Current Population Survey (CPS).
Excluded are:
Data on members of the resident armed forces are added to the CPS estimates to derive estimates of the "total employed", "labor force", and "non-institutional population".
The survey is monthly. It began in March 1940.
The calendar week (Sunday through Saturday) which includes the 12th of the month (known as the survey week).
The survey provides information on the following topics: employment, unemployment, hours of work, wages (median usual weekly earnings), income (data on income for the previous year are collected each March in a supplement to the regular questionnaire), duration of employment (data on job and occupational tenure are collected in occasional supplements to the regular questionnaire), duration of unemployment, discouraged workers, occasional workers (who are counted if they are working during the survey week), industry, occupation, status in employment, level of education (in March of each year) and usual activity.
Information on the informal sector is provided to the extent that survey respondents report on their activities. As regards underemployment, the only data available are statistics on persons working part time for economic reasons.
Included in the total are:
Each employed person is counted only once. Those who held more than one job are counted in the job at which they worked the greatest number of hours during the survey week.
Excluded are persons whose only activity consisted of work around the house (painting, repairing, or own housework); volunteer work for religious, charitable and similar organisations; and unpaid apprentices and trainees. Those are considered as not in the labour force.
Persons who were waiting to be recalled to a job from which they had been laid off (with or without pay), or were waiting to start a new job within 30 days need not be looking for work to be classified as unemployed."
Included in the unemployed are full- and part-time students seeking full- or part-time work, provided they are currently available for work (if they are seeking work for some future date, such as the summer months, they are considered as inactive).
The "specific efforts to find employment" include any of the following steps: registering at a public or private employment agency; writing letters of application; seeking assistance from friends or relatives; placing or answering ads; being on a union or professional register; obtaining assistance from a community organisation; waiting at a designated labour pickup point, etc.
All civilians aged 16 years of age and over who are not classified as employed or unemployed are defined as "not in the labour force". These persons are further classified as "engaged in own housework", "in school", "unable to work" because of long-term physical or mental illness, "retired" and "other". The "other" group includes the voluntarily idle, seasonal workers for whom the survey week fell in an "off" season and who were not reported as looking for work, and persons who did not look for work because they believed that no jobs were available because of personal factors (age, lack of education or training, etc.) or because of the prevailing job market situation.
Both employed and unemployed persons are classified by industry, occupation, status in employment and level of education. Unemployed are classified by their industry, occupation and status in their previous job.
The classification system is not linked to the International Classification of Status in Employment (ICSE).
This classification is not linked to the International Standard Classification of Education (ISCED).
The CPS sample was last redesigned in April 1984-July 1985 to incorporate data from the 1980 Census into the sampling frame, as is done after every decennial census. At the same time, the structure of the sample design was changed in order to provide more reliable subnational (State and local) estimates. It was redesigned as 51 separate subsamples, one for each of the States plus the District of Columbia.
The CPS sample is located in 729 areas comprising over 1,000 counties and independent cities.
Information is obtained for about 105,000 individuals aged 16 years and over each month. The sampling fraction is about one in 1,750 in terms of both households and persons.
The full sample is based on a stratified multi-stage design.
Selection of sample areas: The entire area of the United States, consisting of 3,137 counties and independent cities, is divided into 1,973 primary sampling units (PSUs). With some minor exceptions, a PSU consists of a county or a number of contiguous counties. Metropolitan areas within a State are used as a basis for forming PSUs. Outside metropolitan areas, counties normally are combined, except where the geographic area of the sample county is very large. A typical PSU includes urban and rural residents of both high and low economic levels and encompasses, to the extent feasible, diverse occupations and industries.
The 1,973 PSUs are grouped into strata within each State. Then one PSU is selected from each stratum with probability of selection proportionate to the population size of the PSU. PSUs in strata by themselves are self-representing and generally are the most populated PSUs in each State. Other strata are formed by combining PSUs that are similar in such characteristics as population growth, proportions of Blacks and Hispanics, population distribution by occupation, industry, age, and sex. PSUs selected from these strata are non-self-representing, since each one chosen represents the entire stratum.
Selection of sample households: The sample design is essentially State based and the State sampling ratios range roughly from one in every 200 households to one in every 2,500 households in each stratum of the State. The sampling ratio used within a sample PSU depends on the probability of selection of the PSU and the sampling ratio for the State. In a sample PSU with a probability of selection of one in 10 in a State with a sampling ratio of one in 2,500, the within-PSU sampling ratio that results is one in 250, thereby achieving the desired ratio of one in 2,500 for the stratum.
Within each designated PSU, several steps are involved in selecting the housing units to be enumerated. First, the 1980 census enumeration districts (EDs) are ordered so that the sample would reflect the demographic and residential characteristics of the PSU. Within each ED, the housing units are sorted geographically and are grouped into clusters of approximately four housing units. Then a systematic sample of these clusters of housing units is selected.
