Assessing and adjusting for publication bias in the relationship between anxiety and the error-related negativity.

Many clinical neuroscience investigations have suggested that trait anxiety is associated with increased neural reactivity to mistakes in the form of an event-related potential called the error-related negativity (ERN). Several recent meta-analyses indicated that the anxiety-ERN association was of a small-to-medium effect size, however, these prior investigations did not comprehensively adjust effect sizes for publication bias. Here, in an updated meta-analysis (k = 58, N = 3819), we found support for an uncorrected effect size of r = -0.19, and applied a range of methods to test for and correct publication bias (trim-and-fill, PET, PEESE, Peters' test, three-parameter selection model). The majority of bias-correction methods suggested that the correlation between anxiety and the ERN is non-zero, but smaller than the uncorrected effect size (average adjusted effect size: r = -0.12, range: r = -0.05 to -0.18). Moderation analyses also revealed more robust effects for clinical anxiety and anxious samples characterised by worry, however, it should be noted that these larger effects were also associated with elevated indicators of publication bias relative to the overall analysis. Mixed anxiety and sub-clinical anxiety were not associated with the amplitude of the ERN. Our results suggest that the anxiety-ERN relationship survives multiple corrections for publication bias, albeit not among all sub-types and populations of anxiety. Nevertheless, only 50% of the studies included in our analysis reported significant results, indicating that future research exploring the anxiety-ERN relationship would benefit from increased statistical power.