Theory 90/Selection & Bias/Meta-analysis
Effect Sizes Inflate
The claim
Meta-analytic effect sizes are inflated in proportion to the ratio of publication bias to study precision, so a field of small studies overstates its effect.
01Mechanism
Studies with null results are less likely to be published; the surviving literature over-represents significant findings.
02Prediction
Inflation should track the ratio across fields.
03Refutation
Effect sizes stable across precision would remove the bias.