The identification of the sample housing units within an ED is made wherever possible from the list of ED addresses compiled during the 1980 census, or, if the addresses are incomplete or inadequate, by area sampling methods. The address lists are used in about two-thirds of the cases, primarily in urban areas, and area sampling in applied in the remainder. The address list sample is supplemented by a selection of the appropriate proportion of units newly constructed in the PSU since the census date and obtained from records of building permits in that area.
In those EDs where area sampling methods are used, mainly rural areas, the EDs are subdivided into segments, i.e. small land areas having well-defined boundaries and, in general, an expected size of about eight to 12 housing units or other living quarters. For each subdivided ED, one segment is designated for the sample. When a selected segment contains about four households, all units are included in the sample. When the size of the segment is several times four units, the interviewer uses a systematic sampling pattern to achieve the equivalent of a four-household cluster which is canvassed completely. The remaining housing units in the segment are then available for further samples.
Under this system, 75 per cent of the sample segments are common from month to month and 50 per cent from year to year.
At each monthly visit, a questionnaire is completed for each household member aged 16 and over, and the interviewer asks a series of standard questions on economic activity during the preceding week (containing the 12th of the month). Additional questions are asked each month to help clarify the information on labour force status.
The information obtained for each person in the sample is subjected to an edit by the regional offices of the Bureau of the Census. The field edit serves to catch omissions, inconsistencies, illegible entries and errors at the point where correction is possible.
Although the CPS interviewers are chiefly part-time workers, most have had several years of experience on the survey. They are given intensive training when first recruited and further training each month before the survey. Through editing of their completed questionnaires, repeated observation during enumeration and a systematic reinterview of part of their assignments by the field supervisory staff, the work of the interviewers is monitored and errors or deficiencies are brought directly to their attention.
After the field edit, the questionnaires are forwarded to the Jeffersonville, Indiana, office of the Bureau of the Census by the end of the week after enumeration. The raw data are transferred to computer tape and transmitted to the computers in the Bureau of the Census' Washington office where they are checked for completeness and consistency.
The estimation procedure involves weighting the data from each sample person by the inverse of the probability of the person being in the sample. This gives a rough measure of the number of actual persons that the sample person represents. Since 1985, almost all sample persons within the same State have the same probability of selection. These estimates are then adjusted for non-interviews and the ratio estimation procedure is applied.
Non-interview adjustment: The weights for all interviewed households are adjusted to the extent needed to account for occupied sample households for which no information was obtained because of absence, impassable roads, refusals, or unavailability of the respondents for other reasons. This non-interview adjustment is made separately by combinations of similar sample areas that are not necessarily contained within a State. Within each combination of sample areas there is a further breakdown by residence.
First-stage ratio estimate: In the CPS, a portion of the 729 sample areas is chosen to represent other areas not in the sample; the remainder of the sample areas represent only themselves. The first-stage ratio estimation procedure, designed to reduce the portion of the variance resulting from requiring sample areas to represent non-sample areas, is not applied to sample areas which represent only themselves. The adjustment is made at the State level for each of the 43 States which contain non-sample areas by race cells of black and non-black. The procedure corrects for differences that existed in each cell at the time of the 1980 census between the race distribution of the population in sample areas and the known race distribution of the State.
Second-stage ratio estimate: The sample proportions of persons in specific categories are adjusted to the distribution of independent current estimates of the civilian non-institutional population in the same categories. First, the sample estimates are adjusted within each State and the District of Columbia to an independent control for the population 16 years and over. The second step involves an adjustment by Hispanic origin to a national estimate for eight age-sex categories by Hispanic and non-Hispanic. In the third step, a national adjustment is made by the race categories of white, black and other races to independent estimates by age and sex. The entire second-stage procedure is iterated six times, each time beginning at the weights developed the previous time. The controls by State for the civilian non-institutional population aged 16 years and over are an arithmetic extrapolation of the trend in the growth of this segment of the population using the two most recent July 1st estimates, adjusted as a last step to a current estimate of the US population of this group.
Beginning January 1985, the "inflation-deflation" method of deriving independent population controls was reintroduced into the CPS estimation procedure (it had been temporarily discontinued during the period January 1982 to December 1984). With this method, the independent controls are prepared by inflating the 1980 census counts to include estimated undercounts by age, sex and race, aging this population forward to each subsequent month and later age by adding births and net migration, and subtracting deaths. These post-censal population estimates are then deflated to census level to reflect the pattern of net undercount in the most recent census by age, sex and race. Estimates on vital statistics for the resident population, death of military personnel overseas, institutional populations, and net civilian immigration, are obtained from various independent administrative sources (Department of Defence, Immigration and Naturalization Service, etc.) and from computations made on the basis of the 1980 census total population.
Beginning January 1986, two changes were introduced in the estimation of the independent population controls. An explicit allowance for net undocumented immigration since 1 April 1980 was added to the estimated level of legal immigration. In addition, an increase in the estimate of emigration of legal foreign-born residents has been incorporated into the postcensal population estimates since 1980.
