2 ms·
Because of conditional probabilities, your analysis is wrong. This xkcd explains the issue. Someone else linked to it already, but it exactly describes the pr
by hedora 9y ago
Because of conditional probabilities, your analysis is wrong. This xkcd explains the issue. Someone else linked to it already, but it exactly describes the problem -- with p=0.05 >> 5% of published results are incorrect:
https://xkcd.com/882/ https://xkcd.com/882/
- vacri 9y agoYou're making the same mistake in reverse. The p-value is not a marker of correctness, but of confidence. If something doesn't reach the p-value cutoff, that doesn't mean that the effect doesn't exist, just that it didn't meet the confidence level. A paper could be entirely correct in its hypothesis, yet not meet the confidence level in its experimentation.