Stub 03 · Survey design
Katz and Krueger: survey design when alternative arrangements reshape measured employment
Lawrence F. Katz and Alan B. Krueger’s work on alternative work arrangements—especially papers that revisit Contingent Worker Supplement concepts with independent survey instruments—illustrates a recurring measurement tension: when employment relationships diversify, standard household questions may understate or misclassify nontraditional earnings unless the questionnaire opens explicit gates for those arrangements.
This stub reviews design lessons from that literature for readers who track how secondary and nontraditional earnings enter measured employment. It is not a biography, not a rating of any labor market, and not advice about individual work choices.
Why a parallel instrument was needed
Official CWS waves are infrequent. Researchers who wanted a more recent snapshot sometimes fielded their own surveys that mirrored CWS categories—independent contractors, on-call workers, temporary-help agency workers and contract-firm workers—while documenting differences in sampling frame, mode and response rates. Those design choices affect whether a rise in alternative arrangements is read as an economic shift or as an artifact of who answered which questions.
Katz and Krueger’s RAND-American Life Panel work (around 2015) and related discussion papers are often cited for estimates of alternative-arrangement prevalence relative to earlier BLS CWS benchmarks. Subsequent debate—including responses that emphasize the 2017 CWS results—shows that prevalence estimates remain contested when instruments, weights and reference periods differ. Contested evidence belongs on the same page as the citation, not in a footnote that softens a headline claim.
Secondary earnings and the main-job bias
Alternative-arrangement modules typically focus on the nature of the main job. Secondary jobs, informal tasks and digitally mediated gigs that respondents do not classify as “jobs” can remain outside the punched stub even when the main-job classification is carefully designed. Survey designers therefore face a trade-off: longer batteries improve coverage but raise burden and nonresponse.
Proxy respondents add another layer. A household member answering for another adult may know the primary employer more reliably than irregular side work. Instruments that rely heavily on proxies inherit that asymmetry. Self-response panels can reduce it for some groups while introducing panel conditioning and attrition bias of their own.
What the critique implies for reading tables
When a table reports the share of workers in alternative arrangements, the reader needs the definition, the reference week or month, whether multiple arrangements can be coded, and whether unpaid family work or unpaid waiting is excluded. Two papers that use the phrase “gig” may still measure different gates.
Comparison with administrative tax data—again following themes in Abraham’s survey–admin gap research—can show that nontraditional earnings appear more frequently in tax records than in household interviews. That gap does not by itself prove which source is “correct”; it shows that instruments train different openings onto the same underlying activity.
What this note does not claim
This sheet describes survey design and published debate. It does not assert a single correct national rate of alternative arrangements, prescribe how individuals should classify their own work or forecast income. Ridge Pass does not publish testimonials or success narratives. Attribution belongs to Katz, Krueger, BLS and the authors who engage their estimates.