Limitations of the Data

Home , Census, Data collection, Sample (statistics)

Appendix III Limitations of the Data

Introduction.—The data presented in this unit selected can be assigned a number, Statistical Abstract came from many greater than zero and less than or equal to sources. The sources include not only Fed- one, representing its likelihood or probability eral statistical bureaus and other organiza- of selection. tions that collect and issue statistics as their principal activity, but also governmental ad- For large-scale sample surveys, the prob- ministrative and regulatory agencies, private ability sample of units is often selected as a research bodies, trade associations, insur- multistage sample. The first stage of a mul- ance companies, health associations, and tistage sample is the selection of a probabil- private organizations such as the National ity sample of large groups of population Education Association and philanthropic members, referred to as primary sampling foundations. Consequently, the data vary units (PSU’s). For example, in a national multistage household sample, PSU’s are considerably as to reference periods, defini- often counties or groups of counties. The tions of terms and, for ongoing series, the second stage of a multistage sample is the number and frequency of time periods for selection, within each PSU selected at the which data are available. first stage, of smaller groups of population The statistics presented were obtained and units, referred to as secondary sampling units. In subsequent stages of selection, tabulated by various means. Some statistics smaller and smaller nested groups are cho- are based on complete enumerations or sen until the ultimate sample of population censuses while others are based on units is obtained. To qualify a multistage samples. Some information is extracted sample as a probability sample, all stages from records kept for administrative or regu- of sampling must be carried out using prob- latory purposes (school enrollment, hospital ability sampling methods. records, securities registration, financial accounts, social security records, income tax Prior to selection at each stage of a multi- returns, etc.), while other information is ob- stage (or a single stage) sample, a list of tained explicitly for statistical purposes the sampling units or sampling frame for through interviews or by mail. The estima- that stage must be obtained. For example, tion procedures used vary from highly so- for the first stage of selection of a national household sample, a list of the counties and phisticated scientific techniques, to crude county groups that form the PSU’s must be ‘‘informed guesses.’’ obtained. For the final stage of selection, Each set of data relates to a group of indi- lists of households, and sometimes persons viduals or units of interest referred to as the within the households, have to be compiled in the field. For surveys of economic entities target universe or target population, or sim- and for the economic census, the Census ply as the universe or population. Prior to Bureau generally uses a frame constructed data collection the target universe should from the Bureau’s Standard Statistical Es- be clearly defined. For example, if data are tablishment List (SSEL). The SSEL contains to be collected for the universe of house- all establishments with payroll in the United holds in the United States, it is necessary to States including small single establishment define a ‘‘household.’’ The target universe firms as well as large multi-establishment may not be completely tractable. Cost and firms. other considerations may restrict data collection to a survey universe based on some Wherever the quantities in a table refer to available list, such list may be inaccurate an entire universe, but are constructed from and out of date. This list is called a survey data collected in a sample survey, the table frame or sampling frame. quantities are referred to as sample estimates. In constructing a sample estimate, The data in many tables are based on data an attempt is made to come as close as is obtained for all population units, a census, feasible to the corresponding universe or on data obtained for only a portion, or quantity that would be obtained from a com- sample, of the population units. When the plete census of the universe. Estimates data presented are based on a sample, the based on a sample will, however, generally sample is usually a scientifically selected differ from the hypothetical census figures. probability sample. This is a sample se- Two classifications of errors are associated lected from a list or sampling frame in such with estimates based on sample surveys: a way that every possible sample has a (1) sampling error—the error arising from known chance of selection and usually each the use of a sample, rather than a census, Appendix III 971 to estimate population quantities and (2) a standard error of 10 units. An approxi- nonsampling error—those errors arising mately 90 percent confidence interval (plus from nonsampling sources. As discussed or minus 1.6 standard errors) is from 184 below, the magnitude of the sampling error to 216. for an estimate can usually be estimated from the sample data. However, the magni- All surveys and censuses are subject to nonsampling errors. Nonsampling errors are tude of the nonsampling error for an esti- of two kinds—random and nonrandom. mate can rarely be estimated. Conse- Random nonsampling errors arise because quently, actual error in an estimate exceeds of the varying interpretation of questions (by the error that can be estimated. respondents or interviewers) and varying The particular sample used in a survey is actions of coders, keyers, and other proces- sors. Some randomness is also introduced only one of a large number of possible when respondents must estimate values. samples of the same size which could have These same errors usually have a nonran- been selected using the same sampling dom component. Nonrandom nonsampling procedure. Estimates derived from the dif- errors result from total nonresponse (no us- ferent samples would, in general, differ from able data obtained for a sampled unit), par- each other. The standard error (SE) is a tial or item nonresponse (only a portion of a measure of the variation among the esti- response may be usable), inability or unwill- mates derived from all possible samples. ingness on the part of respondents to pro- The standard error is the most commonly vide correct information, difficulty interpret- used measure of the sampling error of an ing questions, mistakes in recording or estimate. Valid estimates of the standard keying data, errors of collection or process- errors of survey estimates can usually be ing, and coverage problems (overcoverage calculated from the data collected in a prob- and undercoverage of the target universe). ability sample. For convenience, the stand- Random nonresponse errors usually, but ard error is sometimes expressed as a per- not always, result in an understatement of cent of the estimate and is called the sampling errors and thus an overstatement of the precision of survey estimates. Esti- relative standard error or coefficient of mating the magnitude of nonsampling errors variation (CV). For example, an estimate of would require special experiments or ac- 200 units with an estimated standard error cess to independent data and, conse- of 10 units has an estimated CV of 5 per- quently, the magnitudes are seldom avail- cent. able. A sample estimate and an estimate of its Nearly all types of nonsampling errors that standard error or CV can be used to con- affect surveys also occur in complete cen- struct interval estimates that have a pre- suses. Since surveys can be conducted on scribed confidence that the interval includes a smaller scale than censuses, nonsampling the average of the estimates derived from errors can presumably be controlled more all possible samples with a known probabil- tightly. Relatively more funds and effort can ity. To illustrate, if all possible samples were perhaps be expended toward eliciting re- selected under essentially the same general sponses, detecting and correcting response conditions, and using the same sample de- error, and reducing processing errors. As a sign, and if an estimate and its estimated result, survey results can sometimes be standard error were calculated from each more accurate than census results. sample, then: 1) Approximately 68 percent of the intervals from one standard error be- To compensate for suspected nonrandom low the estimate to one standard error errors, adjustments of the sample estimates above the estimate would include the aver- are often made. For example, adjustments age estimate derived from all possible are frequently made for nonresponse, both samples; 2) approximately 90 percent of the total and partial. Adjustments made for ei- intervals from 1.6 standard errors below the ther type of nonresponse are often referred estimate to 1.6 standard errors above the to as imputations. Imputation for total non- estimate would include the average esti- response is usually made by substituting for mate derived from all possible samples; and the questionnaire responses of the nonre- 3) approximately 95 percent of the intervals spondents the ‘‘average’’ questionnaire re- from two standard errors below the estimate sponses of the respondents. These imputa- to two standard errors above the estimate tions usually are made separately within would include the average estimate derived various groups of sample members, formed from all possible samples. by attempting to place respondents and nonrespondents together that have ‘‘similar’’ Thus, for a particular sample, one can say design or ancillary characteristics. Imputa- with the appropriate level of confidence tion for item nonresponse is usually made (e.g., 90 percent or 95 percent) that the av- by substituting for a missing item the re- erage of all possible samples is included in sponse to that item of a respondent having the constructed interval. Example of a confi- characteristics that are ‘‘similar’’ to those of dence interval: An estimate is 200 units with the nonrespondent. 972 Appendix III

