaliel Ramos,observed: “To talk about race andracism is to agitate the still watersof national solidarity.” This viewcoincides with that observed bythe Evaluation B.13 author thatfocus group participants saw“Puerto Rican” as a unitary or con-

tinuous concept that cannot becaptured by checking multiple racecategories.

In addition to interest in why thequestions on race and Hispanic ori-gin were being asked in PuertoRico, considerable interest wasgenerated in the resulting data.Census results showed that thegreat majority of those who identi-fied themselves as Hispanic inPuerto Rico also reported them-selves as White alone (80.7 per-cent), while 7.9 percent identifiedthemselves as Black or AfricanAmerican alone, and just under 7percent reported themselves as ofSome Other Race alone. A head-line in the newspaper El Nuevo Día,dated April 16, 2001, read“Rechazo boricua a su origennegro” (Puerto Ricans reject theirblack origins). Persons inter-viewed for the article indicatedthat these Census results do notreflect the racial reality of PuertoRico.

Although the race question andresulting data caused controversywithin Puerto Rico, two factsremain:

• Race and Hispanic originappeared on the questionnaireas a direct result of the requestfrom the Government of PuertoRico for the same questionnairecontent as stateside; and

• Puerto Rico respondents ofHispanic origin seemed to haveless trouble with the questionsthan their stateside counterpartsof Hispanic origin, as measuredby lower item nonresponserates.

4.4 Field data collection

The research question posed is:How well did we perform the fielddata collection activities? Topicsreviewed were field verification for

Non-ID housing units, nonresponse

followup (NRFU), an operational

analysis of Puerto Rico enumera-

tion, a statistical profile of Local

Census Offices, and the date of ref-

erence used by respondents when

reporting age and date of birth.

Three of the field data collection

evaluations (field verification,

NRFU, and date of reference) pres-

ent Puerto Rico data in combina-

tion with U.S. data. Two of these

three evaluations (field verification

and date of reference) each provide

one table with Puerto Rico specific

information. The third evaluation

(NRFU) provides six tables with

Puerto Rico specific data.

Evaluation H.9, Local Census Office

Profile for Census 2000, provides

hundreds of Puerto Rico specific

tables. Evaluation H.8, Operational

Analysis of Enumeration of Puerto

Rico, primarily focuses on the

update/leave (U/L) operation.

H.2 Assessment of Field

Verification

This evaluation focuses on the Be

Counted/Telephone Questionnaire

Assistance Field Verification opera-

tion. During this operation, enu-

merators visited the location of

units without an assigned Census

identification number to verify

their existence before they were

included in Census 2000. Puerto

Rico is included in the overall

analysis, with one table providing

Puerto Rico specific data. The

table shows that there were 690

assignment areas (AAs) in Puerto

Rico with field verification cases.

(For Census 2000, Puerto Rico had

6,225 AAs.) These 690 AAs repre-

sent 0.16 percent of the total num-

ber of AAs within the U.S. and

Puerto Rico that had field verifica-

tion cases.

16 Puerto Rico U.S. Census Bureau

Page 25: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

H.5 Nonresponse Followup forCensus 2000

The objective of the NRFU opera-tion was to obtain a completedquestionnaire from all householdsin mailback areas that did notrespond by mail, through theInternet or via a TelephoneQuestionnaire Assistance operator.Since all of Puerto Rico was enu-merated by L/E in 1990, NRFU wasa new operation for Puerto Rico inCensus 2000. From the point ofview of the Boston RegionalCensus Center (RCC), NRFU inPuerto Rico was very successfuland all NRFU operations at theLocal Census Offices (LCOs) werefinished ahead of schedule.

In this evaluation, Puerto Rico sta-tistics are combined with those ofthe U.S. Puerto Rico specific data,however, are available from sixtables in the appendices whichprovide state level data. The datain the appendices show that whilePuerto Rico had 1,357,301 housingunits potentially eligible for fol-lowup, the NRFU workload uni-verse in Puerto Rico was 699,540housing units.

