The Current Population Survey (CPS) is administered to a scientifically selected multistage probability-based sample of households designed to represent the civilian noninstitutional population of each state (and the District of Columbia) and the United States as a whole. The sample is made up of addresses, not the names of people or families at those addresses. The sample size is determined by specific criteria that ensure the statistical reliability of the unemployment rate at the national and state level.
The current sample design was developed after the 2010 decennial census and was implemented beginning in 2014. This sample will be used until a replacement sample is developed after the next decennial census.
The CPS starts with a probability sample of about 74,000 assigned housing units each month. Approximately 62,000 housing units are eligible for interview each month, after excluding those which are destroyed, vacant, converted to nonresidential use, contain only people whose usual place of residence is elsewhere, or are ineligible for other reasons. Interviews are completed for about 54,000 housing units. In a given month, about 13 percent of housing units are not interviewed, because of the temporary absence of the occupants (they might be on vacation, for example), other failures to make contact after repeated attempts, inability of people contacted to respond, unavailability for other reasons, and refusals to cooperate. Information is obtained each month on about 105,000 people ages 16 and older.
In developing the survey sample, a number of criteria are considered, including an acceptable degree of statistical reliability for labor force estimates for all states, even those with the smallest populations. The reliability of estimates for a wide range of national labor force characteristics also is considered and optimized, within budgetary limitations. The reliability criteria of the sample design are explained in the state-based design section of Redesign of the sample for the Current Population Survey.
The entire area of the United States is divided into 1,987 Primary Sampling Units (PSUs). Each PSU consists of a county or a group of contiguous counties and is defined within state boundaries.
Metropolitan areas within a state are used as a basis for forming many PSUs. Outside of metropolitan areas, two or more counties normally are combined to form a PSU, except when the geographic area of an individual county is too large. Combining counties to form PSUs provides greater heterogeneity: A typical PSU includes urban and rural residents of varying economic levels and encompasses, to the extent feasible, diverse occupations and industries. Another important consideration is that the PSU be sufficiently compact so that, with a small sample spread throughout it, the PSU can be efficiently canvassed without undue travel costs.
The 1,987 PSUs are grouped into strata within each state. Then, one PSU is selected from each stratum, with the probability of its selection proportional to the population of the PSU. In the first stage of sampling, 852 sample areas are chosen from the strata.
Nationally, there are 506 PSUs in strata by themselves. These strata are self-representing and generally are the most populous PSUs in each state. The remaining 346 strata are formed by combining PSUs that are similar in such characteristics as unemployment, the proportion of housing units with three or more people, the number of people employed in various industries, and average monthly wages for various industries. The single PSU randomly selected from each of these strata is non-self-representing, because it represents not only itself but the entire stratum. The probability of selecting a particular PSU in a non-self-representing stratum is proportional to its population in 2010. For example, within a stratum, the chance that a PSU with a population of 50,000 would be selected for the sample is twice that for a PSU having a population of 25,000.
Because the sample design is state based, the sampling ratio differs by state and depends on the size of the state’s population, as well as both national and state reliability requirements. The state sampling ratios range roughly from 1 in every 200 households to 1 in every 3,000 households.
The sampling ratio occasionally is modified slightly to hold the size of the sample relatively constant, given the overall growth of the population. (The procedure is called “sample maintenance reduction.”) 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 1 in 10 and a state sampling ratio of 1 in 3,000, a within-PSU sampling ratio of 1 in 300 achieves the desired overall ratio of 1 in 3,000 for the stratum. On average, a person in the CPS sample represents about 2,500 people in the population.
The current within-PSU sample design was developed with the use of block-level data from the 2010 census. Normally, census blocks are bounded by streets and other prominent physical features, such as rivers or railroad tracks. County, minor civil division, and census place limits also serve as block boundaries. In cities, blocks can be bounded by four streets and be quite small in land area. In rural areas, blocks can be several square miles in size.
Within each block, housing units are sorted geographically and grouped into clusters. A systematic sample of these clusters is then independently selected from each stratum by using the appropriate within-PSU sampling ratio. The geographic clustering of the sample units reduces travel costs for field representatives.
The vast majority of addresses in the CPS sample are typical housing units, such as single-family homes, apartments, and condominiums. However, the CPS sample does include a relatively small number of housing arrangements known as “group quarters.” Examples of civilian, noninstitutional group quarters that are within the scope of the CPS sample are homeless shelters, group housing for agricultural or mining workers, and convents and monasteries. College dormitories also are noninstitutional group quarters, but beginning in late 2017, they were excluded from the CPS sample for cost and operational considerations. Most group quarters are of an institutional nature (for example, prisons and nursing homes) or military (for example, barracks) and therefore have never been part of the CPS sample.
Part of the sample is changed each month. Each full monthly sample is split into 8 different subsamples called rotation groups. Each rotation group is itself a representative sample of the U.S. population. A given rotation group is interviewed for a total of 8 months, divided into two equal periods. The group is in the sample for 4 consecutive months, leaves the sample during the following 8 months, and then returns for another 4 consecutive months.
Each month, one of the rotation groups is being interviewed for the first time; another is being interviewed for the second time; and so on. (The rotation group in the fifth month in the sample is returning after the 8-month break.) Under this system, 75 percent of the sample is common from month to month and 50 percent is common from the same month a year earlier. This month-to-month and year-to-year overlap in the sample yields better estimates of change and improves the reliability of the data without overburdening sampled households.
A full discussion of the sample design process is available in Chapter 3 of Design and methodology: Current Population Survey, Technical Paper 66. Important updates to the process described in Technical Paper 66 are found in Redesign of the Sample for the Current Population Survey (April 2014).