For an estimate calculated from a sample Other (nonsampling) Errors: Since 1950, survey, the total error in the estimate is evaluation programs have been conducted composed of the sampling error, which can to provide information on the magnitude of usually be estimated from the sample, and some sources of nonsampling errors such the nonsampling error, which usually cannot as response bias and undercoverage in be estimated from the sample. The total each census. Results from the evaluation error present in a population quantity ob- program for the 1990 census indicate that tained from a complete census is composed of only nonsampling errors. Ideally, esti- the net under coverage amounted to about mates of the total error associated with data 1.5 to 2 percent of the total resident popula- given in the Statistical Abstract tables tion. should be given. However, due to the un- Sources of Additional Material: U.S. availability of estimates of nonsampling er- Bureau of the Census, The Coverage of rors, only estimates of the levels of sam- Population in the 1980 Census, PHC80-E4; pling errors, in terms of estimated standard Content Reinterview Study: Accuracy of errors or coefficients of variation, are avail- Data for Selected Population and Housing able. To obtain estimates of the estimated Characteristics as Measured by Reinter- standard errors from the sample of interest, view, PHC80-E2; 1980 Census of Popula- obtain a copy of the referenced report which tion, vol. 1., (PC80-1), appendixes B, C, appears at the end of each table. and D. Principal data bases.—Beginning below Source and Title: U.S. Bureau of the Cen- are brief descriptions of 35 of the sample sus, Current Population Survey (CPS) surveys and censuses that provide a sub- Tables: See tables citing Current Population stantial portion of the data contained in this Reports primarily in section 1, but also in Abstract. sections 2, 3, 4, 8, 12, 13, 14, 18, 23, and 29. Many Bureau of Labor Statistics’ (BLS) SECTION 1. POPULATION tables in section 13 are CPS based. Universe, Frequency, and Types of Data: Source and Title: Bureau of the Census, Nationwide monthly sample survey of civil- Census of Population ian noninstitutional population, 15 years old Tables: See tables citing Census of Popula- or over, to obtain data on employment, tion in section 1 and also in sections 2, 4, unemployment, and a number of other 6, 8, 13, 14, 21, 25, and 29. characteristics. Universe, Frequency, and Types of Data: Type of Data Collection Operation: Multi- Complete count of U.S. population con- stage probability sample of about 60,000 ducted every 10 years since 1790. Data households in 729 PSU’s in 1993. Over- obtained on number and characteristics of sampling in some States and the largest inhabitants. MSA’s to improve reliability for those areas Type of Data Collection Operation: In of employment data on annual average 1970, 1980, and 1990 complete census for basis. A continual sample rotation system is some items—age, sex, race, marital status, used. Households are in sample 4 months, and relationship to household head. In out for 8 months, and in for 4 more. Month- 1970, other items collected from a 5% and to-month overlap is 75%; year-to-year over- a 15% probability (systematic) sample of lap is 50%. the population. In 1980, approximately 19% Data Collection and Imputation Proce- of the housing units were included in the dures: For first and fifth months that a sample; in 1990, approximately 17%. household is in sample, personal interviews; Data Collection and Imputation Proce- other months, approximately, 85% of the dures: In 1970, extensive use of mail ques- data collected by phone. Imputation is done tionnaires in urban areas; personal for both item and total nonresponse. Adjust- interviews in most rural areas. In 1980 and ment for total nonresponse is done by a 1990, mail questionnaires were used in predefined cluster of units, by MSA size even more areas than in 1970, with per- and residence; for item nonresponse impu- sonal interviews in the remainder. Extensive tation varies by subject matter. telephone and personal followup for nonre- Estimates of Sampling Error: Estimated spondents was done in the censuses. Impu- CV’s on national annual averages for labor tations were made for missing force, total employment, and nonagricultural characteristics. employment, 0.2%; for total unemployment Estimates of Sampling Error: Sampling and agricultural employment, 1.0% to 2.5%. errors for data are estimated for all items The estimated CV’s for family income and collected by sample and vary by character- poverty rate for all persons in 1986 are istic and geographic area. The CV’s for 0.5% and 1.5%, respectively. CV’s for sub- national and State estimates are generally national areas, such as States, would be very small. larger and would vary by area. Appendix III 973

Other (nonsampling) Errors: Estimates of SECTION 3. HEALTH AND NUTRITION response bias on unemployment are not available, but estimates of unemployment Source and Title: U.S. National Center for are usually 5% to 9% lower than estimates Health Statistics, National Health Interview from reinterviews. Four to 5.0% of sample Survey (NHIS) households unavailable for interviews. Tables: 178, 179, 204, 207, 208, 210, 214, 215, 219, 221 and 222. Sources of Additional Material: U.S. Bureau of the Census and Bureau of Labor Universe, Frequency, and Types of Data: Statistics, Concepts and Methods Used in Continuous data collection covering the Labor Force Statistics from Current Popula- civilian noninstitutional population to obtain tion Survey (Census series P-23, No. 62; information on personal and demographic BLS Report No. 463) and Bureau of the characteristics, illnesses, injuries, impair- Census, Current Population Survey (Tech. ments, and other health topics. Paper 40) and Bureau of Labor Statistics, Type of Data Collection Operation: Multi- Employment and Earnings, monthly, stage probability sample of 42,000 house- Explanatory Notes and Estimates of Error, holds (in 376 PSU’s) selected in groups of tables 1-A through 1-H and BLS Handbook about four adjacent households. of Methods, Chapter 1 (Bulletin 2414.) Data Collection and Imputation Proce- dures: Personal household interviews with SECTION 2. VITAL STATISTICS extensive followup of nonrespondents. Data are adjusted for nonresponse by imputation Source and Title: U.S. National Center for procedure based on ‘‘average’’ characteris- Health Statistics (NCHS), Vital Registration tics of persons in interviewed households in System the same geographic area. Tables: See tables citing Vital Statistics of Estimates of Sampling Error: Estimated the United States; 314 in section 5; and CV’s: For physician visits by males, 1.5%; 1346 in section 29. for workdays lost by males, 3.5%; for per- Universe, Frequency, and Types of Data: sons injured at home, 4.7%. Annual data on births and deaths in the Other (nonsampling) Errors: Response rate United States. was 95.7% in 1985 for the NHIS. Type of Data Collection Operation: Mortal- Sources of Additional Material: U.S. ity data based on complete file of death National Center for Health Statistics, records, except 1972, based on 50% ‘‘Current Estimates from the National Health sample. Natality statistics 1951-71, based Interview Survey, U.S., 1983,’’ Vital and on 50% sample of birth certificates, except Health Statistics, series 10. a 20% to 50% in 1967, received by NCHS. (See section 13 for information pertaining to Beginning 1972, data from some States table 173, section 15 for table 167, and received through Vital Statistics Cooperative section 27 for table 174.) Program (VSCP) and complete file used; data from other States based on 50% SECTION 4. EDUCATION sample. Beginning 1986, all reporting areas Source and Title: U.S. Department of Edu- participated in the VSCP. cation, National Center for Education Statis- Data Collection and Imputation Proce- tics, Higher Education General Information dures: Reports based on records from Survey (HEGIS), Fall Enrollment in Institu- registration offices of all States, District of tions of Higher Education; beginning 1986, Columbia, New York City, Puerto Rico, Integrated Postsecondary Education Data Virgin Islands, and Guam. Survey (IPEDS), Fall Enrollment Estimates of Sampling Error: For recent Tables: 277, 278, 281 and 282. years, CV’s for births are small due to large Universe, Frequency, and Types of Data: portion of total file in sample (except for Annual survey of all institutions and very small estimated totals). branches listed in the Directory to obtain Other (nonsampling) Errors: Data on births data on total enrollment by sex, level of and deaths believed to be at least 99% enrollment, type of program, racial/ethnic complete. characteristics (in alternate years) and Sources of Additional Material: U.S. attendance status of student, and on first- National Center for Health Statistics, Vital time students. Statistics of the United States, vol. I and Type of Data Collection Operation: vol. II, annual, and Monthly Vital Statistics Complete census. Report. Data Collection and Imputation Proce- (See section 1 above for information pertaining dures: Survey package is usually mailed in to tables 99-104.) the spring with surveys due at varying dates 974 Appendix III