The following NRFU operationalchallenges in U/L areas, noted bythe author, also occurred in PuertoRico: surname in the incorrect fieldon the listing page, address regis-ters started with the address of thefirst nonresponding housing uniton each block, and U/L adds werenot processed in time to updatethe NRFU registers. Because theU/L adds were not keyed in time,enumerators may have added themissing units again during theNRFU operation and thus inflatedthe percentage of added address-es. In Puerto Rico, NRFU wasresponsible for adding 28,793addresses and deleting 78,680addresses.

H.8 Operational Analysis ofEnumeration of Puerto Rico

The author looks at the way theaddress list was compiled, charac-teristics of the U/L operation andoperational problems whichoccurred in the NRFU andCoverage Improvement Followup(CIFU) operations. An analysis ofthe debriefing questionnaires com-pleted by the Assistant Managerfor Field Operations (AMFO) foundthree main problem areas: latearrival of training materials, maps,and merging long-form and short-form questionnaires in the LocalCensus Offices (LCOs).

Training materials for Puerto Ricofield operations were adapted foraddressing conventions and geo-graphic and questionnaire differ-ences. In addition, all materials forenumerators, crew leaders andfield operations supervisors wereto be translated into Spanish.While materials for Puerto Ricowere adapted and translated on aflow basis, they could not be final-ized until the stateside materialswere completed. This meant thatPuerto Rico materials were alwaysavailable after stateside materialswere available. Occasionally, therewas not enough time in the sched-ule to allow for the translation offield operations supervisor materi-als. Any problems with assem-bling and shipping materials exac-erbated an already tight schedule.

The AMFOs felt that numerous mapupdates were required, especiallyin rural areas. While U/L is prima-rily a rural procedure in the U.S., inPuerto Rico it was used islandwide.Because of scale issues, map spot-ting in urban areas was difficult.Furthermore, the process of pass-ing on map updates to subsequentfield operations was consideredproblematic.

The LCOs prepared the U/L enu-merator assignments. One of thedifficult tasks, reported in theAMFO debriefing questionnaires,was merging the long form andshort form questionnaires so thatthey appeared in the order foundon the U/L listing pages.

There were 1,471,225 addresses inPuerto Rico, including addressesfrom the address listing operationand adds from the U/L operation,U/L adds accounted for 7.6 percentof the Puerto Rico workload. Ofthe 111,787 U/L adds in PuertoRico, 83.7 percent (93,607) wereincluded in the final counts.

Deletes accounted for almost 8.4percent of the Puerto Rico U/Lworkload (as compared to 5.2 per-cent of the stateside workload).Some of Puerto Rico’s higher deleterate may have been due to achange in Puerto Rico’s addresslisting procedures which allowedfor the inclusion of vacant dam-aged living quarters in the addresslisting registers. This change wasmade to take into account theeffects of Hurricane Georges,which occurred in late September1998, just before address listingwas to begin. At that time it wasestimated that Hurricane Georgeshad damaged or demolished over100,000 housing units in PuertoRico. Address listing procedureswere modified in anticipation thatmany of these damaged unitswould be repaired or reconstructedon the same site in upcomingmonths.5

Corrections accounted for 38.45percent of the stateside U/L work-load and 51.06 percent of thePuerto Rico workload. PuertoRico’s higher correction rate is not

U.S. Census Bureau Puerto Rico 17

5 Monaghan, Brian, Memorandum forArthur Dukakis, Inclusion of VacantDamaged Living Quarters in Address ListingRegisters for Puerto Rico, October 16, 1998.

Page 26: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

18 Puerto Rico U.S. Census Bureau

surprising given the address listingprocessing error and the decisionnot to key LUCA field verificationcorrections.

The AMFOs reported that the NRFUworkload for Puerto Rico was larg-er than it needed to be. Not all ofthe Puerto Rico questionnaires sentin were processed before the NRFUcut date. The maps also werecited by the AMFOs as NRFU andCIFU problems. There were prob-lems with crowded map spots andU/L map updates not being passedon to the NRFU and CIFU opera-tions. The AMFOs did report suc-cess in retaining qualified stafffrom earlier operations to work onNRFU and CIFU.

