Public Opinion

Polarization Is Not Associated with Trust in Government

Apr 30 '10

Clive Crook likes this passage from a new paper (pdf) by William Galston:

bq. Can we honestly say that today’s mistrust–between the political parties, and between citizens and their government–remains within Madisonian bounds? Can we judge our party system healthy if it fosters this mistrust? If we knew how to change it, would we choose to perpetuate a situation in which the very process of self-government stands in such disrepute? These are not the questions of an aging academic looking back with nostalgia. They are the concerns of a citizen looking forward with alarm. Our adversaries around the world will never be able to harm us as much as we are now harming ourselves. And if our party system remains as it is, this process of self-destruction will only get worse.

Crook says that “The essay is essential reading, and I’ll have more to say about it later.” Since Crook focuses on that passage, so will I. Actually, let’s back up a couple sentences in Galston’s paper to the empirical proposition that animates that passage. Galston writes:

bq. There is evidence, finally, that rising polarization is one of the forces contributing to sharply declining trust in government.

He cites John Hibbing and Elizabeth Theiss-Morse’s Stealth Democracy (previously discussed on this blog here). However, Hibbing and Theiss-Morse aren’t systematically testing this relationship. That is, they are not looking at whether polarization affects trust. You could read that implication into their findings — e.g., the finding that citizens don’t like conflictual political processes, which perhaps polarization breeds. But that’s not a direct test.

Let’s do a direct test. First, here are two graphs, one of the percent who trust the government (data discussed here) and one of the the difference between the median House Republican and Democrat in the 85th-110th Congresses (data here, courtesy of Keith Poole). As the the difference between the medians gets larger, so does polarization.



Already you can see the problem. The initial decline in trust precedes the increase in polarization. And then trust goes up and down and up and down, but polarization just goes — to quote my toddler as he ascends any staircase — “up up up.”

By matching the survey data to the last year of each Congress, you can plot trust against polarization.


The orange line is a linear fit. The gray line is a non-linear fit. The gray line would have us believe that a very modest increase in polarization leads to a huge decline in trust, but somehow the remaining increase in polarization — i.e., the last 30 or so years of history — had no effect on trust. The story looks even more shaky.

Let’s put the final nail in the coffin. Here is an updated version of my earlier graph comparing trust and the state of the economy. (As before, I used Doug Hibbs’s data, but modified his calculations to generated an estimate of economic growth in midterm years):


This relationship still looks sturdy. Now, what happens if you regress trust on both polarization and economic growth? The effect of economic growth is substantively and statistically significant (b=6.5, se=1.7 for my nerds). The effect of polarization is neither (b=-24.6; se=19.7).

I don’t know what more Clive Crook wanted to say about this essay, but perhaps this will save him some time:

Polarization does not affect trust in government.