Data Collection In Census 2000

1. Introduction1With the goal to obtain a completed questionnaire for every housingunit in Census 2000, the UnitedStates Census Bureau used threebasic data collection methods andother special strategies to ensuredelivery of questionnaires to everyhousing unit. The basic methodsincluded “door-to- door” canvassing, a variation of which has beendone since the first census in1790, the use of the postal serviceto both deliver questionnaires andreceive questionnaire responses(mailout/mailback), which was initiated in the 1970 census, and thepersonal delivery of census questionnaires to the respondents withinstructions for them to mail thecompleted questionnaires back tothe Census Bureau (update/leave).This report, Data Collection inCensus 2000, synthesizes resultsfrom evaluations and assessmentsthat pertain to these data collection methods. This includes, invarying degrees, operations suchas Nonresponse Followup,Update/Leave, Urban Update/Leave, Update/Enumerate,List/Enumerate, CoverageImprovement Followup, and Field1Sources for this section include: (a)Census 2000 Operational Plan; (b) 200 Yearsof Census Taking: Population and HousingQuestions, 1790 – 1990; (c) TwentyCensuses–Population and Housing Questions;(d) Measuring America: The DecennialCensuses From 1790–2000; (e) Two HundredYears and Counting: The 1990 Census; (f)1990 Census of Population and HousingHistory (Parts A and D); (g) 1980 Census ofPopulation and Housing–History; (h) 1970Census of Population and Housing–Procedural History; (i) Procedural History ofthe 1960 Censuses of Population andHousing; (j) The 1950 Censuses–How TheyWere Taken, Procedural Studies of the 1950Censuses; and (k) Procedural History of the1940 Census of Population and Housing.U.S. Census BureauVerification. The report highlightsthe major challenges and successes of data collection operations,cites results from the evaluationsand assessments, and includes recommendations for planning anddesigning future data collectionoperations.BackgroundCensus-taking efforts were initiated shortly after our first President,George Washington, was inaugurated in 1790. In accordance withArticle I, section 2, of the UnitedStates Constitution, the first enumeration of the inhabitants of theUnited States began on the firstMonday of August in that sameyear.From 1790 to 1960, the basicmethod of collecting census information from the populationremained the same. It is referredto as “door-to-door” enumeration.Throughout this period, however,the census schedules, field procedures, training, and questionnaireswere continually improved fromthe previous experience. Changesor improvements related to datacollection are highlighted below.Up to 1840, the household, ratherthan the individual, remained theunit of enumeration in the population census. The U.S. marshalsresponsible for collecting dataentered only the names of the“household heads” on the censusschedules.From 1850 through 1870, the actthat governed the taking of theSeventh, Eighth, and NinthDecennial Censuses initiatedchanges in the census data collection process. The marshals wererequired, for the first time, to subdivide their districts into “knowncivil divisions” such as counties,townships, or wards and they werealso responsible for checking thecompleted census work to ensurethat the returns of their assistantswere properly completed.Three important changes were initiated for the 1880 census.Specially appointed agents (expertsassigned to collect technical data)along with supervisors and enumerators replaced marshals andtheir assistants. Enumerators wereforbidden to disclose census information. In prior censuses, censusschedules were posted publicly.Third, enumerators were givendetailed maps to follow so thatthey could account for every streetor road, and not stray beyond theirassigned boundaries.The 1890 census utilized, for thefirst time in history, a separateschedule for each family. The1900 census featured the first U.S.censuses conducted outside of thecontinental states and territories.The 1910 census had severalnotable features. Most importantlyfrom a data collection perspective,this was the first time thatprospective census employeeswere required to take open competitive examinations that wereadministered throughout the country. Previously, starting in 1880,appointees had been given noncompetitive tests prior to workingas census enumerators.Data Collection in Census 2000 1

