Most programs at BLS collect multiple items of data from respondents. For example, the Job Openings and Labor Turnover Statistics (JOLTS) program collects employment, hires, and separations. Sometimes respondents may not report all three of these items. Rather than ignoring an incomplete response, BLS uses models to fill in missing items of data. This process of filling in missing items is called imputation.
For more information
If a program uses imputation, that information can be found in the Calculation section of the Handbook of Methods.
In-depth research on imputation
Imputation at three federal statistical agencies: https://www.bls.gov/advisory/fesacp1101703.pdf
Impact of imputation on income estimates: https://www.bls.gov/opub/mlr/2009/08/art3full.pdf
JOLTS item imputation: https://www.bls.gov/osmr/research-papers/2013/pdf/st130120.pdf