@mvanassen.bsky.social
Most important is that a rejection of nullhypothesis of the pcurve analysis does not tell you WHY all true effect sizes do not all appear to be zero... this can be for different reasons i-iii
October 30, 2025 at 1:56 PM
Most important is that a rejection of nullhypothesis of the pcurve analysis does not tell you WHY all true effect sizes do not all appear to be zero... this can be for different reasons i-iii
how large this probability/power is depends on the strength of the effect size of statistic in (i), the amount of p-hacking in (ii), amount of effect size heterogeneity in (iii), number of statistics in the analysis, and relative number of statistics in the p-curve analysis affected by (i)-(iii).
October 30, 2025 at 1:55 PM
how large this probability/power is depends on the strength of the effect size of statistic in (i), the amount of p-hacking in (ii), amount of effect size heterogeneity in (iii), number of statistics in the analysis, and relative number of statistics in the p-curve analysis affected by (i)-(iii).
(i)-(iii) lead to a rejection of the null-hypothesis of the p-curve analysis with a probability larger than alpha (the power of the test)...
October 30, 2025 at 1:54 PM
(i)-(iii) lead to a rejection of the null-hypothesis of the p-curve analysis with a probability larger than alpha (the power of the test)...
The null is false if at least one mu <> 0. This may occur if (i) the null is false for only one statistics (this may be an outlier), (ii) p-hacking or something else affects the distribution of one or more statistics, or (iii) effect size heterogeneity (can be seen as a special case of (i)).
October 30, 2025 at 1:53 PM
The null is false if at least one mu <> 0. This may occur if (i) the null is false for only one statistics (this may be an outlier), (ii) p-hacking or something else affects the distribution of one or more statistics, or (iii) effect size heterogeneity (can be seen as a special case of (i)).
Ha!
This is, fortunately, not rocket science.
P-curve tests the hypothesis that ALL test statistics arise from H0, for instance, mu = 0).
This is, fortunately, not rocket science.
P-curve tests the hypothesis that ALL test statistics arise from H0, for instance, mu = 0).
October 30, 2025 at 1:52 PM
Ha!
This is, fortunately, not rocket science.
P-curve tests the hypothesis that ALL test statistics arise from H0, for instance, mu = 0).
This is, fortunately, not rocket science.
P-curve tests the hypothesis that ALL test statistics arise from H0, for instance, mu = 0).