Table 1.

Predictors of over/underestimation, United States vs. all other countries

PredictorNonbehavioral (k = 40, n = 566)Behavioral, all (k = 42, n = 608)Biobehavioral (k = 20, n = 308)Behavioral (k = 22, n = 300)
(Intercept)0.42 [0.40, 0.46]0.55 [0.51, 0.56]0.51 [0.47, 0.54]0.57 [0.50, 0.59]
United States vs. rest−0.02 [−0.06, 0.00]0.03 [0.02, 0.06]0.03 [0.00, 0.07]0.04 [0.01, 0.07]
Study size (SE)0.43 [0.27, 0.53]0.11 [0.07, 0.23]0.20 [0.11, 0.31]0.06 [0.01, 0.29]
Pub. order0.02 [0.00, 0.03]0.00 [−0.01, 0.01]0.01 [0.00, 0.05]0.00 [−0.02, 0.01]
USA*SE−0.21 [−0.47, 0.22]−0.19 [−0.31, −0.03]−0.16 [−0.34, 0.12]−0.22 [−0.46, −0.02]
USA*pub. order−0.02 [−0.05, 0.01]0.00 [−0.02, 0.03]−0.02 [−0.06, 0.01]0.01 [−0.02, 0.05]
  • Likelihood of a primary study within a meta-analysis to deviate from the summary effect size (deviation score, i.e., double square root-transformed absolute value of deviation values in Fig. 1), depending on its size (SE), its chronological order of appearance (pub. order) within the meta-analysis (z-scaled by meta-analysis), and origin of corresponding author (United States vs. all other countries). Study effects (upper four rows) are estimated without interaction effects. The latter were estimated in a hierarchically well-formulated model, whose main effects (i.e., the same factors that appear above) are omitted. Studies are classified as in Fig. 1.