Monday, February 8, 2016

What impact is "socialism" having on the election?

Vox recently asked some political scientists how they thought Bernie Sanders would fare in a general election. Of course, the closest way we have to "objectively" answer this sort of question is just to poll people: and by that approach, Sanders generally does better than Clinton (an average of three points better against the three Republican front-runners).

But even that relatively rigorous approach is pretty useless this far away from election day, so instead Vox's scientists generally answered a different question: how would a radical democratic socialist movement candidate fare?

This approach is understandable, though perhaps problematic for different reasons. Oddly, however, instead of attempting to answer the new question with any kind of rigor, their answers "read as guestwork rather than estimates" as Mike Konczal put it. The article involves all kinds of conjecture about the role of loss aversion, ideological spectrum fits and the performance of "movement" candidates, but little direct interrogation of the question they set out for themselves.

Here, I'd like to tackle this a little more rigorously. Unfortunately there aren't any polls available that ask respondents directly how socialism is impacting their decision, but we can at least approach an answer by focusing on the economic and ideological considerations at hand.

Preference by household income

Loss aversion may, as Bruce Miroff speculates, play some role in voter preferences - but this is just one of all kinds of economic considerations that may impact decisions. Instead of trying to deduce how people in different economic situations are thinking about the election, we could just do a behavioral analysis to see what their decisions actually are. To do this, I drilled into some poll data to compare how the two Democratic frontrunners would do, on average, against the two Republican frontrunners, Donald Trump and Ted Cruz.


The first point to notice here is that household income never plays a decisive role in whether either candidate can beat the Republican; in every bracket, the Democrat wins every time. The only question here is which Democrat wins and by how much - so let's compare margins of victory between Sanders and Clinton.


The general trend is clear: Sanders wins by a greater margin than Clinton in every household income bracket except for the second. If there's any economic trend here, it's that Democrats win slimmer victories among richer voters - but by this measure, if Sanders' ideology has any additional impact it is helping him, particularly among poor voters, where he gets a nearly 10% bump over Clinton. (The more accurate way to put that though would be that something about Clinton is hurting her, since her 10% disadvantage among the poor is accounted for almost entirely by more votes for Republicans.)

Preference by ideological identification

In addition to economic considerations, Vox also proposes that voters may make their choice "by identifying which candidate fits closest to them on an ideological spectrum." Using the exact same approach, we can investigate how Clinton and Sanders would do, on average, against Trump and Cruz, and break the results down by ideological self-identification. We can then compare Clinton and Sanders' margins of victory/defeat to see if there is any kind of ideological trend.


Again, two trends stand out. First, the most significant predictor of performance is partisan identification: Democrats do better then Republicans among liberals, and worse among conservatives. But the second trend is a lot more significant: Clinton can only outperform Sanders among the very conservative.

I see three possible explanations for this. The simplest is that Clinton is simply too far to the right. That's why she even loses conservative leaners to Sanders, ties with him among moderate conservatives, and can only outperform him among the far-right. Sanders, while to Clinton's left, clearly isn't so far left that it's alienating a significant number of voters.

The second possible explanation is that people tend to dislike Clinton even more than they dislike socialism. This could be due to sexism, or her wealth, or her record, or all kinds of things - but whatever the case, more people vote against Clinton than against Sanders in every ideological bracket:


Clinton can peel off a few more moderate-and-very conservatives than Sanders can, but that advantage is almost completely overwhelmed by the number of conservatives who vote against her; and it's nothing compared to the enormous advantages Sanders has among liberals.

One last consideration

The third possible explanation for Sanders' ideological advantage is that voters just don't know him well enough yet. This could also account for his advantages across nearly every income bracket. If some combination of increased exposure and increased criticism manage to reverse enough of the advantages we've noted so far, he could perform worse than Clinton might have, and on top of that he could even lose.

Of course, this is coming awfully close to saying that Sanders will lose if something makes him lose. But we don't need to dismiss that vague possibility to point out that the electability argument against socialism has failed. To the extent that it can be affirmed or dismissed by what's actually happening in the polls, it should be dismissed; to the extent that the polls give us no insight into what's going on, we are no longer dealing in science.