The Number of Senders and Total Judgments Matter More Than Sample Size in Deception-Detection Experiments.

Hundreds of experiments have examined people's ability to distinguish truths from lies. Meta-analyses suggest that the findings from larger scale experiments converge and that findings discrepant from the meta-analytic average of 54% occur in only smaller experiments. Study size (number of data points, or total number of judgments) is a joint function of the sample size and the number of judgments per research participant. Furthermore, because senders vary more than judges, experiments involving few senders may not be replicable. A number of simulations are reported in which the sample size, the number of unique senders, and the number of judgments per research participant are varied. The findings demonstrate that stability is more a function of the number of judgments than the sample size and that experiments involving too few senders risk idiosyncratic findings that are less likely to be replicable. Implications for research design are discussed.