3 ms·

The statistical analysis to be done should be decided before looking at the data. http://www.stat.columbia.edu/~gelman/research/unpublished/p_hacking.pdf http:

• Someone 11y ago
  "Blinding should be pretty easy and cheap - have an assistant cover up the head with a sheet as the measurement is made." But you would need to move the body, or have an experimenter who doesn't know whether he is in Japan (using Japanese living abroad or westerners living in Japan for the experiment is for the follow-up study) Also, but that's relatively minor, covering the head is insufficient. There are lots of other cues, such as skin color, tattoos, missing little fingers, body size, etc. that correlate well with being Japanese as opposed to being a westerner. I would have the assistant cut out everything from the stomach down (even though there is a sphincter, if the size difference is small, specifying where the intestines start will be a challenge) and empty it before handing it over to the experimenter (indirectly, to prevent communication between assistant and experimenters)
  • yummyfajitas 11y ago
    "Missing little fingers"? I've never been to Japan, but as far as I know, most folks aren't in the Yakuza. Also, I wouldn't be super worried about blinding in this case - a length measurement is not particularly subjective. I cannot see how unconscious bias would reasonably add 20-50% (postulated effect size) to intestine length. I'd personally consider the results valid even without blinding, provided a rigid procedure was chosen for measuring intestine length.