Description Usage Arguments Value See Also Examples
View source: R/pcurve.binomial.tests.R
This function tests the null hypothesis that there are at
least as many p values in the 0.03 - 0.04 bin as in the
0.04 - 0.05 bin. Note that p values of exactly 0.04 are
excluded, because they do not fall in either bin. The test
uses a one-tailed sign test. Significantly more p values in
the smaller bin is consistent with collections of p values
with evidential value. Significantly more p values in the
larger bin is consistent with p-hacking or publication
bias. This is a more sensitive test of p-hacking or
publication bias than the related
binomial.all.test
1 |
p |
a vector of p values between 0.0 and 0.05 (inclusive) |
a list giving the number of p values in each bin, and the p value of the two-tailed sign test.
binomial.all.test
,
binomial.sns.test
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 | # here are some p values you might get from a strong effect size
p <- rexp(1000, 200)
p <- p[p<0.05]
# let's add some you might get from p-hacking and/or publication bias
h <- -1 * rexp(100, 200) + 0.05
h <- h[h>0.00]
p <- c(p, h)
# the binomial.all.test should show significant right skew
# that's expected - it uses all the data from 0.00 to 0.05
binomial.all.test(p)
# the binomial.bias.test is more sensitive to p-hacking and/or
# publication bias - it uses just the data from 0.03 to 0.05
binomial.bias.test(p)
|
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.