Table 1. Definitions of key terms used in this manuscript.
Key term | Definition |
---|---|
Reproducibility | The similarity of outcomes between replicate studies. This can be measured, e.g., by the CI95 of the mean effect size estimates of a sample of replicate studies (depicted by the shaded area in Fig 2A). |
FNR | False negative rate: proportion of true positives that yield negative test outcomes. FNR = false negative ÷ (true positive + false negative). |
FPR | False positive rate: proportion of true negatives that yield positive test outcomes. FPR = false positive ÷ (false positive + true negative). |
DOR | Diagnostic odds ratio: ratio of the odds of the test being positive in the case of a true positive relative to the odds of the test being positive in the case of a true negative. DOR = (true positive ÷ false positive) ÷ (false negative ÷ true negative). |
pc | Coverage probability: the probability with which the CI95 of an effect size estimate includes the true effect size. |
psa | The proportion of studies reporting both a significant effect for α = 0.05 and a CI95 for the effect size estimate that includes the true effect. |
I2 |
I2 is a descriptive statistic of the ratio of excess dispersion to total dispersion. I2 = (Q − df ÷ Q) × 100%, where Q is the weighted sum of squares of study effect sizes and df gives the degrees of freedom. |
Abbreviations: CI95, 95% confidence interval; DOR, diagnostic odds ratio; FNR, false negative rate; FPR, false positive rate; pc, coverage probability.