H.9 Local Census Office Profile forCensus 2000

This profile covers 16 general top-ics, including counts by housingunit types, householder demo-graphics, response rates and work-loads for various field operations.In all, there are over 1400 statis-tics produced for each of the LCOsin the U.S. and Puerto Rico. Whenshowing the totals for the variousstatistics, the tables give the U.S.totals with and without PuertoRico.

H.10 Date of Reference for Ageand Birth Date used byRespondents of Census 2000

This evaluation analyzes how wellrespondents used Census Day astheir date of reference whenanswering the age and date ofbirth questions. For Census 2000,the ‘average’ date of reference wasApril 20, 2000.

Appendix D, State Return Rates asof December 31, 2000 and StateDate of Reference, shows thatPuerto Rico’s date of reference wasMay 3, 2000 and its return rate asof December 31, 2000 was 63.9percent. The author explains that

a state’s return rate seems to becorrelated with the date of refer-ence for that state. As the rateincreases, the date of reference forthe state is closer to April 1, 2000.Since Puerto Rico’s return rate wasless than that of any U.S. state, itis not surprising that it is the onlystate (equivalent) with a referencedate in May.

Independent Analysis of Field DataCollection Activities

The state of the address list andmail response rate for Puerto Ricoare the two overarching issuesaffecting the evaluations in thiscategory. Field Verification (FV),Update/Leave (U/L) and Non-response Followup (NRFU) were theoperations most affected by thedifferent addressing conventionsused in Puerto Rico and/or thecondition of the address list attheir respective stages. PuertoRico’s lower return rate appears tobe associated with a May date ofreference for respondents(Evaluation H.10) and affectedNRFU workloads.

The workload for Puerto Rico fornon-MAF ID processing was 40,330addresses. The FV workload con-sisted of those non-MAF ID ques-tionnaires that could not bematched to the existing addresslists. Because the existing GEOmatching and geocoding softwarecould not process the uniqueaddressing conventions found inPuerto Rico, GEO entered into acontract with Seek Data to matchand geocode non-MAF ID question-naires from Puerto Rico. This oper-ation marked the beginning of anongoing relationship with SeekData to clean up and standardizeaddress components for the PuertoRico MAF.

The U/L and NRFU operations inPuerto Rico followed the statesideschedule and the same basic enu-

meration procedures. Thus state-side successes and challenges alsoaffected Puerto Rico. The inclusionof Puerto Rico in the automatedCost and Progress reports, as wellas the LCO statistics reported inEvaluation H.9, are examples ofsuccesses. Stateside operationalchallenges affecting Puerto Ricoare evident in some of the sameNRFU listing page errors.

Unique to Puerto Rico was the useof a basically rural procedure toenumerate the entire Island. In1990, Puerto Rico included fourMetropolitan Statistical Areas andone Consolidated MetropolitanStatistical Area. Using a rural pro-cedure in this environment createsits own set of problems - mostnotably map spotting.

The state of the Puerto Rico MAF,after all census data collectionoperations were completed, is thereason DMD has funded GEO’s con-tract with Seek Data to clean upthe existing MAF. Seek Data is cur-rently working with GEO to devel-op a new data model for PuertoRico MAF addresses that will betterenable the Census Bureau to imple-ment automated address process-ing activities. This will allow us toexplore additional enumerationmethodologies for Puerto Rico forthe 2010 census and better sup-port Puerto Rico CommunitySurvey activities.

4.5 Special places/groupquarters

The special place/group quartersevaluation that provides PuertoRico specific data is E.5, GroupQuarters Enumeration. One tablein the appendices provides datausing a state-by-state breakout.The Appendix C table providescounts of the group quarters (GQ)population by GQ category. PuertoRico had a total of 46,774 personsin GQs, distributed as follows:

Page 27: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

17,283 in correctional institutions;853 in juvenile institutions; 7,311in nursing homes; 3,356 in hospi-tals; 2174 in colleges/universities;1,199 in military facilities; 6,419 ingroup homes; and 8,179 in serv-ice- based and other GQs.

