P-value myths:
"p-value is the probability that your results occurred by chance"
"p-value is the probability that the null hypothesis is true" or the "false positive risk"
Actual definition: "p-value is the chance of seeing a difference at least as big as we have done, if, in fact, there were no real effect"
graph/quotes via @david_colquhoun https://doi.org/10.1098/rsos.171085
@academicchatter #SciComm #AcademicMastodon #science #AcademicChatter
@erinnacland @david_colquhoun @academicchatter
I do so wish that all science Ph.D. programs, at least, required a rigorous but highly applied statistics course (including data sets with a realistically complex structure) so that we would stop seeing literature filled with bad inferences from improperly-used statistics.
"Condition A has p < 0.05 compared to control, but condition B doesn't, so we can conclude that A and B are different" is a really common and wrong inference.
(Amusingly, or perhaps depressingly, ChatGPT does better than a lot of authors of Science and Nature articles.)
