The American Voter

"This was the beginning of modern political science – of empirically oriented, big data, statistically tested work. The book was published in 1960 but it drew on surveys taken during each of the three previous US presidential elections – a smaller number in 1948 and 1952, and a much larger number in 1956. So the book was largely based on the 1956 US election, and that will become important in a moment. These were large samples of several thousand, randomly selected from American voters. They were all asked the same questions: it was a standardised survey, about 45 minutes long. It probed all sorts of attitudes people had toward politics, and various structural factors – race, geography, income – that might influence those attitudes. Finally, the data was coded and analysed at the individual level. That’s important because it allows you to assess the correlation between, say, age and voting, controlling for all sorts of other things at the same time – income, region, religion, race… You could factor out all these other confounding variables and see just what the relationship is between the two target variables. The headline finding of the book is that the strongest correlate of a person’s vote is that person’s party identification, and the strongest correlate of a person’s party identification is their parents’ party identification. These authors were all psychologically oriented, and they see this as early childhood imprinting. They took identification really seriously: they saw it as an identity thing, where you say “I am a Republican” in the same way as “I am an American”, “I am gay”, “I am Catholic”. So the combination of the identity emphasis and the early childhood socialization emphasis suggested to them that party identification and voting patterns are rusted on. You’re not going to change them. The issues, they found, don’t really matter – people don’t have much information about the policy issues, or any coherence ideologically in their attitudes. That’s the headline story, that people are psychologically driven rather than rational, issue-driven voters. No, the story – the specific findings – have not stood the test of time very well. Just four years after the book was published, the story took a serious hit when the Republicans nominated Barry Goldwater for president. Barry Goldwater was one of the few senators who voted against the 1964 Civil Rights Act. Black Americans had been rusted on Republican voters from the beginning – that was the party of Mr. Lincoln the Great Emancipator, after all – and overnight, they flipped and became a solidly Democratic constituency. So much for that theory of early childhood socialization and a fixed party identification driving everything. How well-informed are voters? There is a big body of literature suggesting that voters are about as informed as they need to be to make rational choices. How many Americans could name all nine members of the Supreme Court ? But do you need to be able to name all nine justices to know that you didn’t like them overturning Roe v Wade? You just need to know that the key justices were appointed by Trump, and if you want to get Roe v Wade back, you’d better vote against Trump, because he’s the guy who keeps putting these anti-abortion people on the Supreme Court. There are lots of models of low-information rationality, stories about how voters can use cues and shortcuts that help them arrive at the vote that they would have arrived at had they been perfectly informed. Don’t read The American Voter for its substantive conclusions, necessarily, but appreciate it for its contribution in starting out modern political science in its big data, statistically analysed fashion."