In an e-mail message responding to my post earlier today (here) about misinterpretations of the 1968 New Hampshire primary result, Adam Berinsky has pointed me to a paper (“The Factual Basis of ‘Belief Systems’: A Reassessment,” Critical Review 18 (Winter 2006), 233-254) in which Sam Popkin argues that Converse et al. erred in depicting the majority of McCarthy’s New Hampshire voters in 1968 as hawks rather than doves. As I read Popkin’s evidence, though, I’m not convinced. He does present some polling data from earlier in the year (three weeks before the primary at the latest) to the effect that very few of those polled (many of whom presumably ended up not voting or did vote, but not for McCarthy) considered McCarthy a hawk. Even if that point were granted, it would still refute just a subpoint of Converse et al.’s argument. Their main point, again, was that most of those who voted for McCarthy were hawks.
To try to refute Converse et al.’s main point more directly, Popkin introduces data from a Louis Harris survey conducted a month before the primary. The timing of that survey puts his argument under a cloud, but let’s set that objection aside.
Popkin’s smoking gun is that in the Harris poll “The vast majority of the McCarthy voters were doves by any defnition: 83 percent believed that the United States should be doing more to bring about negotiations; 61 percent thought we should ‘pull out our troops now. Only 7 percent of the McCarthy voters wanted a total military victory, and 13-15 percent wanted to carry the war into North Vietnam or step up the military effort.”
Pretty strong stuff, eh? Well, not really when you look at it more closely. The respondents in that Harris poll were 292 registered New Hampshire Democrats and independents who said they might vote in the primary. Of these, 16% said they would vote for McCarthy. Going back to my previous point, let’s acknowledge of that some of those folks probably didn’t vote (or, if Popkin is right, some of them may have figured out in the ensuing month that McCarthy wasn’t a hawk after all and ended up voting for Johnson as the more hawkish of the two). But having acknowledged those possibilities, let’s just set them aside again and get to the main problem with Popkin’s ostensibly damning evidence.
The problem with Popkin’s evidence becomes clear via some math so simple that even I can do it — so simple, in fact, that I can do it in my head: 16% (the percentage of potential McCarthy voters) of 292 (the number of respondents in the Harris survey) equals 46. This means that Popkin’s main countercharge is based on the responses of a not-so-grand total of 46 possible voters a month before the primary. That’s the sum and substance of the evidence that Popkin introduces in refutation of Converse’s claim that most McCarthy voters in the New Hampshire primary were hawks. In my view, that just won’t wash as reliable evidence. I don’t have a dog in the fight, which goes back to Philip Converse versus V.O. Key, about the (ir)rationality of the American electorate, but I do like to see the arguments of the contenders based on data that I, as a reasonably well-informed consumer, consider credible.