Composite estimate procedure: In deriving statistics for a given month, a composite estimating procedure is used which takes account of net changes from the previous month for continuing parts of the sample (75 per cent) as well as the sample results for the current month. Also included is an additional term which is an estimate of the net difference between incoming and continuing parts of the current month's sample.
| Total | Males | Females | |
|---|---|---|---|
| Employment (size of est.) | 117,113,000 | 64,246,000 | 52,866,000 |
| Standard error | 282,000 | 194,000 | 220,000 |
| Unemployment (size of est.) | 6,546,000 | 3,593,000 | 2,953,000 |
| Standard error | 128,000 | 96,000 | 87,000 |
An adjustment is made in weights for interviewed households to account for non-interviews (see under Weighting the sample). In addition, for a relatively few households, some of the information is omitted because of lack of knowledge on the part of the respondent or because of interviewer error. In processing the completed questionnaires, entries usually are supplied for omitted items on the basis of the distributions of these items for persons of similar characteristics.
The official seasonal adjustment procedure for the labour force series is the X-11 ARIMA program, which is an adaptation of the standard ratio-to-moving-average method and provides for "moving"adjustment factors to take account of changing seasonal patterns.
209 different major series are directly adjusted twice each year, after receipt of June and December data, with six months of projected factors drawn from each run and historical revisions drawn from the end-of-year run. In addition, several hundred seasonally adjusted series are produced by arithmetically combining or aggregating the directly adjusted series with each other, or, in some cases, with series on population or resident armed forces levels, which are not seasonally adjusted because they are not considered to have any significant seasonal variation.
Classification errors may be particularly large in the case of persons with marginal attachments to the labour force. These errors may be caused by interviewers, respondents, or both, or may arise from faulty questionnaire design. Interviewers may not always ask the questions in the prescribed fashion. To the extent that varying the wording of the question causes differences in response, errors or lack of uniformity in the statistics may result. Similarly the data are limited by the adequacy of the information possessed by the respondent and the willingness to report accurately.
Non-sampling errors occurring in the interview phase of the survey have been studied by means of a reinterview program. This program is used to estimate various sources of error as well as to evaluate and control the work of the interviewers. A random sample of each interviewer's work is inspected through reinterview at regular intervals. The results indicate that the data published from the CPS are subject to moderate systematic biases.
Although these is a quality control program on coding and a close control on all other phases of processing and tabulation of the returns, some processing errors are almost inevitable. However, the net error arising from processing in considered to be fairly negligible.
The national sample survey of households, called the Monthly Report of Unemployment, was initiated by the Works Progress Administration in 1940. It was transferred to the Bureau of the Census in late 1942 and became the Monthly Report on the Labor Force. In 1948 its title was changed to the present Current Population Survey. In 1959, responsibility for analyzing and publishing the CPS labour force data was transferred to the Bureau of Labor Statistics, although the Bureau of the Census continues to collect and tabulate the statistics.
The lower age limit for official statistics on the labour force, employment and unemployment was raised from 14 to 16 years of age in January 1967. Definitional changes were also introduced at that time. Several modifications have been brought to include increased populations from new States and areas (inclusion of Alaska, Hawaii in 1960) and regular adjustments have been made after each of the decennial censuses.
Several modifications were brought to the estimation procedures successively in 1974, 1978, 1979, 1982, 1983, 1985 and 1986.
Beginning January 1983, the occupational and industrial classification systems used in the 1980 census were introduced into the CPS. These systems differ from those developed for the 1970 census, which were used in the CPS from January 1971 through December 1982.
Various changes have been made in the design of the CPS sample. One major change made after every decennial census is to change the sample design to make use of the recently collected census materials. Also, the number of sample areas and the number of sample persons have been regularly increased to take account of new States, growing population groups, new construction units and mobile homes, etc.
For survey results, see:
US Department of Labor, Bureau of Labor Statistics: "News - The Employment Situation" (monthly press release) (Washington, DC). Data are published three weeks after the reference period.
idem: "Employment and Earnings" (monthly) (ibid.). Data are published about five weeks after the reference period.
idem: "Monthly Labor Review" (ibid).
For methodological information, see:
idem: "Employment and Earnings" (Explanatory Notes) (monthly) (ibid.).
idem: "Employment and Earnings" ("Changes in the Estimation Procedure in the Current Population Survey Beginning in January 1986") (ibid., February 1986).
idem: "BLS Handbook of Methods", Bulletin 2285, Chapter I (ibid., April 1988).
idem: "The Current Population Survey: A Historical Perspective and BLS' Role" by John E. Bregger in "Monthly Labor Review" (pp. 8-14) (ibid., June 1984).
Bureau of the Census: "The Current Population Survey: Design and Methodology", Technical Paper No. 40 (ibid., January 1978).
In addition to regular publications, the labour force data are available on both diskettes and magnetic tapes. Non-published but regularly tabulated results can be made readily available upon request. Special data that is not regularly tabulated can be generated through the use of individual record (micro) tapes.