in the summer and fall; mail and phone earned degrees and other formal awards, followup procedures for nonrespondents. conferred by ﬁeld of study, level of degree, Missing data are imputed by using data of sex, and by racial/ethnic characteristics (in similar institutions. alternate years). Estimates of Sampling Error: Not Type of Data Collection Operation: applicable. Complete census. Other (nonsampling) Errors: Approximately Data Collection and Imputation Proce- 87% response rate. dures: Survey package is usually mailed in Sources of Additional Material: U.S. the spring with surveys due at varying dates Department of Education, National Center in the summer and fall; mail and phone fol- for Education Statistics, Fall Enrollment in lowup procedures for nonrespondents. Higher Education, annual. Missing data are imputed by using data of Source and Title: U.S. Department of similar institutions. Education, National Center for Education Estimates of Sampling Error: Not Statistics, Higher Education General Infor- applicable. mation Survey (HEGIS), Financial Statistics Other (nonsampling) Errors: For 1989-90, of Institutions of Higher Education; begin- approximately 92.3% response rate. ning 1986, Integrated Postsecondary Edu- Sources of Additional Material: U.S. cation Data Survey (IPEDS), Financial Department of Education, National Center Statistics of Institutions of Higher Education for Education Statistics, Earned Degrees Tables: 229, 232, 281, and 284. Conferred, annual. Universe, Frequency, and Types of Data: (See sections 1 and 9 for information pertaining Annual survey of all institutions and to the Bureau of the Census and section 3 branches listed in the Education Directory, above for the National Center for Health Colleges and Universities to obtain data on Statistics.) ﬁnancial status and operations, including current funds revenues, current funds SECTION 5. LAW ENFORCEMENT, expenditures, and physical plant assets. COURTS, AND PRISONS Type of Data Collection Operation: Com- Source and Title: U.S. Federal Bureau of plete census. Investigation, Uniform Crime Reporting Data Collection and Imputation Proce- (UCR) Program dures: Survey package is usually mailed in Tables: 308-313, 315, 316, 322-324, and the spring with surveys due at varying dates 326. in the summer and fall; mail and phone Universe, Frequency, and Types of Data: followup procedures for nonrespondents. Monthly reports on the number of criminal Missing data are imputed by using data of offenses that become known to law enforce- similar institutions. ment agencies. Data are collected on Estimates of Sampling Error: Not crimes cleared by arrest, by age, sex, and applicable. race of offender, and on assaults on law Other (nonsampling) Errors: For 1990, a enforcement officers. 87% response rate. Imputed expenditures Type of Data Collection Operation: Crime amounted to about 2.8% of total expendi- statistics are based on reports of crime data tures. submitted either directly to the FBI by con- Sources of Additional Material: U.S. tributing law enforcement agencies or Department of Education, National Center through cooperating State UCR programs. for Education Statistics, Financial Statistics Data Collection and Imputation Proce- of Institutions of Higher Education, annual. dures: States with UCR programs collect Source and Title: U.S. Department of data directly from individual law enforce- Education, National Center for Education ment agencies and forward reports, pre- Statistics, Higher Education General Infor- pared in accordance with UCR standards, mation Survey (HEGIS), Degrees and Other to FBI. Accuracy and consistency edits are Formal Awards Conferred. Beginning 1986, performed by FBI. Integrated Postsecondary Education Data Estimates of Sampling Error: Not Survey (IPEDS), Degrees and Other Formal applicable. Awards Conferred. Other (nonsampling) Errors: Coverage of Tables: 298-303. 97% of the population (96.6% in MSA’s, Universe, Frequency, and Types of Data: 85.8% in ‘‘other cities,’’ and 85.6% in rural Annual survey of all institutions and areas) by UCR program, though varying branches listed in the Education Directory, number of agencies report. Some error may Colleges and Universities to obtain data on be present through incorrect reporting. Appendix III 975

Sources of Additional Material: U.S. Fed- SECTION 9. STATE AND LOCAL eral Bureau of Investigation, Crime in the GOVERNMENT FINANCES AND United States. EMPLOYMENT Source and Title: U.S. Bureau of Justice Source and Title: U.S. Bureau of the Statistics (BJS), National Crime Victimiza- Census, Census of Governments tion Survey Tables: See tables in section 9 citing Census Tables: 317-321. of Governments and tables 452 and 454 in Universe, Frequency, and Types of Data: section 8. Monthly survey of individuals and house- Universe, Frequency, and Types of Data: holds in the United States to obtain data Survey of all governmental units in the on criminal victimization of those units for United States conducted every 5 years to compilation of annual estimates. obtain data on government revenue, expen- Type of Data Collection Operation: ditures, debt, assets, employment and National probability sample survey of employee retirement systems, property val- about 50,000 interviewed households ues, public school systems, and number, in 376 PSU’s selected from a list of size, and structure of governments. addresses from the 1980 census, supple- Type of Data Collection Operation: Com- mented by new construction permits and plete census. List of units derived through an area sample where permits are not classification of government units recently required. authorized in each State and identification, Data Collection and Imputation Proce- counting, and classification of existing local dures: Interviews are conducted every 6 governments and public school systems. months for 3 years for each household in Data Collection and Imputation Proce- the sample; 8,000 households are inter- dures: Data collected through field and viewed monthly. Personal interviews are office compilation of financial data from offi- used in the first and fifth interviews; the cial records and reports for States and large intervening interviews are conducted by local governments; mail canvass of selected telephone whenever possible. data items, like State tax revenue and Estimates of Sampling Error: CV’s in 1985: employee retirement systems; and collec- 2.2% for crimes of violence; 14.8% for esti- tion of local government statistics through mate of rape counts; 5.5% for robbery central collection arrangements with State counts; 2.5% for assault counts; 1.4% for governments. personal larceny counts; 2.2% for burglary Estimates of Sampling Error: Not counts; 1.8% for household larceny; 1.3% applicable. for all household crimes; and 4.8% for Other (nonsampling) Errors: Some non- motor vehicle theft counts. sampling errors may arise due to possible Other (nonsampling) Errors: Respondent inaccuracies in classification, response, and recall errors which may include reporting processing. incidents for other than the reference Sources of Additional Material: U.S. period; interviewer coding and processing Bureau of the Census, Census of Govern- errors; and possible mistaken reporting or ments, 1987, various reports, and State classifying of events. Adjustment is made Government Finances in 1990, GF 90, for a household noninterview rate of about No. 3. 4% and for a smaller within-household non- Source and Title: U.S. Bureau of the interview rate. Census, Annual Surveys of State and Local Sources of Additional Material: U.S. Government Bureau of Justice Statistics, Criminal Tables: See tables citing Public Employment Victimization in the United States, annual. and Governmental Finances in section 9; (See section 2 for details on table 314 table 256 in section 4; table 333 in section and section 9 for details on table 333.) 5; and table 598 in section 12. SECTION 7. PARKS AND RECREATION Universe, Frequency, and Types of Data: Sample survey conducted annually to (See section 27 for details on table 405 and obtain data on revenue, expenditure, debt, 427.) and employment of State and local governments. Universe is all governmental units in SECTION 8. ELECTIONS the United States (about 83,000). Type of Data Collection Operation: Sample (See section 1 above for information per- of about 22,000 units includes all State gov- taining to tables 459 and 460 and section ernments, county governments with 9 for information pertaining to tables 452 50,000+ population, municipalities and and 454.) townships with 25,000+ population, all 976 Appendix III

