How surveys count jobs beyond a primary payroll

Statistical agencies have long distinguished a person’s main job from additional paid work, but the boundary is not trivial. In the U.S. Current Population Survey (CPS), multiple jobholding is typically identified when a respondent reports more than one job during the reference period. The published rate therefore depends on how “job” is framed, which week is referenced, and whether informal or intermittent work is recalled at all.

Katharine Abraham and James Haltiwanger, writing with colleagues on measurement of nonstandard and contract work, have emphasized that household surveys can miss short spells and informal arrangements that never feel like a second “job” to the respondent. The Bureau of Labor Statistics (BLS) multiple-jobholding series remains a core public benchmark, yet BLS documentation itself notes definitional limits: unpaid family work, certain unpaid activities, and some contingent arrangements sit outside the classic second-job count.

Time-use data add another contour. The American Time Use Survey (ATUS) records minutes allocated to work activities, which can reveal secondary work that does not appear as a named second employer in a labor-force module. Researchers comparing CPS employment questions with time diaries have argued that diary evidence sometimes shows more fragmented paid activity than a single employment status question captures. That does not automatically mean the CPS is “wrong”; it means the instruments answer different questions—one about job holding as understood by respondents, another about minutes of paid activity.

International surveys raise parallel issues. Modules on secondary employment in European labour force surveys vary in whether they ask about all paid activities in a reference week or only about jobs the respondent considers ongoing. When platforms mediate short tasks, a respondent may describe the activity as “gigs” rather than “a second job,” which can suppress reporting even when earnings occur.

Methodological critiques focus on three recurring risks. First, reference-period mismatch: a one-week window understates intermittent secondary work that occurs monthly. Second, social desirability and fatigue: long questionnaires may truncate follow-ups about additional work. Third, classification: independent contracting, on-call shifts, and platform tasks may be coded as self-employment on a main job rather than as multiple jobholding, depending on skip patterns.

Cyan Atlas treats multiple-jobholding statistics as maps with known projection systems. Published BLS rates are useful for tracking direction over time under a stable definition. They are less reliable as a census of every paid hour outside a primary payroll. Readers comparing a headline multiple-jobholding rate with tax-based counts of platform payers should expect divergence: the instruments observe different populations and different notions of what counts as a job.

None of this literature functions as guidance for individual employment decisions. It documents how measurement choices shape the public picture of work beyond a primary job—and why two careful studies can report different contours of the same landscape.