Hey, everybody, I think you know what it's time for . . . .
That’s right-- another episode of Macau's favorite game show...: "Make a prediction, know it all!," or "MAPKIA!"!
To get the technicalities out of the way, here's the posting of the "official statement of contest terms & conditions," as mandated by the Gaming Commission:
I, the host, will identify an empirical question -- or perhaps a set of related questions -- that can be answered with CCP data. Then, you, the players, will make predictions and explain the basis for them. The answer will be posted "tomorrow." The first contestant who makes the right prediction will win a really cool CCP prize (like maybe this or possibly some other equally cool thing), so long as the prediction rests on a cogent theoretical foundation. (Cogency will be judged, of course, by a panel of experts.)
Okay, this is a tricky one!
It’s going to take (a) a Feynmanite/Selbstian level of analytical thought, (b) a Fredrickian resistance to the seductive tug of WEKS, plus (c) a Barry-Bonds-sized dose of political-psychology HGH (& yes former Freud expert & current stats legend Andrew Gelman and Josh " 'Hot Hand Fallay' Fallacy" Miller both remain eligible for this MAPKIA pending their appeals for testing positive in the aftermath of their stunning post “CCP-APPC Political Polarization IQ Test”™ victories).
Let’s start by creating a “political partisanship index.” The recipe for that is as follows:
- Take a left-right political outlook scale formed by standardizing the sum of the sums of responses to conventional 7-point political-party identification and 5-point liberal-conservative ideology survey items. A very nice feature of this approach when one uses it with a nationally representative sample is that “0” is “moderate Independent,” while -1 and +1 SD are “liberal Democrat” and “conservative Republican,” respectively. Scores in the vicinity of -1.8 and +1.8 will be “Extremely liberal, Strong Democrat” and “Extremely conservative, Strong Republican,” respectively. In case you’ve forgotten how nicely this simple scale performs in picking in partisan polarization on contested issues, check out the policy-polarization figure below or watch a re-run of the wildly popular episode on the “CCP-APPC PPQ IQ Test”™).
- Then take the absolute value of the scores on this Left_right scale. The result is a “Partisanship Index” (PI), one that registers the intensity of one’s left-right outlooks without regard to their valence. Thus, if one is either a “liberal Democrat” or a “conservative Republican,” one gets a PI score of “1.0.” If one is either an “Extremely liberal, Strong Democrat” and “Exremely conservative, Strong Republican,” one gets a PI score of 1.8. A milqetoast politically sissy who is a “moderate Independent” will get a score of “0.”
Okay, got that? Good. (If you are curious for what the relationshop between Left_right and PI looks like without smoothing--and why the intercept for zero on y axis is slightly above zero--good for you! Click here).
Now here is the MAPKIA question:
What is the profile of a “super partisan”? On the basis of characteristics like (a) gender, (b) race, (c) income, (d) education, (e) science comprehension (measured by OSI), (f) science curiosity (measured with SCS), (g) religiosity, (h) cultural worldivews (measured with the CCW scales) etc. or appropriate combinations thereof, who is the most partisan “type” of person (i.e., gets the highest PI score) in U.S. society????
You know the rules: don’t just gesture toward an answer in some vague discursive way; be specific, both about what your conjecture and why, and tell me how to test it using the sort of data that typically appears in a CCP data set.
Realize that basically the question is, What's the relationship between the specified characteristics and partisanship? If you want to specificy simple correlations between partisanship and one or more of these attributes or (better still) combinations of them, that's fine!
But if you have some more clever way to specify how the characteristics should be combined into some latent-variable "identity" variable or how the relationship between the characteristics (individually or in combination) should be related to the Partisanship index (in that regard, you might want to check out "yesterday's" post on how science curiosity and science comprehension relate to each other), go for it!
Now, an important proviso: Do not tell me to just jam every one of these characteristics onto the right hand side of a goddam linear regression and “see what comes out statistically significant.” The reason is that the results of such an analysis will be gibberish.
Actually, the R2 will be fine & might be interesting if you want to get an idea of the the upper limit of the possibilities for explaining PI. But the parameter estimates will be meaningless in relation to our task, which is to identify the sorts of real-world people who are super partisans.
And with that . . . mark, get set, MAPKIA!