school districts with 10,000+ enrollment in Estimates of Sampling Error: Estimated October 1986, and other governments CV’s for tax year 1987: Adjusted gross meeting certain criteria; probability sample income less deficit .13%; salaries and for remaining units. wages .20%; and tax-exempt interest Data Collection and Imputation Proce- received 4.51%. (State data not subject to dures: Field and office compilation of data sampling error.) from official records and reports for States Other (nonsampling) Errors: Processing and large local governments; central collec- errors and errors arising from the use of tion of local governmental financial data tolerance checks for the data. through cooperative agreements with a Sources of Additional Material: U.S. Inter- number of State governments; mail canvass nal Revenue Service, Statistics of Income, of other units with mail and telephone fol- Individual Income Tax Returns, annual. lowups of nonrespondents. Data for nonresponses are imputed from previous year SECTION 12. SOCIAL INSURANCE data or obtained from secondary sources, AND HUMAN SERVICES if available. Estimates of Sampling Error: CV’s for Source and Title: U.S. Social Security estimates of major employment and finan- Administration, Benefit Data cial items are generally less than 2% for Tables: 594 and 595. most States and less than 1.2% for the Universe, Frequency, and Types of Data: majority of States. All persons receiving monthly benefits Other (nonsampling) Errors: Nonresponse under Title II of Social Security Act. Data on rate is less than 15% for number of units. number and amount of benefits paid by Other possible errors may result from unde- type and State. tected inaccuracies in classification, Type of Data Collection Operation: Data response, and processing. based on administrative records. Data Sources of Additional Material: U.S. based on 100% files, as well as 10% and Bureau of the Census, Public Employment 1% sample files. in 1990, GE 90, No. 1, Governmental Data Collection and Imputation Proce- Finances in 1989-90, GF 90, No. 5, and dures: Records used consist of actions pur- Census of Governments, 1987, various suant to applications for benefits, updated reports. by subsequent post-entitlement actions. SECTION 10. FEDERAL GOVERNMENT Estimates of Sampling Error: Varies by size of estimate and sample file size. Source and Title: U.S. Internal Revenue Service, Statistics of Income, Individual Other (nonsampling) Errors: Processing Income Tax Returns errors, which are believed to be small. Tables: 533-527. Sources of Additional Material: U.S. Social Universe, Frequency, and Types of Data: Security Administration, Annual Statistical Annual study of unaudited individual income Supplement to the Social Security Bulletin. tax returns, forms 1040, 1040A, and Source and Title: U.S. Social Security 1040EZ, filed by U.S. citizens and resi- Administration, Supplemental Security dents. Data provided on various financial Income (SSI) Program characteristics by size of adjusted gross Tables: 609-612. income, marital status, and by taxable and Universe, Frequency, and Types of Data: nontaxable returns. Data by State, based All eligible aged, blind, or disabled persons on 100% file, also include returns from receiving SSI benefit payments under SSI 1040NR, filed by nonresident aliens plus program. Data include number of persons certain self-employment tax returns. receiving federally administered SSI, Type of Data Collection Operation: Annual amounts paid, and State administered stratified probability sample of approxi- supplementation. mately 125,000 returns broken into sample Type of Data Collection Operation: Data strata based on the larger of total income or based on administrative records. total loss amounts as well as the size of business plus farm receipts. Sampling rates Data Collection and Imputation Proce- for sample strata varied from .025% to dures: Data adjusted to reflect returned 100%. checks and overpayment refunds. For fed- Data Collection and Imputation Proce- erally administered payments, actual dures: Computer selection of sample of tax adjusted amounts are used. return records. Data adjusted during editing Estimates of Sampling Error: Not for incorrect, missing, or inconsistent entries applicable. to ensure consistency with other entries on Other (nonsampling) Errors: Processing return. errors, which are believed to be small. Appendix III 977

Sources of Additional Material: U.S. Social by urban wage earners and clerical workers Security Administration, Annual Statistical prior to 1978, and urban consumers there- Supplement to the Social Security Bulletin. after. Both indexes continue to be pub- (See section 1 above for information pertaining lished. to the Current Population Survey, section 3 Type of Data Collection Operation: Prior to for information pertaining to the National Cen- 1978, sample of various consumer items in ter for Health Statistics, and section 9 for 56 urban areas; thereafter, in 85 PSU’s, information pertaining to Annual Surveys of except from January 1987 through March State and Local Government.) 1988, when 91 areas were sampled. Data Collection and Imputation Proce- SECTION 13. LABOR FORCE, dures: Prices of consumer items are EMPLOYMENT, AND EARNINGS obtained from about 57,000 housing units, Source and Title: U.S. Bureau of Labor and 19,000 other reporters in 85 areas. Statistics (BLS), Current Employment Sta- Prices of food, fuel, and a few other items tistics (CES) Program are obtained monthly; prices of most other Tables: 666-668, 673; in section 3, table 173; commodities and services are collected in section 21, table 1008; and in section 22, every month in the five largest geographic table 1075. areas and every other month in others. Universe, Frequency, and Types of Data: Estimates of Sampling Error: Estimates of Monthly survey covering about 6 million standard errors are not available at present. nonagricultural establishments to obtain Other (nonsampling) Errors: Errors result data on employment, hours, and earnings, from inaccurate reporting, difficulties in by industry. defining concepts and their operational Type of Data Collection Operation: Sample implementation, and introduction of product survey of over 390,000 establishments in quality changes and new products. March 1993. Sources of Additional Material: U.S. Data Collection and Imputation Proce- Bureau of Labor Statistics, The Consumer dures: Cooperating State agencies mail Price Index: 1987 Revision, Report 736, questionnaires to sample establishments to and BLS Handbook of Methods, Chapter develop State and local estimates; informa- 19, Bulletin 2414. tion is forwarded to BLS where national Source and Title: U.S. Bureau of Labor estimates are prepared. Statistics, Producer Price Index (PPI) Estimates of Sampling Error: Estimated Tables: 760, 767-769, and in section 21, CV’s for employment, 0.1%, for average table 1058; in section 24, table 1154; and weekly hours paid, 0.4% and for average in section 25, table 1203. hourly earnings, 0.3%. Universe, Frequency, and Types of Data: Other (nonsampling) Errors: Estimates of Monthly survey of producing companies to employment adjusted annually to reflect determine price changes of all commodities complete universe. Average adjustment is produced in the United States for sale in 0.2%. commercial transactions. Data on agricul- Sources of Additional Material: U.S. ture, forestry, fishing, manufacturing, min- Bureau of Labor Statistics, Employment and ing, gas, electricity, public utilities, and a Earnings, monthly, Explanatory Notes and few services. Estimates of Error, tables 2-A through 2-G Type of Data Collection Operation: and BLS Handbook of Methods, Chapter 2, Probability sample of approximately 3,100 Bulletin 2414 (Sept. 1992). commodities and about 75,000 quotations (See section 1 above for information pertaining per month. to the Current Population Survey.) Data Collection and Imputation Proce- dures: Data are collected by mail. If trans- SECTION 14. INCOME, EXPENDITURES, action prices are not supplied, list prices are AND WEALTH used. Some prices are obtained from trade publications, organized exchanges, and (See section 1 above for information pertain- government agencies. To calculate index, ing to the Bureau of the Census.) price changes are multiplied by their relative weights based on total net selling value of SECTION 15. PRICES all commodities in 1982. Source and Title: U.S. Bureau of Labor Sta- Estimates of Sampling Error: Not tistics (BLS), Consumer Price Index (CPI) applicable. Tables: 760-764, 777, 778, and in section 3, Other (nonsampling) Errors: Not available table 167. at present. Universe, Frequency, and Types of Data: Sources of Additional Material: U.S. Monthly survey of price changes of all types Bureau of Labor Statistics, BLS Handbook of consumer goods and services purchased of Methods, Chapter 16, Bulletin 2414. 978 Appendix III

