list-experiment

A list experiment estimates sensitive topic prevalence by asking respondents to count matching items rather than directly endorsing them. This reduces direct-response bias while requiring diagnostic checks to correct nonstrategic misreporting.

List experiment design variants include conventional single lists, placebo lists with known baseline items, and double list experiments (DLE) with carryover diagnostics. Placebo designs help identify nonstrategic misreporting and improve prevalence estimation accuracy.

Power analysis for list experiments uses simulation-based methods to estimate the detectable prevalence effect size given a sample. This Skill provides power simulation guidance alongside design diagnostics to ensure your survey is adequately powered before data collection.

Yes, list experiment diagnostics include No False Positives (NFC) and No Design Effect (NDE) tests. These diagnostic checks are integrated with the R list package to validate estimator assumptions and detect nonstrategic misreporting in your survey data.

You should not use a list experiment when sensitive topic prevalence is expected to be extremely high or low, as the difference-in-means estimator loses precision. Additionally, if design diagnostics indicate nonstrategic misreporting, alternative survey techniques may be more appropriate.

To estimate sensitive prevalence in R, use the list package with a three-control-item list design and one sensitive item. Apply estimators like difference-in-means, NLSreg, or MLreg, then run combined estimator diagnostics and placebo checks to validate results.