Question for dopey PhDs

Factual Questions
1

Finishing up my Ed.D. and I do have a master’s in math so get as technical as you want.

I have two lists of rankings that are probably related. As an example:
What do you prefer to play / What do you prefer to watch on TV
Baseball 1 / 2
Basketball 3 / 3
Football 2 / 1

My original thought was to measure the total deviation so in the above example the deviation would be 1+0+1=2. I’m not real happy with this but I’m not really sure what the standard deviation (0.8165) adds to the analysis. Of course, I may just be overthinking the issue.

So how about a matched pair t-test? That would work if I want to see IF the two lists are significantly different, but does the score itself tell me about the amount of difference between the two lists.

My ultimate goal is given a whole group of people, to see if the amount of difference for an individual correlates to another number like beers drunk in a week.

2

Since you only have ranks and not a continuous measure indicating the “absolute” enjoyment of doing/watching, I’d say the obvious answer is a Spearman’s rank correlation coefficient is one piece of data

But I suppose I shouldn’t be answering this question. Although I have a Ph.D., I don’t consider myself a “dopey Ph.D.”

3

Well, we are on the Straight Dope.

But I do wonder why you think this question needs to be answered by someone with a Ph.D. (in any subject, apparently)

4

Quite. I have a Ph.D. and I have no idea about this. It is not at all my field. On the other hand, I am sure plenty of people without doctorates understand enough statistics to be able to help you.

5

Because as of 1986, a Ph.D. is technically a degree in how to conduct research and not a specific field.

Plus whenever I tell a PhD that I’m getting an EdD, I keep getting told that it is not a real doctorate since they had to do real research :rolleyes:

6

This is not for PhDs, its stats 101. The measure of association of choice for two ordinal variables is the Spearman’s rho.