SECTION 17. BUSINESS ENTERPRISE Data Collection and Imputation Proce- dures: Computer selection of sample of tax Source and Title: U.S. Internal Revenue return records. Data are adjusted during Service, Statistics of Income, Sole Propri- editing for incorrect, missing, or inconsistent etorship Returns and Statistics of Income entries to ensure consistency with other Bulletin entries on return. Data not available due to Tables: 846-849. regulations are not imputed. Universe, Frequency, and Types of Data: Estimates of Sampling Error: Estimated Annual study of unaudited income tax CV’s for tax year 1988 (latest available): returns of nonfarm sole proprietorships, For number of partnerships, .51%; business form 1040 with business schedules. Data receipts, .78%; net income, 3.03%; net loss, provided on various financial characteristics 2.21% and total assets, 1.22%. by industry. Other (nonsampling) Errors: Processing Type of Data Collection Operation: errors and errors arising from the use of Stratified probability sample of approxi- tolerance checks for the data. mately 31,000 sole proprietorships for tax Sources of Additional Material: U.S. Inter- year 1990. The sample is classified based nal Revenue Service, Statistics of Income, on presence or absence of certain business Partnership Returns and Statistics of schedules; the larger of total income or Income Bulletin, vol. 10, No. 1 (summer loss; and size of business plus farm 1990). receipts. Sampling rates vary from .043% to Source and Title: U.S. Internal Revenue 100%. Service, Corporation Income Tax Returns Data Collection and Imputation Proce- Tables: 846-848 and 854-856. dures: Computer selection of sample of tax Universe, Frequency, and Types of Data: return records. Data adjusted during editing Annual study of unaudited corporation for incorrect, missing, or inconsistent entries income tax returns, Forms 1120 and 1120 to ensure consistency with other entries on (A, F, L, PC, REIT, RIC, and S), filed by return. corporations or businesses legally defined Estimates of Sampling Error: Estimated as corporations. Data provided on various CV’s for tax year 1990 are not available; for financial characteristics by industry and size 1987 (the latest available): For sole propri- of total assets, and business receipts. etorships, business receipts, 1.66%; net Type of Data Collection Operation: income, (less loss), 1.33%; depreciation Stratified probability sample of approxi- 2.17%; interest expense 2.80%; and mately 85,000 returns for 1987, distributed employee benefit programs 7.55%. by sample classes generally based on type Other (nonsampling) Errors: Processing of return, size of total assets, size of net errors and errors arising from the use of income or deficit, and selected business tolerance checks for the data. activity. Sampling rates for sample strata Sources of Additional Material: U.S. Inter- varied from .25% to 100%. nal Revenue Service, Statistics of Income, Data Collection and Imputation Proce- Sole Proprietorship Returns (for years dures: Computer selection of sample of tax through 1980) and Statistics of Income Bul- return records. Data adjusted during editing letin, vol. 10, No. 1 (summer 1990). for incorrect, missing, or inconsistent entries Source and Title: U.S. Internal Revenue to ensure consistency with other entries on Service, Statistics of Income, Partnership return and to achieve statistical definitions. Returns and Statistics of Income Bulletin Estimates of Sampling Error: Estimated Tables: 846-848, 850, and 851. CV’s for 1988: Number of returns in sub- Universe, Frequency, and Types of Data: groups ranged from 1.4% with assets under Annual study of unaudited income tax $100,000, to 0% with assets over $100 mil.; returns of partnerships, Form 1065. Data for amount of net income and amount of provided on various financial characteristics income tax, .18% by industry. Other (nonsampling) Errors: Processing Type of Data Collection Operation: errors and errors arising from the use of Stratified probability sample of approxi- tolerance checks for the data. mately 28,000 partnership returns from a Sources of Additional Material: U.S. Inter- population of 1,660,000 filed during calen- nal Revenue Service, Statistics of Income, dar year 1990. The sample is classified Corporation Income Tax Returns, annual. based on combinations of gross receipts, net income or loss, and total assets, and on SECTION 18. COMMUNICATIONS industry. Sampling rates vary from .04% to (See section 1 for information pertaining to 100%. table 904, and section 26 for table 916.) Appendix III 979

