Fig 6. Summary of all the replication analyses (with exception of the replication bayes factor).
The first two rows show the partial correlation (attempting to adjust for Session order) between the AB magnitude change scores (tDCS—baseline) in the anodal and cathodal sessions, for study 1 [20] (in yellow) and the present study (in blue). The first is significantly negative, the second is slightly positive and not significant, because its 95% confidence interval (CI) overlaps with zero. The third row shows the 90% CI around the correlation in the present study. Because this interval does not overlap with the “small telescopes” effect size, (indicated by the x: −r33% = -.27), this correlation is significantly smaller. The fourth row shows the 95% prediction interval (PI) around the correlation in [20]. Because this interval does not overlap with the correlation in the present study, both correlations are not consistent. The final two rows show the overall effect when the two correlations are meta-analyzed, and when one correlation is computed over the pooled data from both studies. Neither are significant (95% CI overlaps with zero). Thus, our replication analyses all suggest that we failed to replicate London & Slagter [20], and when the results are examined in combination, no evidence in support of a negative relationship between the AB magnitude change scores (tDCS—baseline) in the anodal and cathodal sessions is obtained.