The 2021 OPM FEVS Sampling Plan

Survey Population 

The survey population is primarily made up permanent, full- and part-time employees, as well as employees in phased retirement. In 2021, eligibility was expanded to include non-permanent and additional work schedules, but only if the participating agencies opted-in those populations; some elected to include all of them, some elected none, and others allowed some and not others. Political appointees are not eligible to participate. 

Two variables are used to group (or ‘stratify’) the eligible employees into subgroups: (1) organization subgroup (e.g., bureaus, offices) and (2) supervisory status (e.g., non-supervisors, supervisors, and Senior Executive Service). 

Sample Size and Sample Selection 

The total eligible population is 1,695,533 and total sample selected for 2021 is 934,734 employees. For agencies with fewer than 10,000 employees, all eligible employees are included for the survey this year (a census for these agencies). For larger agencies/departments, sample sizes were calculated in two steps. First a statistical formula based on probability calculated the number of completed surveys required to accurately reflect the population. To account for non-response, sample sizes were inflated on an agencywide basis so the expected number of completed surveys returned would meet or exceed the sample size calculated in step one. Sample sizes were inflated based on agency response rates to the 2020 OPM FEVS with the minimum inflation rate based on the Government-wide response rate. Agencies with response rates below the Government-wide rate received a larger inflation to account for expected higher rate of non-response. 

If an agency’s inflated sample (described in the above paragraph) was more than three-quarters of the agency, the agency was changed to a full census. In the remaining large agencies/departments, sample sizes were adjusted so that each agency/department had an overall sample of approximately fifty percent of the agency/department population. It should be noted that whether a random sample or a census was taken, most, if not all, supervisors and executives in the agencies were selected to ensure a sufficient enough number of returned completed surveys. This is because they constitute a much smaller proportion of the population, and a random sample would create issues with underrepresentation. This disproportionate sampling is corrected at the end of the survey when we weight the final data to produce results that are representative of the whole population and not just of those who responded.