SECTION 19. ENERGY include value, quantity, and shipping weight Source and Title: U.S. Energy Information of exports by commodity, country of desti- Administration, Residential Energy Con- nation, Customs district of exportation, and sumption Survey mode of transportation. Tables: 943, 944 and table 1233 in section Type of Data Collection Operation: 25. Shipper’s Export Declarations are required Universe, Frequency, and Types of Data: to be filed for the exportation of merchan- Triennial survey of households and fuel dise valued over $1,500. Customs officials suppliers. Data are obtained on energy- collect and transmit the documents to the related household characteristics, housing Bureau of the Census on a flow basis for unit characteristics, use of fuels, and energy data compilation. Value data for shipments consumption and expenditures by fuel type. valued under $1,501 are estimated, based Type of Data Collection Operation: Prob- on established percentages of individual ability sample of 7,183 eligible units in 129 country totals. PSU’s. For responding units, fuel consump- Data Collection and Imputation Proce- tion and expenditure data obtained from dures: Statistical copies of Shipper’s Export fuel suppliers to those households. Declarations are received on a daily basis Data Collection and Imputation Proce- from Customs ports throughout the country dures: Personal interviews. Extensive fol- and subjected to a monthly processing lowup of nonrespondents including mail cycle. They are fully processed to the questionnaires for some households. extent they reflect items valued over Adjustments for nonrespondents were made $1,500. Estimates for shipments valued at in weighting for respondents. Most item $1,500 or less are made, based on estab- nonresponses were imputed. lished percentages of individual country Estimates of Sampling Error: Estimated totals. CV’s for household averages: For consump- Estimates of Sampling Error: Not tion, 1.3%; for expenditures, 1.0%; for vari- applicable. ous fuels, values ranged from 1.4% for Other (nonsampling) Errors: Clerical and electricity to 5.9% for LPG. complex computer checks intercept most Other (nonsampling) Errors: Household processing errors and minimize otherwise response rate of 86.7%. Nonconsumption significant reporting errors; other nonsam- data were mostly imputed for mail respon- pling errors are caused by undercounting of dents (5.2% of eligible units). Usable exports to Canada due to the nonreceipt of responses from fuel suppliers for various some Shipper’s Export Declarations. fuels ranged from 82.8% for electricity to Sources of Additional Material: U.S. 55.7% for fuel oil. Bureau of the Census, U.S. Merchandise Sources of Additional Material: U.S. Trade: Exports, General Imports, and Energy Information Administration, House- Imports for Consumption, SITC, Commodity hold Energy Consumption and Expendi- by Country, FT 925. tures, 1990 and Housing Characteristics, Source and Title: U.S. Bureau of the 1990. Census, Foreign Trade—Import Statistics Tables: See Bureau of the Census citations SECTION 21. for import statistics in source notes in sec- TRANSPORTATION—LAND tions 22 and 28 and also table 1185 in section 24; and 1354 in section 29. (See section 1 for information pertaining to Universe, Frequency, and Types of Data: table 1030, section 13 for table 1008, and The import entry documents collected by section 15 for table 1058 and section 27 for Customs are processed each month to table 1010.) obtain data on the movement of merchandise imported into the United States. Data SECTION 22. TRANSPORTATION—AIR obtained include value, quantity, and ship- AND WATER ping weight by commodity, country of origin, Source and Title: U.S. Bureau of the Customs district of entry, and mode of Census, Foreign Trade—Export Statistics transportation. Tables: See Bureau of the Census citations Type of Data Collection Operation: Import for export statistics in source notes in sec- entry documents are required to be filed for tions 22 and 28 and also table 1185 in sec- the importation of goods into the United tion 24; and 1354 in section 29. States valued over $1,000 or for articles Universe, Frequency, and Types of Data: which must be reported on formal entries. The export declarations collected by Cus- Customs officials collect and transmit statis- toms are processed each month to obtain tical copies of the documents to the Bureau data on the movement of U.S. merchandise of the Census on a flow basis for data com- exports to foreign countries. Data obtained pilation. Estimates for shipments valued 980 Appendix III

under $1,001 and not reported on formal Other (nonsampling) Errors: Nonsampling entries are based on established percent- errors are due to incompleteness of the ages of individual country totals. census mailing list, duplications on the list, Data Collection and Imputation Proce- respondent reporting errors, errors in editing dures: Statistical copies of import entry reported data, and in imputation for missing documents, received on a daily basis from data. Evaluation studies are conducted to Customs ports of entry throughout the measure certain nonsampling errors such country, are subjected to a monthly pro- as list coverage and classification error. cessing cycle. They are fully processed to Results from the evaluation program for the the extent they reflect items valued at 1987 census indicate the net under cover- $1,001 and over or items which must be age amounted to about 7.2% of the nations reported on formal entries. total farms. Estimates of Sampling Error: Not Sources of Additional Material: U.S. applicable. Bureau of the Census, 1992 Census of Other (nonsampling) Errors: Verification of Agriculture, Volume 2, Subject Series— statistical data reporting by Customs offi- Part 1, Agriculture Atlas of the U.S., Part 2, cials prior to transmittal and a subsequent Coverage Evaluation, Part 3, Rankings of program of clerical and computer checks States and Counties, Part 4, History, Part 5, are utilized to hold nonsampling errors aris- ZIP Code Tabulation of Selected Items, ; ing from reporting and/or processing errors and Volume 4 Farm and Ranch Irrigation to a minimum. Survey. Sources of Additional Material: U.S. Source and Title: U.S. Department of Agri- Bureau of the Census, U.S. Merchandise culture, National Agricultural Statistics Serv- Trade: Exports, General Imports and ice (NASS), Basic Area Frame Sample Imports for Consumption, SITC, Commodity Tables: See tables citing NASS in source by Country, FT 925. notes in section 23, which pertain to this or (See section 13 for information pertaining to the following two surveys. table 1075.) Universe, Frequency, and Types of Data: Two annual area sample surveys of U.S. SECTION 23. AGRICULTURE farm operators: June agricultural survey collects data on planted acreage and livestock Source and Title: U.S. Bureau of the inventories; and a February Farm Costs Census, Census of Agriculture. and Returns survey that collects data on Tables: 1096-1103. total farm production, expenses and specific Universe, Frequency, and Types of Data: commodity costs of production. Complete count of U.S. farms and ranches Type of Data Collection Operation: conducted once every five years with data Stratified probability sample of about 16,000 at the national, state, and county level. Data land area units of about 1 sq. mile (range published on farm numbers and related from 0.1 sq. mile in cities to several sq. items/characteristics. miles in open grazing areas). Sample Type of Data Collection Operation: includes 60,000 parcels of agricultural land. Complete census for— number of farms; About 20% of the sample replaced annu- land in farms; estimated market value of ally. land and buildings, machinery and equip- Data Collection and Imputation Proce- ment, agriculture products sold; total crop- dures: Data collection is by personal enu- land; irrigated land; total farm production meration. Imputation is based on expenses; farm operator characteristics; enumerator observation or data reported by livestock and poultry inventory and sales; respondents having similar agricultural char- and selected crops harvested. acteristics. Data Collection and Imputation Proce- Estimates of Sampling Error: Estimated dures: Data collection is by mailing ques- CV’s range from 1% to 2% for regional esti- tionnaires to all farmers and ranchers. mates to 3% to 6% for State estimates of Nonrespondents are conducted by tele- livestock inventories. phone and correspondence followups. Other (nonsampling) Errors: Minimized Imputations were made for all nonresponse through rigid quality controls on the collec- item/characteristics. tion process and careful review of all Estimates of Sampling Error: Variability in reported data. the estimates is due to the sample selection Sources of Additional Material: U.S. and estimation for items collected by Department of Agriculture, SRS, Scope and sample and census nonresponse estimation Methods of the Statistical Reporting Serv- procedures. The CV’s for national and state ice, (name changed to National Agricultural estimates are generally very small. Approxi- Statistics Service), Miscellaneous Publica- mately 85% response rate. tion No. 1308, September 1983 (revised). Appendix III 981

