Why gig-work prevalence estimates keep moving

Few labor statistics have traveled as widely in public debate as estimates of “gig work” prevalence—and few have been as sensitive to questionnaire design. Lawrence Katz and Alan Krueger’s influential analyses of alternative work arrangements helped frame a decade of discussion. Later work, including critiques and replications using different modules and years, showed that headline shares move when questions change, when online samples replace probability surveys, and when researchers redefine what counts as alternative.

The Contingent Worker Supplement (CWS) to the CPS remains a reference point, but its cycles are infrequent and its concepts (contingent work, alternative arrangements, electronically mediated employment) are not interchangeable. A rise or fall between cycles may reflect a real labor-market shift, a wording change, or both. Commentators who collapse those categories into a single “gig economy percentage” import precision the instruments do not offer.

Abraham and coauthors have argued that traditional surveys understate offline informal work as well as online platform work. Other researchers caution against treating every independent contractor as a gig worker; many contractors have long-term business models unrelated to app-mediated task markets. Prevalence therefore depends on the atlas legend: include all non-employee arrangements, or only platform-mediated tasks, or only respondents who call their work a side activity.

Online convenience samples often report higher gig participation than CPS-based estimates. That gap can reflect true differences in who answers online surveys, incentives that attract platform workers, or looser definitions. Without a transparent sampling frame, such figures are poor substitutes for official series—even when they surface useful qualitative patterns.

Cyan Atlas treats contested prevalence numbers as overlapping isolines rather than competing truths. When two studies diverge, the editorial task is to show which population, reference period, and definition each line represents. Declaring a single national gig share without those footnotes overstates what the evidence can support.

The literature reviewed here does not prescribe employment choices and does not claim that any arrangement is desirable or undesirable. It explains why careful researchers keep revising maps of contingent and platform work—and why readers should expect the contours to shift when the survey instrument changes.