Tag Archives: p-hacking

Chocolate causes you to lose weight. Or does it?

So, this person who made a bunch of open access journals look like the fools they are a while back (turns out most don’t actually do peer review – whoops!) just posted about a more recent sting, this time on the science media.

Basically, they carried out a study that showed chocolate helps you lose weight, and reported it as such.

It was then spread far and wide for obvious reasons.

Mainly because chocolate is delicious and most of us want to keep eating it without gaining the weight it seems to drag along with it.

Turns out it was a crappy study, on purpose.

The authors say they purposely engaged in a practice called p-hacking, which is pretty common in science unfortunately. Basically, the researchers measured a whole bunch of other variables too, so by random chance you might expect some of them to LOOK like chocolate affects them.

I read as their main point that science media sensationalizes everything scientists find without really asking how much support there is for the conclusion they are spreading around the world.

The only problem is I keep coming back to their study. They DID find that chocolate helps you lose weight. Their ‘extra’ variables were things like blood protein, sodium, etc., which in my opinion are things that I wouldn’t expect chocolate to affect anyways.

Hmm. So, clearly, science media sensationalize everything without regard to the quality of the research.

But I keep coming back to the important question – can chocolate help me lose weight? (Or rather, can I keep eating lots of chocolate but do something about my ever-growing midsection?)

I’m going to conclude that the authors set up a bunch of ‘straw man’ hypotheses they didn’t really believe in to prove a point, and that they in fact were NOT engaging in p-hacking. Maybe I’m being nitpicky, but I don’t see chocolate affecting sodium as that good of a hypothesis.

BUT on the other hand the main points they made were really good! I keep going around in circles on this…

Whatever. I don’t think I’ll be able to solve this tonight.

I’m gonna go eat some chocolate.

Advertisements