A coding problem in the classic study, Nature and Origins of Mass Opinion

Gaurav Sood writes:

In your 2015 piece, you mention: “In my research I’ve been strongly committed, in many different ways, to the model in which voter preferences and attitudes should be taken seriously.”

One of the reasons people in political sciene think voters are confused is because of data presented in a book by Zaller—Nature and Origins of Mass Opinion.

Recently Paul Sniderman re-analyzed the data, taking issue with how “conflicts” are coded. The point is narrow but vital. To make it easy, I have taken screenshots of the two relevant pages and included their links here and here [I’ve updated the links — ed.].

The chapter also touches upon another point that is in your wheelhouse—how key claims go unscrutinized. When writing a super brief review of the book, here are a few lines I came up with on that point: “What is more startling and sobering is that something regularly taught in graduate courses and so well cited is so under-scrutinized and so underthought. The citation/scrutiny ratio is pretty high. And tells a bunch about biases of academics and chances of scientific progress. It is a strange fate to be cited but not be scrutinized.”

I’ll have to take a look at Sniderman’s book now and then talk these ideas over with my colleagues in the political science department. I’m writing this post in early Oct and it’s scheduled for the end of Apr so this should allow enough time for me to get some sense of what’s going on.

In any case, Sood’s remark about “the citation/scrutiny ratio” is interesting in its own right. It often seems that people love to be cited but hate to be scrutinized, most famously when researchers in psychology have complained about “bullying” when outsiders do close readings of their articles and point out things that don’t make sense.

On the other hand, some people love scrutiny: they feel their work is strong and they welcome when outsiders make criticisms and reveal flaws. That’s how I feel: citation and scrutiny should go together.

I don’t really know Zaller so I can’t say how he’ll react to Sniderman’s comments. A quick web search led to this article by Larry Bartels who writes that an “apparent evolution of Zaller’s views is a testament to his open-mindedness and intellectual seriousness.” So that’s encouraging. I also came across an article by Sniderman and John Bullock, “A Consistency Theory of Public Opinion and Political Choice: The Hypothesis of Menu Dependence,” that seems relevant to this discussion.

7 thoughts on “A coding problem in the classic study, Nature and Origins of Mass Opinion

    • “I’m writing this post in early Oct and it’s scheduled for the end of Apr so this should allow enough time for me to get some sense of what’s going on.”

      OK I’ll bite. How have your ideas about this evolved since then? Have we learned anything about appropriate or best practices coding for turning an abstract concept (conflict) into a measure (deaths? soldiers sent?) into a discrete 0/1 variable?

        • I probably would’ve completely forgotten about the whole question until my post showed up whatever months later. But I’m teaching a lot right now so I can’t keep anything that isn’t a lecture tomorrow straight. Also I would not have a half a year of pre-written blog posts handy either.

Leave a Reply

Your email address will not be published. Required fields are marked *