Personnel selection · Research companion

The evidence

The figures from the manuscript, with plain-language reading guides. Study 1 is a stakes-moderated meta-analysis built on the public Stanek & Ones (2023) database; Study 2 is the simulation behind the calculator.

1 · Only the faked trait moves

Forest plot of meta-analytic correlations between each Big Five trait and general mental ability, split by sample type: applicants, incumbents, students, community, and military. For Conscientiousness the applicant estimate is negative while the incumbent estimate is positive; for the other traits applicant estimates resemble the other sample types.
How to read it: each row is a meta-analytic estimate of the correlation between a Big Five trait and cognitive ability for one kind of sample; horizontal bars are 95% confidence intervals. Applicants (red) are people tested with a job on the line. For four of the five traits, applicants look like everyone else. For Conscientiousness — the trait hiring systems weight most, and the one most worth faking — the applicant estimate sits on the other side of zero from the incumbent estimate. That selective distortion is what a compensatory faking account predicts.

2 · The same contrast, study by study

Study-level forest plot of observed Conscientiousness by GMA correlations in applicant samples, split into ten US samples with an N-weighted mean of +.01 and ten non-US samples with an N-weighted mean of minus .27.
How to read it: every individual applicant sample in the database, one row per study, with its sample size and confidence interval. The negative applicant correlation is not one bad study — and it is not uniform either. US applicant samples average +.01; non-US samples (Australia/New Zealand, Canada, Europe) average −.27. The paper treats that regional split as a finding to explain, not a nuisance: region is confounded with instrument family and national selection practices, so the simulations are run separately per context — which is why the calculator asks you to pick one.

3 · The mechanism, end to end

Three-panel mechanism explainer. Panel A: scatter of Conscientiousness inflation against general mental ability for fakers, slope negative. Panel B: scatter of composite scores as scored against composite scores under honest responding, with faked-in and displaced applicants marked; 31 percent of hires change. Panel C: bar chart of mean true job performance for hires under honest scoring, hires as scored, displaced applicants, and faked-in applicants, with annual dollar values.
How to read it, left to right: (A) fakers add more inflation the lower their ability — the same pattern as the scatter on the overview page, here in the pool calibrated to r = −.27. (B) What that does to the ranking: every red dot above the hiring line is an applicant who got the job only because of faking; every gray dot pushed below it is the honest applicant they displaced. In this pool, 31% of hires change. (C) The bill: the applicants faked into seats truly perform 0.19 SD on the job against 0.51 SD for the people they displaced — roughly $9,100 per swapped hire per year under the standard utility convention (performance SD in dollars = 40% of salary).