Table 4. Results of the bare-bones meta-analysis of the success bootstrapping organized by type of evaluation criterion.
| Evaluation criteria | k | N | Δ | SDΔ | 95% CI | 80% CI | Q | I2(%) | τ2 | 75% |
|---|---|---|---|---|---|---|---|---|---|---|
| Subjective | 4 | 76 | .03 | .00 | -.19 - .25 | .03 - .03 | .60 n.s. | 0.00 | 0.00 | 520 |
| Publ. bias | +2 | 81 | .02 | .00 | -.16 - .06 | .02 - .02 | 44.41*** | 88.7 | 0.01 | > 10,000 |
| Objective | 33 | 857 | .08 | .00 | .01 - .14 | .08 - .08 | 4.78 n.s. | 0.00 | 0.01 | 778 |
| Publ. bias | +9 | 1,020 | .10 | .00 | .06 - .12 | .10 - .10 | 216*** | 81.1 | 0.00 | 639 |
| Test | 15 | 177 | .07 | .00 | -.08 - .21 | .07 - .07 | 8.68n.s. | 0.00 | 0.00 | 197 |
| Publ. bias | +3 | 330 | -.01 | .01 | -.12 - .09 | -.14 - .11 | 149.33*** | 88.6 | 0.03 | 86.14 |
k = number of judgment tasks;
N = number of success indices;
Δ = the success of bootstrapping (see Eq 2);
SDΔ = standard deviation of true score correlation; 95% CI = confidence interval; 80% CI = 80% credibility interval including lower 10% of the true score and the upper 10% of the true score; 75% = percent variance in observed correlation attributable to all artifacts; Publ. bias = publication bias-corrected estimation by the trim-and-fill method (see [63]); + = the number of missing tasks indicated by the trim-and-fill method.