Source and Title: U.S. Department of Other (nonsampling) Errors: In addition to Agriculture, National Agricultural Statistics above, replicated sampling procedures used Service (NASS), Multiple Frame Surveys to monitor effects of changes in survey pro- Tables: See tables citing NASS in source cedures. notes in section 23, which pertain to this or Sources of Additional Material: U.S. the following survey. Department of Agriculture, SRS, Scope and Universe, Frequency, and Types of Data: Methods of the Statistical Reporting Serv- Surveys of U.S. farm operators to obtain ice, (name changed to National Agricultural data on major livestock inventories, Statistics Service), Miscellaneous Publica- selected crop acreages and production, tion No. 1308, September 1983 (revised). grain stocks, and farm labor characteristics; (See section 1 above for information pertaining and to obtain farm economic data for price to the Census of Population and Current indexing. Population Survey.) Type of Data Collection Operation: Primary frame is obtained from general or special SECTION 24. NATURAL RESOURCES purpose lists, supplemented by a probability sample of land areas used to estimate for (See section 15 for table 1154, and section list incompleteness. 22 for table 1185.) Data Collection and Imputation Proce- SECTION 25. CONSTRUCTION dures: Mail, telephone, or personal inter- AND HOUSING views used for initial data collection. Mail nonrespondent followup by phone and per- Source and Title: U.S. Bureau of the sonal interviews. Imputation based on aver- Census, Monthly Survey of Construction age of respondents. Tables: 1212 and 1214-1216. Estimates of Sampling Error: Estimated CV Universe, Frequency, and Types of Data: for number of hired farm workers is about Survey conducted monthly of newly con- 3%. Estimated CV’s range from 1% to 2% structed housing units (excluding mobile for regional estimates to 3% to 6% for State homes). Data are collected on the start, estimates of livestock inventories. completion, and sale of housing. (Annual figures are aggregates of monthly esti- Other (nonsampling) Errors: In addition to mates.) above, replicated sampling procedures used Type of Data Collection Operation: Prob- to monitor effects of changes in survey pro- ability sample of housing units obtained cedures. from building permits selected from 17,000 Sources of Additional Material: U.S. places. For nonpermit places, multistage Department of Agriculture, SRS, Scope and probability sample of new housing units Methods of the Statistical Reporting Serv- selected in 169 PSU’s. In those areas, all ice, (name changed to National Agricultural roads are canvassed in selected enumera- Statistics Service), Miscellaneous Publica- tion districts. tion No. 1308, September 1983 (revised). Data Collection and Imputation Proce- Source and Title: U.S. Department of dures: Data are obtained by telephone Agriculture, National Agricultural Statistics inquiry and field visit. Service (NASS), Objective Yield Surveys Estimates of Sampling Error: Estimated CV Tables: See tables citing NASS in source of 3% to 4% for estimates of national totals, notes in section 23, which pertain to this or but are as high as 20% for estimated totals the preceding survey. of more detailed characteristics, such as Universe, Frequency, and Types of Data: housing units in multiunit structures. Surveys for data on corn, cotton, potatoes, Other (nonsampling) Errors: Response rate soybeans, wheat, and rice to forecast and is over 90% for most items. Nonsampling estimate yields. errors are attributed to definitional prob- Type of Data Collection Operation: Ran- lems, differences in interpretation of ques- dom location of plots in probability sample tions, incorrect reporting, inability to obtain of fields. Fields selected in June from Basic information about all cases in the sample, Area Frame Sample (see above). and processing errors. Data Collection and Imputation Proce- Sources of Additional Material: U.S. dures: Enumerators count and measure Bureau of the Census, Construction plant characteristics in sample fields. Pro- Reports, series C20, Housing Starts; C22, duction measured from plots at harvest. Housing Completions; and C25, New One- Harvest loss measured from post harvest Family Houses Sold and For Sale. gleanings. Source and Title: U.S. Bureau of the Estimates of Sampling Error: CV’s for Census, Value of New Construction Put in national estimates of production are about Place 2-3%. Tables: 1205-1207. 982 Appendix III

Universe, Frequency, and Types of Data: Tables: See tables citing Census of Housing Survey conducted monthly on total value of in source notes in section 25. all construction put in place in the current month, both public and private projects. Universe, Frequency, and Types of Data: Construction values include costs of materi- Census of all occupied and vacant housing, als and labor, contractors’ profits, overhead excluding group quarters, conducted every costs, cost of architectural and engineering 10 years as part of the decennial census work, and miscellaneous project costs. (see section 1 above) to determine charac- (Annual figures are aggregates of monthly teristics of U.S. housing. estimates.) Type of Data Collection Operation: For Type of Data Collection Operation: Varies 1970, 1980, and 1990, a complete count of by type of activity: Total cost of private one- some housing items (e.g. Owned or rented, family houses started each month is distrib- and value). In 1970, other items collected uted into value put in place using fixed from 5% and 15% probability samples patterns of monthly construction progress; selected from two sets of detailed questions using a multistage probability sample, data on housing (these two sets having some for private multifamily housing are obtained common items). In 1980, approximately by mail from owners of multiunit projects. 19% of the housing units were included in Data for residential additions and alterations the sample; in 1990, approximately 17%. are obtained in a quarterly survey measur- Data Collection and Imputation Proce- ing expenditures; monthly estimates are dures: In 1970, a self-enumeration census interpolated from quarterly data. Estimates using a mail-out/mail-back procedure was of value of private nonresidential buildings, used in most areas. In 1980 and 1990, mail and State and local government construc- questionnaires were used in even more tion are obtained by mail from owners (or areas than in 1970, with personal interviews agents) for a probability sample of projects. in the remainder. Followup for nonrespon- Estimates of farm nonresidential construc- dents and identification of vacant units done tion expenditures are based on U.S. by phone and personal visit. Department of Agriculture annual estimates of construction; public utility estimates are Estimates of Sampling Error: Sampling obtained from reports submitted to Federal errors for data are estimated for all items regulatory agencies and from private utility collected by sample and vary by character- companies; estimates for all other private istic and geographic area. construction (nonbuilding) are obtained by phasing F. W. Dodge contract award data; Other (nonsampling) Errors: Evaluation estimates of Federal construction are based studies for 1980 estimated the underenu- on monthly data supplied by Federal agen- meration of occupied housing units at 1.5%. cies. The missed rate in 1980 for all units was 2.6% or approximately 2.3 million units, 1 Data Collection and Imputation Proce- million of which were vacant housing units. dures: See ‘‘Type of Data Collection Operation.’’ Imputation accounts for Sources of Additional Material: U.S. approximately 20% of estimated value of Bureau of the Census, 1980 Census of construction each month. Population and Housing, The Coverage of Estimates of Sampling Error: CV estimates Housing in the 1980 Census, PHC80-E1, for private nonresidential building construc- July 1985. tion range from 3% for estimated value of Source and Title: U.S. Bureau of the Cen- industrial buildings to 9% for religious build- sus, American Housing Survey ings. CV is approximately 2% for total new private nonresidential buildings. Tables: See tables citing American Housing Other (nonsampling) Errors: For directly Survey in source notes. measured data series based on samples, Universe, Frequency, and Types of Data: some nonsampling errors may arise from Conducted nationally in the fall in odd num- processing errors, imputations, and misun- bered years to obtain data on the approxi- derstanding of questions. Indirect data mately 103 million occupied or vacant series are dependent on the validity of the housing units in the United States (group underlying assumptions and procedures. quarters are excluded). Data include char- Sources of Additional Material: U.S. acteristics of occupied housing units, vacant Bureau of the Census, Construction units, new housing and mobile home units, Reports, series C30, Value of New financial characteristics, recent mover Construction Put in Place. households, housing and neighborhood Source and Title: U.S. Bureau of the quality indicators, and energy characteris- Census, Census of Housing tics. Appendix III 983