Independent Analysis of SpecialPlaces/Group QuartersEnumeration

The E.5 evaluation provides PuertoRico specific data on the popula-tion associated with each type ofgroup quarters. However, PuertoRico also participated in the follow-ing operations: Special Place LUCA,Special Place Advance Visits,Military, Maritime, T-Night, MobileFood Vans, Service BasedEnumeration, Shelter, Soup Kitchenand Targeted Non-ShelteredOutdoor Locations. Operationally

these operations followed thestateside procedures and sched-ules, thus conclusions drawn fromstateside evaluations can probablybe applied to Puerto Rico as well.(Refer to the Special Place/GroupQuarters Enumeration Topic Reportby Florence H. Abramson, issuedSeptember 17, 2003.)

In Puerto Rico, the long and shortform Individual Census Reports(ICRs) and Individual CensusQuestionnaires (ICQs) were avail-able in both Spanish and English.Between ICRs and ICQs, there wereeight forms for Puerto Rico. Inaddition there were Puerto RicoMilitary Census Reports (MCRs) andShipboard Census Reports (SCRs)available only in English. In PuertoRico, the Facility Questionnaire wasonly available in Spanish and was a

paper version as there was notenough time to translate the paperversion into a computer assistedtelephone interview (CATI) instru-ment.

Numerous forms and the latearrival of training materials werethe issues that most affectedPuerto Rico Special Place/GroupQuarters enumeration. There wasinsufficient lead time to adapt andtranslate all of the SpecialPlace/Group Quarters trainingmaterials. Various imperfect solu-tions were tried, including usingstateside materials or only adaptedmaterials. Sometimes, when therewas not enough time to translatethe materials, the positions werefilled with bilingual personnel inorder to meet the enumerationschedule.

U.S. Census Bureau Puerto Rico 19

Page 28: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 29: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

The following list of recommenda-tions represents a summary ofthose provided by the authors ofthe evaluations. The specificstudy(ies) and/or assessment isprovided in parentheses.

• Include Puerto Rico in any cogni-tive testing or efforts to fieldtest different versions of therace question.The findings from the race ques-tion suggest that the residentsof Puerto Rico interpreted thisquestion differently than thoseof Hispanic origin and eventhose of Puerto Rican origin inthe 50 states and DC.Consequently, it will be impor-tant to include Puerto Rico infuture census tests. (B.12, B.13)

• Investigate further the use ofthe Some Other Race Category.Investigate further the use ofthe Some Other Race categoryof the race question to assurethat the range of responsesexcludes those that better fitthe Hispanic origin question.(B.12)

The Census Bureau also is con-ducting research on eliminatingthe Some Other Race categoryfrom the race question.

• Improve the census message. Provide more extensive publiceducation to the Puerto Ricanpopulation on the larger missionof the census, the rationale forasking questions about race andethnicity, and the intended usesof the data. (B.13)

Develop advertising appeals forcensus participation more

attuned to the Puerto Rican pop-ulation. (A.8)

• Conduct a survey on respon-dents’ views on data collection,census materials, and questions.Consider using the results of theB.13 study, along with theresults of the A.8 study, to cre-ate a survey to be administeredto a probability sample of resi-dents on the Island. The surveycould seek the respondents’views of different approaches todata collection and their reac-tions to any new materialsdeveloped, including any allow-able changes made to the struc-ture and content of the ques-tions on race and Hispanicorigin. (A.8, B.13)

• Promote the use of the Internetform.While not specific to PuertoRico, consider how to promul-gate the Internet form as anoption and convince the publicthat there is sufficient datasecurity. (A.2.b) For Puerto Rico,provide a Spanish Internetresponse option in 2010 torespond to the increasing num-bers of Internet users on theisland.

