Screeners should not reveal the qualifying answer
Aliases: masked screener · demand characteristics · screener bias
What it is
Eligibility screener masking prevents applicants from easily inferring which response earns admission. Asking “Do you use this product at least three times every day?” exposes both the target behavior and threshold, inviting exaggeration from people seeking compensation or access. Masking does not conceal study risks. It embeds eligibility in neutral questions with plausible nonqualifying options so strategic responding is harder.
Why it happens
Screening is a classification task with an explicit reward. Once applicants infer the rule, answers measure experience plus guesses about the study, reconstructed memory, and impression management. Yes/no items are easy to game, and visible thresholds supply a reusable script. Frequency ranges, recent concrete episodes, cross-checks, and unrelated decoy categories increase the cost of fabricating a coherent account, although none proves identity.
Studying it
Begin from the eligibility construct and distinguish mandatory conditions, quota variables, and descriptive variables. Cognitive interviews can reveal interpretation and whether the target is obvious. Compare wording variants on pass rates, later no-shows, in-study behavior, and inconsistency. After consent, high-risk recruitment may verify essential experience through product records, artifacts, or a brief follow-up while collecting only necessary data. Handle repeat or implausibly fast applicants with prespecified rules.
Where it stops holding
Complete concealment is neither possible nor always ethical. Research involving children, health, occupational risk, or sensitive data must disclose participation requirements and risks. Rare populations may need direct self-identification, while excessive decoys add burden and can exclude intended participants. Inconsistency may reflect memory, terminology, or time-window interpretation rather than deception.
Applying it
- Replace threshold-revealing yes/no items with neutral ranges covering the full behavior distribution.
- Verify experience through a recent episode and optional cross-check rather than unnecessary identity evidence.
- During pretesting, ask who respondents think qualifies; rewrite items when the admission answer is obvious.
- Monitor pass rate, later inconsistency, and no-shows. Remove masking complexity that reduces enrollment without improving evidence quality.