Type of Data Collection Operation: The single-location firms (generally those with national sample was a multistage probability fewer than 10 employees) not mailed cen- sample with about 51,300 units eligible for sus questionnaires were estimated from interview in 1987. Sample units, selected administrative records of IRS and SSA. within 394 PSU’s, were surveyed over a Data for nonrespondents were imputed from 5-month period. related responses or administrative records Data Collection and Imputation Proce- from IRS and SSA. Approximately 8% of dures: For 1987, the survey was conducted total value of shipments was represented by by personal interviews. The interviewers fully imputed records in 1987. obtained the information from the occupants Estimates of Sampling Error: Not appli- or, if the unit was vacant, from informed cable. persons such as landlords, rental agents, or Other (nonsampling) Errors: Based on knowledgeable neighbors. evaluation studies, estimates of nonsam- Estimates of Sampling Error: For the pling errors for 1972 were about 1.3% for national sample, illustrations of the S.E. of estimated total payroll; 2% for total employ- the estimates are provided in the appendix ment; and 1% for value of shipments. Esti- B of the 1987 report. As an example, the mates for later years are not available. estimated CV is about 0.5% for the esti- Sources of Additional Material: U.S. mated percentage of owner occupied units Bureau of the Census, 1987 Census of with two persons. Manufactures, Industry Series, Geographic Other (nonsampling) Errors: Response rate Area Series, and Subject Series. was about 97%. Nonsampling errors may Source and Title: U.S. Bureau of the result from incorrect or incomplete Census, Annual Survey of Manufactures responses, errors in coding and recording, and processing errors. For the 1985 Tables: See tables citing Annual Survey of national sample, approximately 6% of the Manufactures in source notes. total housing inventory was not adequately Universe, Frequency, and Types of Data: represented by the AHS sample. Conducted annually to provide basic mea- Sources of Additional Material: U.S. sures of manufacturing activity for intercen- Bureau of the Census, Current Housing sal years for all manufacturing Reports, series H-150 and H-170, American establishments having one or more paid Housing Survey. employees. (See section 1 above for information pertaining Type of Data Collection Operation: Sam- to the Census of Population, section 15 per- pling frame is 350,000 establishments in the taining to table 1203, and section 19 for table 1987 Census of Manufactures (see above), 1233.) supplemented by Social Security Adminis- tration lists of new manufacturers and new Section 26. MANUFACTURES manufacturing establishments of multi- Source and Title: U.S. Bureau of the establishment companies identified annually Census, Census of Manufactures by the Census Bureau’s Company Organi- Tables: See tables citing Census of Manu- zation Survey. A probability sample of about factures in source notes in section 26 and 55,000 establishments is selected. All also table 916 in section 18 and table 1358 establishments of companies with more in section 29. than $500 million of manufacturing ship- Universe, Frequency, and Types of Data: ments in 1987 are included with certainty. Conducted every 5 years to obtain informa- All establishments with 250+ employees are tion on labor, materials, capital input and also included with certainty along with a output characteristics, plant location, and probability sample of smaller establish- legal form of organization for all plants in ments. the United States with one or more paid Data Collection and Imputation Proce- employees. Universe was 350,000 manu- dures: Survey is conducted by mail with facturing establishments in 1987. phone and mail followups of nonrespon- Type of Data Collection Operation: Com- dents. Imputation (for all nonresponse plete enumeration of data items obtained items) is based on previous year reports, or from 200,000 firms. Administrative records for new establishments in survey, on indus- from Internal Revenue Service and Social try averages. Security Administration are used for Estimates of Sampling Error: Estimated 150,000 smaller single-location firms, which standard errors for number of employees, were determined by various cutoffs based new expenditure, and for value added totals on size and industry. are given in annual publications. For U.S. Data Collection and Imputation Proce- level industry statistics, most estimated dures: Five mail and telephone followups standard errors are 2% or less, but vary for larger nonrespondents. Data for small considerably for detailed characteristics. 984 Appendix III

Other (nonsampling) Errors: Response rate Tables: 1292-1295, 1309, 1313, 1317, and is about 85%. Nonsampling errors include table 174 in section 3, tables 405 and 427 those due to collection, reporting, and tran- in section 7, and table 1010 in section 21. scription errors, many of which are cor- Universe, Frequency, and Types of Data: rected through computer and clerical Provides monthly estimates of retail sales checks. by kind of business and geographic area, Sources of Additional Material: U.S. and end-of-month inventories of retail Bureau of the Census, Annual Survey of stores; monthly wholesale sales and end-of- Manufactures, and Technical Paper 24. month inventories; and annual receipts of selected service industries. Section 27. DOMESTIC TRADE AND Type of Data Collection Operation: Prob- SERVICES ability sample of all firms from a list frame and, additionally, (before August 1993) for Source and Title: U.S. Bureau of the Cen- retail and service an area frame. The list sus, Census of Wholesale Trade, Census of frame is the Bureau’s Standard Statistical Retail Trade, Census of Service Industries Establishment List (SSEL) updated quar- Tables: See tables citing the above censuses terly for recent birth Employer Identification in source notes in section 27 and table (EIN) Numbers issued by the Internal Rev- 1358 in section 29. enue Service and assigned a kind-of- Universe, Frequency, and Types of Data: business code by the Social Security Conducted every 5 years to obtain data on Administration. The largest firms are number of establishments, number of included monthly; a sample of employees, total payroll size, total sales, others is included every three months on a and other industry-specific statistics. In rotating basis. Prior to August 1993, an 1992, the universe was all employer estab- area sample was used to account for lishments primarily engaged in wholesale employer births and nonemployers in retail trade, and employer and nonemployer trade. To account for sales and inventories establishments in retail trade or service for employer births and nonemployers, a industries. benchmarking operation was initiated. Type of Data Collection Operation: All Data Collection and Imputation Proce- wholesale firms with paid employees sur- dures: Data are collected by mail question- veyed; all retail and service large employer naire with telephone followups for firms surveyed (i.e. all employer firms nonrespondents. Imputation made for each above the payroll size cutoff established to nonresponse item and each item failing edit separate large from small employers) plus a checks. 10-percent sample of smaller employer Estimates of Sampling Error: For the 1995 firms. Firms with no employees were not monthly surveys, CV’s are about 0.7% for required to file a census return. estimated total retail sales, 1.3% for whole- Data Collection and Imputation Proce- sale sales, 1.8% for wholesale inventories. dures: Mail questionnaire is utilized with For dollar volume of receipts, CV’s from the both mail and telephone followups for non- Service Annual Survey vary by kind of busi- respondents. Data for nonrespondents and ness and range between 1.5% to 15.0%. ‘‘nonselected’’ small employer firms in retail Sampling errors are shown in monthly publi- trade and service industries are obtained cations. from administrative records of IRS and the Other (nonsampling) Errors: Imputation Social Security Administration. rates are about 18% to 23% for monthly retail sales, 27% to 32% for retail invento- Estimates of Sampling Error: Not ries, 23% to 28% for wholesale sales, about applicable. 25% to 30% for monthly wholesale invento- Other (nonsampling) Errors: Response rate ries and 14% for the Service Annual Sur- in 1992 of 88% for single establishment vey. firms; 89% for multi-establishment firms. Sources of Additional Material: U.S. Item response ranged from 60% to 90% Bureau of the Census, Current Business with higher rates for less detailed questions. Reports, Monthly Retail Trade, Monthly Sources of Additional Material: U.S. Wholesale Trade, and Service Annual Bureau of the Census, Appendix A of Cen- Survey. sus of Retail Trade; Census of Service Industries; Census of Wholesale Trade; and Section 29. OUTLYING AREAS History of the 1987 Economic Censuses, (See section 1 for information pertaining to April 1992. tables 1347 and 1348, section 2 for table Source and Title: U.S. Bureau of the 1346, section 22 for table 1354, sections 26 Census, Current Business Surveys and 27 for table 1358.)