• Create a new Puerto Ricoaddress list.At the point of questionnairedelivery, the initial errors in theaddress list for Puerto Rico havebeen compounded from theeffects of several operations. Itmight be advisable for futureefforts in Puerto Rico to create anew address list rather than do

a dependent listing startingfrom the current listing. (F.10)

• Consider mailout/mailback for2010.Build on the experience fromCensus 2000 and the PuertoRico Community Survey andconsider mailout/mailback datacollection in 2010 for at leastthe urban parts of the island.(H.8)

• Improve census maps and pro-vide more map training.Include reference points andchange the scale of maps tomake them easier for enumera-tors to use. Provide moreextensive map training for enu-merators and look into the feasi-bility of using Global PositioningSystem technology to improvethe accuracy of enumerator mapspotting. Attempt to automatechanges and corrections to cen-sus maps from early censusfield operations so that they canbe utilized in subsequent fieldoperations. (H.8)

• Ensure that field materialsarrive on time.Build enough time into thestateside schedule for fieldmaterials preparation so thattraining materials for PuertoRico can be adapted and trans-lated and arrive in the LCOswith enough lead time for train-er preparation. (H.8)

U.S. Census Bureau Puerto Rico 21

5. Evaluation Authors’ Recommendations

Page 30: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 31: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

U.S. Census Bureau Puerto Rico 23

The “standardize/customize” conflictaffected decennial operations forPuerto Rico. Standardization offerscost savings, ease of processing,quicker release of data products, andcomparability with stateside data.Customization allows us to preparematerials in Spanish, offer question-naire content that can meet theIsland’s unique data needs, and uti-lize different addressing require-ments and geography. Puerto Ricocustomization requirements, howev-er, compete for limited decennialresources, including staff, time andmoney.

Census 2000 was the first time thatthe Census Bureau and Puerto Ricoreally experienced the benefits ofstandardization. Benefits includedtimely release of data products withinthe existing stateside schedule, theinclusion of Puerto Rico in U.S. sum-mary statistics, and American FactFinder availability. While some evalu-ations suggest the use of the state-side questionnaire in Puerto Rico hadits drawbacks, the use of the samequestionnaire content simplified theprocessing and tabulation of data.These benefits were available to theCensus Bureau and Puerto Ricobecause the Government of PuertoRico requested the same question-naire content. However, we cannotassume that Puerto Rico will requeststateside questionnaire content in2010.

Some customization for Puerto Rico,beyond the scope of questionnairecontent, will always be required. Theuse of Spanish is the most obvious.In addition to questionnaires andpublicity for the general populationin Puerto Rico, materials for a pre-dominantly Spanish speaking laborforce also must be available.

Several evaluations addressed theproblems that resulted from cus-tomizing addresses for Puerto Rico toinclude condominium or urbanizationname, and using the Spanish initialsfor address or location description.This customization was needed toensure deliverability in Puerto Rico,both by Census 2000 enumeratorsand for future use by the USPS formail delivery of the Puerto RicoCommunity Survey. As the evalua-tions suggest, the Census Bureau’sstandard systems and processeswere not sufficiently prepared to han-dle these customization require-ments.

The long term solution for 2010 maybe to consider a total redesign of thelisting page to allow the CensusBureau to successfully parse PuertoRico addresses into their componentparts (e.g., urbanization, house num-ber, direction, street name, streettype). Parsing will allow us to devel-op the intelligence/capability, overtime, to build the various supplemen-tal tables needed for Puerto Ricostandardization - like alternativename tables.

Given the “standardize/customize”challenges, we recommend:

Improve the process for includingand informing all parties when cus-tomization is required

• Increase the use of JointApplications Development.

• Include in-house personnel fromall relevant subject matter andoperational divisions in the plan-ning, specification, development,and testing of materials for PuertoRico.

• Ensure that all team members,such as subject matter expertsand processing staffs, stay activelyinvolved in the continued adapta-tion of requirements and the reso-lution of technical issues through-out the development andimplementation efforts.

• Schedule development activities sothat ample time is allowed for theadaptation and translation ofmaterials for Puerto Rico.

Standardize where appropriate

• Standardize the process for ensur-ing that Puerto Rico customizationrequirements are included in allprocesses and systems.

• Standardize processes, such asenumeration methodology, whenthe Census Bureau determinesthey can be applied to Puerto Ricowithout a resulting loss in dataquality.

6. Topic Report Authors’ Recommendations

Page 32: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

24 Puerto Rico U.S. Census Bureau

• Collect Puerto Rico addressesusing a parsed format developedspecifically for Puerto Rico.

• Develop a Puerto Rico addressstandardizer.

Support early planning

• Explore ways to improve the mail-back response rate for PuertoRico.