There were minor changes in thescope and the data collectionmethods of the 1920 and 1930censuses. However, a profoundchange was introduced with thecensus of 1940. Sampling in the1940 census allowed for the addition of several questions for justfive percent of the households enumerated without unduly increasingthe overall burden on respondentsor on the data processing requirements. It was also the first toinclude a census of housing whichobtained a variety of facts on thegeneral condition of the Nation’shousing inventory.The 1950 population and housingcensus was conducted followingthe conventional method of doorto-door enumeration. However, acensus test conducted in Octoberof 1948 indicated that self-enumeration appeared feasible for use inthe census of agriculture in 1950.Prior to the actual decennial census in April 1950, an experimentwas conducted in six districtoffices and indicated that a general20 percent sampling pattern wouldbe feasible during the census.This sampling pattern was instituted during the 1950 census.As in all previous censuses, the1960 census still relied on door-todoor enumeration. However, itwas the first time that the UnitedStates Postal Service (USPS) wasused extensively to deliver censusself-administered questionnaires.Prior to the door-to-door enumerator canvassing operation, the USPSdelivered a questionnaire containing the 100 percent questions toevery occupied housing unit.Householders were asked to complete the questionnaire and hold ituntil the census enumerator cameto pick it up. This was regarded asthe first stage in the 1960 census.The second stage pertained to thesample questions, which were on a2 Data Collection in Census 2000separate questionnaire. In theurban areas of the country (whichcontained about 80 percent of thepopulation), the enumerators carried the sample questionnaireswith them while canvassing andleft one at every fourth householdasking the occupants to completethe sample questionnaire and mailit back to the Census Bureau.2The 1970 census introduced thefirst data collection method thatdid not require a 100 percent doorto-door canvassing. As in 1960,the USPS delivered self-administered questionnaires to households, but in 1970 the householdrespondent was instructed to mailreturn their responses back to theCensus Bureau. This method isreferred to as mailout/mailback.Approximately 60 percent of thepopulation (essentially in largemetropolitan areas) received andwere asked to return their censusquestionnaires via the mail. Inthese areas, enumerators contacted only those households that didnot return questionnaires or thathad given incomplete answers tothe questions (a nonresponse followup operation). For the remaining 40 percent of the population,predominately located in ruralareas, the mail carriers delivered aquestionnaire to the householdsand the householders were askedto complete and hold them forpickup by a census enumerator.The enumerators were responsiblefor obtaining missing or incom2The first stage of the 1960 processwas designed to concentrate primarily oncoverage, with the goal of providingimproved counts of people and housingunits. During the second stage of the 1960enumeration, the census enumerators concentrated primarily on collecting acceptablesample information on the various subjectscovered by the censuses. The separation ofthe enumeration into two stages was intended to simplify the job of the enumeratorsand it was hoped that if the enumerators foreach stage had fewer tasks to perform, theywould master them better.plete information. Three hundredand ninety-three district officeswere established in the 50 statesand the District of Columbia. Inaddition, six temporary districtoffices and one area office wereestablished for the census ofPuerto Rico. In 181 of the districtoffices, the census was taken inthe conventional (traditional) manner; that is, an enumerator visitedeach house to collect the information. In 167 of the district offices,the census was taken by thedecentralized mail method. Thedecentralized mail methodinvolved listing, by enumerators,the street addresses of every housing unit and the addressing, byclerks, of mailing pieces (questionnaire packages) to be mailed toevery housing unit. In the remaining 45 district offices, where commercial mailing lists were availablefor purchase by the Census Bureau,the centralized (mailout/mailback)method was used to collect thecensus information.The 1980 census basically involvedthe same methods used in 1970.The mailout/mailback method wasused in areas of the country containing 95.5 percent of the population and the conventional method(going door to door) was used forthe remainder of the country. TheUSPS delivered addressed censusquestionnaires to over 80 millionhousing units at the end of March,1980. In addition, by Census Day,all of the mail district offices hadtelephone lines installed for thepurpose of helping respondentscomplete their questionnaires. Alldistrict offices provided the sameservice for respondents whoappeared in person through theuse of walk-in assistance centers.The Census Bureau conducted the1990 Census using the three basiccensus methodologies: “mailout/mailback, enumerator delivery/U.S. Census Bureau

mailback, and mailout with a doorto-door canvass,” (U.S. Departmentof Commerce 1993). Ninety-fivepercent of the population wascounted by mail census procedures. Notably, the 1990 Censuswas the first national census thatused computers at the local districtoffice level to check-in, monitor,and check-out census questionnaires. Field data collection operations were structured, monitored,and evaluated through the use ofcomputer programs available atthe local district office level, the 13regional census centers, and thenational Census BureauHeadquarters. Applicant, employee, and payroll data were alsomaintained on

Randall J. Neugebauer of the Planning, Research and Evaluation Division; Violeta Vazquez of the Decennial Management Division; and Frank A. Vitrano (formerly) of the Planning, Research and Evaluation Division. The Census 2000 TXE Program was coordinated by the Planning, Research and Evaluation Division: Ruth Ann