Book Review: Neither Liberal nor Conservative – Ideological Innocence in the American Public

If there was any kind of standard operating procedure around here, I would say that I was going to do something a bit different today. Today, I’m going to “review” (i.e. organize my thoughts/ramble a bit) about Donald Kinder and Nathan Kalmoe’s book, Neither Liberal nor Conservative – Ideological Innocence in the American Public (publisher’s website). This book is well written, no longer than necessary, and easily digestible.[1] If you are interested in belief systems, you should read it.

The basic idea of this book is in the title. Americans aren’t clearly ideological. They often don’t hold consistent opinions and often don’t endorse any particular ideological label. The meat of the book will be useful for social psychologists and political psychologists to read and digest as it provides a summary and update of some of the ideology literature and provides initial fodder for some interesting research questions.

Converging Evidence from Converse

The first section of the book reviews the seminal work by Converse, both in his famous 1964 book chapter, but also in related work published in The American Voter and other less well known (to me) writings from the same basic period. This is important work that inspired decades of work on belief systems and ideology in political science. It is as relevant now as it was in the early ’60’s.

And indeed, this is part of Kinder and Kalmoe’s point. They summarize Converse as making three basic claims.

  1. Almost no one uses ideology to organize and explain their beliefs and few people have the necessary knowledge to do so.
  2. The typical person does not have well structured (correlated) political beliefs when looking across a broad set of beliefs. Political elites, on the other hand, have better structured political beliefs.
  3. People have very unstable opinions that shift from one year to the next. This is not due to a genuine conversion (and re-conversion) on the issues, but reflects that many people have “non-attitudes” about issues.

The helpful second chapter then also reviews the reams of research (and reports some novel analyses) that has tested Converse’s claims over the last 50 years. In short, they conclude that:

  1. Most people still do not use ideology to organize and explain their beliefs. Converse may have underestimated it, but the numbers are still low.
  2. People who do use ideology to organize their beliefs are unusual people. They are very interested in politics.
  3. Most people still do not have well structured political beliefs and they are unstable, although with improved measures it isn’t quite as bad as Converse estimated.

That is, since Conserve’s original analysis not at a lot has changed. Sure measurement has gotten better and there is a bit more ideological thinking than expected. But the overall numbers are still low. By these standards Americans are still innocent of ideology.

Contemporary Ideological Identification

But wait a minute you say, such an analysis misses the clearly increasing political polarization and partisan hostility.[2] Or if you are aware of the social psychological approach to ideology you might point to the hundreds of papers linking people’s self-reported ideology (on the 1 [strongly liberal] to 7 [strongly conservative] scale) with a host of psychological variables (e.g., personality, need for closure) and various outcomes (e.g., prejudice, policy positions, perceptions of inequality). With all of this evidence in place, sure ideology matters and Americans are guilty of it.

Kinder and Kalmoe would like to convince you that this is wrong. They analyze a lot of data, much of it from the American National Election Studies, on people’s ideological identification. Here are there claims.

  1. Using this measure, people still aren’t that ideological. About a quarter of people opt not to select an ideological identification and another quarter pick the middle option. The rest rarely pick “strong” identification options and so even among the half that have some identity, that identity is not strong.
  2. People are not becoming more polarized on ideological identification (although, they might be on issues, see Pew).
  3. Ideological identification does not do a good job predicting voting, especially when compared to partisan identification, when information is low, and when elections aren’t competitive. Ideology identification may have consequences for political behavior, but primarily among people who are knowledgable and interested in politics. Partisan identification, however, is typically a stronger predictor.
  4. Ideological identification doesn’t even predict political policies very well.
  5. Ideological identification may not be a predictor of policy attitudes and voting behavior, but instead it might be a summary of people’s politics. That is, it is a consequence of the voting decisions people make and the group attachments (e.g. ethnic groups, religious groups) that people have.

My Annoyances

Now, the support for some of these claims is better than others. For example, to claim that ideology is not a cause they use models where ideology at Time 1 is correlated with policy or behavior at Time 2. They argue that this helps establish causal precedence. But I do not think this is the case. If the measures at Time 2 have any kind of stability overtime, as voting intentions and policy attitudes do (even if not substantial!), it breaks down this argument (e.g., the attitude reported at Time 2 may have already existed prior to Time 1).

There are also instances where the authors go out of their way to remove the influence of ideology. E.g., when using ideological identification to predict policy, they also control for values of limited government and equal opportunity. Why? Well, it’s not clear since they don’t control for them anywhere else in the book and there are other values in the data they use (e.g. traditionalism). When ideological identification predicted attitudes towards some social issues (e.g., abortion), did they conclude that ideological identification might be relevant to social issues? No. Instead they added additional control variables (e.g. feeling thermometers towards feminists) until they removed the correlation. An ungenerous interpretation of this is that it looks likes p-hacking in reverse.

Finally, although I know that the authors are not fans of the social psychological work on ideological identification, they sometimes act as if it doesn’t exist. For example, there is an entire chapter on where ideological identification comes from and the vast literature on genetics, personality, negativity bias etc isn’t mentioned. Not once.

 Social Psychologists Take Note!

Despite these annoyances, I’m writing about the book because I like. Why?

First, it makes clear and plain that ideological identification and party identification are not the same thing. These are not just two indicators of the same latent variable (something I’ve argued in emails and other’s have argued on Twitter) and so should be modeled separately. Of course, they are correlated and they likely will share some properties, but that does not mean they are tapping into the same thing. Do psychological theories of politics take the differences between ideological and partisan identification into account? Not in my experience.

Second, it makes clear and plain that ideological identification will primarily have its impact for people with higher levels of political knowledge and interest. Others have shown this too (e.g., John Zaller, Ari Malka, a ton of political scientists). Kinder and Kalmoe show it again. Maybe it will start to stick. If you are anything like me, you might argue that you’ve used measures of ideological identification many times and find clear effects. The work here suggests these effects may only be limited to people with high levels of interest. Is that incorporated into our theories? It is in political science.

Third, the book has a lot of good research ideas. Some of these stem from my annoyances. For example, they might not mention psychology when looking at what predicts ideological identification, but part of the reason is that they are looking at what distinguishes someone who has an ideological identification (either liberal or conservative) from someone who does not. The personality approaches to ideology don’t typically look at this. Social psychologists have a lot of knowledge about what causes identification. We can use it here.

Similarly, some of the evidence they use to argue that ideological identification is a summary is a bit weak. It is suggestive, but it could really benefit from a well constructed social psychology experiment. They draw on Bem’s self-perception theory. The field’s experience with these sorts of experimental designs may be the right fit for answering this question.

The book also raises the question as to whether ideological moderation is an ideology. The authors lean towards “no”, but there is clearly more work available in this area. Some moderates may be legitimately moderate (somewhere between liberal and conservative), some may not have many opinions (not ideological at all), and some might have a mix of liberal and conservative opinions that they use to construct their moderate identification. Each of these groups may have different social and psychological profiles.

And finally, for the psychometricians out there, the whole book raises the question (including in an appendix) as to the best way to measure ideological identification. If this construct is interesting to study (and I think it is), then what is the best way to capture it? Psychometricians attack!

Wrapping Up

This book summarizes theoretically important work on political ideology and ideological identification. It identifies shortcomings in current models of ideological identification and it points a way forward. With all that, you also get inspiration for new research projects aimed at understanding how ideological identification works. Read the book.

[1]: Descriptions that do not apply to this post.

[2]: Some of the time series data only go to 2000. A lot of polarization has occurred since then. Perhaps another update is warranted.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s