• Redesign the Puerto Rico addresslisting page. Conduct a limitedaddress collection field test inPuerto Rico and of the address list-ing processing systems prior to2010 decennial census implemen-tation.

• Dedicate more Census Bureau staffresources to Puerto Rico enumera-tion issues, so that internal exper-

tise can be developed across divi-sions.

Develop Puerto Rico specific evalua-tions

• Build Puerto Rico specific datarequirements and analysis into theprogram design of future censusevaluations.

Page 33: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

Positive first steps have been taken

following Census 2000 towards

improving the Puerto Rico Master

Address File (MAF). The

Geography Division (GEO) has had

a contract since 2000 with Seek

Data, Inc., a private sector compa-

ny experienced in working with

Puerto Rico addresses and geogra-

phy. Seek Data is currently alter-

ing many existing MAF records by

parsing and standardizing the indi-

vidual address components, or by

adding these components to some

MAF records where no addresscomponents exist. Seek Data alsois working with GEO to develop anew data model for Puerto RicoMAF addresses that will betterenable the Census Bureau to imple-ment automated address process-ing activities in the future. This iskey to the Census Bureau’s abilityto maintain the Puerto Rico MAFthrough automated means.

Seek Data also has begun attempt-ing to provide new, city-stylePuerto Rico addresses to the

Census Bureau for potential use in

future MAF activities. As part of

this, they are examining the utility

of the United States Postal Service

file of addresses, which is a key

component of MAF maintenance

stateside. This process has only

recently begun, and it is too early

to predict what results will be

achieved.6

U.S. Census Bureau Puerto Rico 25

7. Actions to Date

6 Sobel, Joel, E-mail communication toMr. G. Maldonado, September 25, 2002.

Page 34: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 35: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

The decision to move fromlist/enumerate to update/leave forCensus 2000 in Puerto Rico repre-sented the first step, in a series ofsteps that still need to be made, ofmoving Puerto Rico tomailout/mailback in future census-es. Beginning with address listing,and continuing with subsequentcensus operations, an address listwas built for Puerto Rico. ThePuerto Rico address list requiredcustomization for deliverability,including Spanish and a fourth linefor development or condominiumname. As the evaluations indicate,

however, the address list wasflawed from the beginning and itseffects on census operations anddata were noted. The GEO is nowworking with Seek Data to repairand enhance this list for futurecensus operations and the PuertoRico Community Survey (known asthe American Community Survey inthe U.S.).

The standardization of question-naire content has resulted in thetimely release of Puerto Rico dataproducts and comparabilitybetween stateside and Puerto Ricodata. However, as the author of

the focus group evaluations pointsout, standardization often comeswith a price - as evidenced by thefocus group respondents’ com-plaints with the Hispanic originand race questions.

The standardization/customizationdebate for Puerto Rico will contin-ue, with choices to be made forenumeration methodologies, con-tent, geography, data products andso on. Standardizing the processfor including customizationrequirements will help us at everystage along the way.

U.S. Census Bureau Puerto Rico 27

8. Summary

Page 36: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 37: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

References

American FactFinder, Summary File 3Detailed Tables for Puerto Rico,P19., Age by Language Spoken atHome by Ability to Speak English forthe Population 5 Years and Over.

Berkowitz, Dr. Susan, Puerto RicoFocus Groups on Why HouseholdsDid Not Mail Back the Census 2000Questionnaires, Census 2000Evaluation A.8, July 2001.

Berkowitz, Dr. Susan, Puerto RicoFocus Groups on the Census 2000Race and Ethnicity Questions,Census 2000 Evaluation B.13, July2001.

Bryant, Bettie J., Puerto Rico 2000Memoranda Series, MemorandumNo. 97-01, Memorandum toDistribution List, Transmission ofAddress Listing RequirementsOverview for Puerto Rico, September10, 1997.

Burgos, Norma, Chairwoman of thePuerto Rico Planning Board, letter toDr. Martha Farnsworth Riche,Director, Bureau of the Census,October 31, 1997.

Carter, Nathan, Be CountedCampaign for Census 2000, Census2000 Evaluation A.3, September2002.

