Main effects on trait anxiety (Model 1). A mixed effects model was run to predict trait anxiety scores from 11 variables, all competing for variance: (A) study context (behaviour vs fMRI), (B) age, (C) whether the study was part of anxiety research, (D) gender, whether the study involved (E) a stressor, (F) a drug administration procedure, (G) psychiatric screening, (H) whether data were provided after participant exclusion, (I) sample type, (J) study duration in minutes and (K) pay rate converted to USD per hour. Effects of categorical factors (A, C–I) are shown as box plots of the raw data; the blue dots and numbers represent the marginal means predicted from the model. Effects of continuous variables (B, J, K) are shown as scatter plots of trait anxiety as a function of the variable (dots: raw data; line: effect of the variable predicted by the model). The effects of study context, age and anxiety research (A–C) were found to be significant both in the mixed effects model and using Bayesian tests (*P < 0.001 and BF10 > 100).
