Hacker News new | ask | show | jobs
by aristophenes 2639 days ago
While this is a very interesting finding, I suspect p-hacking. https://en.wikipedia.org/wiki/Data_dredging

Imagine the awkwardness when someone tries to get funding to confirm.

3 comments

Data dredging is usually establishing a causation or pattern of two loosely correlated variables with no clear undesrstanding of why one have effects on the other. (correlation != causation).

In this particular case they seem to have a clear hypothesis on why there could be a strong link between one and the other: > "Oral sex might induce maternal tolerance towards paternal antigens of the fetus".

Anyway. This doesn't mean that this is a strong study. Just pointing out that at least they seem to have a theory on why oral sex can protect against miscarriages.

We don't know how many different things did they ask for on the survey? Maybe they had 100 different things that they were just spitballing might have some possible effect.
My friend and I were talking about the same thing.

"association became weaker" with a p=.21

doesn't matter - now that this has been reported it is fact