Carter, Nathan and Sarah Brady, Dateof Reference for Age and Birth Dateused by Respondents of Census2000, Census 2000 Evaluation H.10,November 2002.

Censo de Población y Viviendas:1950, Puerto Rico, Forma P93.

Christenson, Matthew, Ph.D, PuertoRico Census 2000 Responses to theRace and Ethnicity Questions,Census 2000 Evaluation B.12, July2003.

El Nuevo Día Interactivo, Rechazoboricua a su origen negro, 16 deabril de 2001.

Farnsworth Riche, Dr. Martha,Director, Bureau of the Census, letterto Planner Norma Burgos,Chairwoman of the Puerto RicoPlanning Board, January 13, 1998.

Hovland, Idabelle B., Memorandumfor the Record, Keying LUCA 99 FieldVerification Results for Puerto Rico,August 24, 1999.

Hovland, Idabelle B., Note toDistribution List, Puerto RicoAddress Listing Page - Final, January28, 1998.

Imel, Jerry D., Local Census OfficeProfile for Census 2000, Final DraftReport, Census 2000 Evaluation H.9,September 2003.

Jonas, Kimball, Group QuartersEnumeration, Census 2000Evaluation E.5, Revision 1, August2003.

McNally, Tracy, Operational Analysisof Enumeration of Puerto Rico,Census 2000 Evaluation H.8, May2003.

Merced, Hector X., E-mail communi-cation to Carl C. Sanders, March 30,1999.

Monaghan, Brian, Memorandum forArthur Dukakis, Inclusion of VacantDamaged Living Quarters inAddress Listing Registers for PuertoRico, October 16, 1998.

Moul, Darlene A., NonresponseFollowup for Census 2000, Census2000 Evaluation H.5, July 2002.

Norris, Sherri, Analysis of ItemNonresponse Rates for the 100Percent Housing and PopulationItems from Census 2000, Census2000 Evaluation B.1.b, September2003.

Owens, Karen L., Evaluation of theLocal Update of Census Addresses99 (LUCA 99), Census 2000Evaluation F.6, May 2002.

Pennington, Robin A., Evaluation ofthe Update/Leave Operation, Census2000 Evaluation F.10, June 2003.

Puerto Rico and Island Areas Branch,Internal branch records, Puerto RicoMail Back and Internet Forms CheckIn, May 18, 2000.

Ruhnke, Megan C., The AddressListing Operation and Its Impact onthe Master Address File, Census2000 Evaluation F.2, January 2002.

San Juan City Magazine, So Are WeRacists????? A Conspiracy of Silence:Racism in Puerto Rico, Dec/Jan1996.

Sobel, Joel, Memorandum forLourdes Flaim, Address ListingOperations for Puerto Rico, October25, 1996.

Sobel, Joel, E-mail communication toIdabelle B. Hovland, May 27, 1999.

Sobel, Joel, E-mail communication toMr. G. Maldonado, September 25,2002.

Tenebaum, Michael, Assessment ofField Verification, Census 2000Evaluation H.2, July 2001.

Thompson, John H., 2000 DecennialCensus, Decision Memorandum No.6, Recommendation that the CensusBureau Use the Update/LeaveMethodology for Data CollectionDuring the Year 2000 Census ofPuerto Rico, signed by Robert W.Marx on August 12, 1996.

U.S. Census Bureau, Census 2000Final Response Rates, Final Update,September 7, 2000.

U.S. Census Bureau, 1990 Census ofPopulation and Housing, History,Part D, Chapter 13, Puerto Rico,Virgin Islands, and the Pacific IslandTerritories, March 1996.

Whitworth, Erin, Internet DataCollection, Census 2000 EvaluationA.2.b, August 2002.

U.S. Census Bureau Puerto Rico 29

Page 38: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

This page intentionally left blank.

Page 39: Issued February 2004 Puerto Rico - Census.gov · 2004-04-15 · Donald L. Evans, Secretary Samuel W. Bodman, Deputy Secretary Economics and Statistics Administration Kathleen B. Cooper,

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

Census 2000 Testing, Experimentation, and Evaluation Program