Sunday, May 25, 2014

No, "tightness" vs "looseness" does not "explain" the political map.

By Hans Noel

A new map is crawling all over my Facebook feed. This one is from a Washington Post write-up of a study by two University of Maryland psychologists. The Post's headline, which overstates what the article argues, is How “tightness” vs “looseness” explains the U.S. political map.

Except it doesn't.

Map from The Washington Post.
The study develops a state-level measure of "tightness"/"looseness" using the following items (as found in the supplemental materials).  The Washington Post then notes that the variation in a scale built out of these items looks a lot like the political variation we see in elections.

1.  Legality of corporal punishment in schools
2.  Percentage of students hit/punished in schools
3.  Rate of executions, 1976-2011
4.  Severity of punishment for marijuana law violations
5.  Legality of same sex civil unions (reversed)
6.  Ratio of dry to total counties
7.  State religiosity
8.  Percentage of individuals with no religious affiliation (reversed)
9.  Percentage of population that is foreign (reversed)

This is a list of very political items.  Items 1 through 6 are all the result of policy decisions, presumably made by policy makers responding to the political situation in their states. Item 5 in particular is currently being politically contested across the United States. Items 7 and 8 are about religiosity, which of course is known to be politically entwined. The same is true for immigration, item 9. So it should not surprise us that the measure is correlated with political variation. If you used this scale in a political science journal, you wouldn't call it "tightness vs. looseness." You'd call it "ideology." 

In other words, "tightness" vs. "looseness" doesn't explain the political map. The political map explains (or simply is) this measure of tightness/looseness.

The authors themselves don't exactly say that the measure explains politics, but they do say it might "have consequences for state outcomes." Which is not surprising either, since the first six items are themselves state outcomes. And the authors don't think they have a measure of ideology, even though I do. The paper is most interesting the further it gets from politics. And why wouldn't it be. Showing that a measure of politics is linked to another measure of politics is not interesting, but showing its connections to more fundamental things can be. 

This is not to say that there might not be something about the cultural construct the authors are labeling "tightness" or "looseness." (I wonder, though, how tight restrictions on guns, loose environmental regulations, and highly structured labor laws fit into the construct.) The problem is that, as with every construct, measurement is hard. This is especially the case with things that are meant to explain ideological differences. In a world in which everything is increasingly political, it's not surprising that everything is correlated with the political. If you take a bunch of items (e.g. 9) that are correlated with almost every political conflict, they will probably be correlated with each other. But that's probably because of the politics.

The problem is reification. All we have is a measure, but treat it as if the construct it is meant to measure is real, and that the measure is that construct. Political scientists are far from immune to reification. There may be a construct that we should call tightness, but putting new labels on old correlations isn't going to measure it. 


  1. What the map illustrates quite dramatically, I think, is that persistent partisan differences in U.S. electoral politics are driven by social and cultural issues to a far greater extent than economic issues.

  2. But the uppermost cheap fifa 14 coins in our minds is the final destination. On a certain day at a fifa 14 coins certain hour, we will pull into the station. Bands will be playing and flags cheap fifa 14 coins waving.