Originally posted by perrymason
View Post
I was, if anything, being rather kind by choosing those categories, which in any case subsume all points in between. In other words, the superset "neck and above" includes the subsets "neck, chin, nose, eyes, forehead, over-head"; just as the superset "chest and below" includes the subsets "First rib, Second rib (etc)... Stomach... Naughty-bits". Taking the latter approach might have come up with the following scores, for example:
1st-3rd rib: 2
4th-6th rib: 3
7th-9th rib: 3
10th-12th: 3
Stomach: 1
Navel: 1
Groin: 0
Note that I've grouped the ribs to give a rough parity in the spans involved. The total still adds up to 13, but compare that with how the 15 "neck and above" sample breaks down:
Overhead: 2
Forehead/top: 5
Nose/eye: 6
Neck/chin: 2
...all of a sudden, we're (a) looking at a more complex statistical test, because of the greater number of categories; (b) there might actually be a statistically significant tendency for people to write in a region between the nose and the top of head.
I can't make that claim, because I didn't test for that - mainly because we didn't have sufficiently fine-grained data, just "chest", and also because I made up the "rib→groin" data purely by way of illustration.
As it stands, there was no statistically significant difference between the two (coarse-grained) groups "neck/above" and "chest/below", at least not in the case of this experiment, for which my gratitude goes to Michael once again.
Comment