How voters decide: A book review with lessons for campaigns

cross posted from Daily Kos

This diary is based on my reading of the book How Voters Decide:Information Processing during Election Campaigns by Richard R. Lau and David P. Redlawsk.  But it’s not a traditional review: I will get that part out of the way quickly.  Nor is it a summary: My skills are not up to summarizing 250 pages of fairly dense text into a diary that anyone would want to read.

Rather, I attempt to take the lessons they teach about how voters decide and how they process information and translate them from scholarly political science into practical tools. 

In an attempt to keep the diary to a reasonable length, I have not tried to make it too organized, but kept it almost as a list of what might be extra-long bullet points.  I hope it is, nonetheless, comprehensible.

It’s below the fold

First, a brief review.  If you are a political scientist interested in voting theory, you should read this book.  But such people likely already *have* read the book.  More generally, the book is aimed at political scientists, but interested and persistent lay-people will also find much of interest.  It’s a very good book.

There are a few caveats: The prose will not remind anyone of Mark Twain.  This is a political science text, and it reads like one.  The English is grammatical, but it takes some work getting through it.  The graphics are poorly done: Occasionally they are misleading, more often, they are simply not well thought out and do not convey information as clearly as they might.

With those caveats, I can enthusiastically recommend this book to audiences that would appreciate it.  The authors note that traditional models of voting are static, and concentrate on trying to predict who will vote for which type of candidate.  They take attributes of people and attempt to use those attributes to predict position on some political spectrum. 

Lau and Redlawsk do not denigrate such efforts, but they are after something else: Not *what* decision voters will make, but *how* they make those decisions.  How do people get information about candidates and use that information to make a decision.  Tellingly, their analysis does not, mostly, involve any real elections.  They are not trying to explain why people voted for (say) Gore or Bush, but how people take information and use it to make decisions.  To get at this, they invented a new and ingenious methodology, in which information about candidates scrolls on a screen, and people have to pick which information to access.  Given that, in real life, almost no one can learn everything about all the candidates (especially in primaries) this method seems to be a realistic portrayal of campaigns. 

A second major innovation is the authors’ definition of ‘correct voting’.  Rather than impose their own ideological views on voters, they define a ‘correct’ vote as the one a voter would make if he or she had access to all the information, and virtually unlimited time to make the decision. 

How can the findings of this book be used by campaigners?

Lau and Redlawsk define four basic methods of making a voting decision:
1.  *Rational Choice* involves a cold, calculating look at the positions of each candidate, how well they match up with the voter’s own views, what the likely outcomes of electing the person would be, and so on.  Rational choice demands a lot of effort, and, often, rational choice voters evaluate candidates based on their own self-interest.

2.  *Early socialization* voters have made a choice about voting earlier in their lives, and nearly always stick to that decision.  In the general election, these voters will almost always vote for the same party, often without much consideration, and their partisanship often colors any attempt at objective evaluation of the candidates.

3.  *Fast and frugal* voters are a generalization of ‘single-issue’ voters.  These people vote on one or a few issues, with no interest in the candidates’ positions on other issues.

4. *Bounded rationality* voters gather a very few bits of information about each candidate, and then use that information to confirm views about the candidates in each party.

From a campaigner’s point of view, we must immediately separate primaries from general elections.  I’ll discuss general elections first.

People in two of these four groups are almost unreachable in general elections.  People in group 2 have made up their minds years ago, and people in group 4 are after a few bits of information.  To get a person in group 2 to change party is nearly impossible; to get a person in group 4 to do so, we must present information showing either that our candidate is not a typical Democrat or that the Republican candidate is not a typical Republican.  Alternatively, we can engage in a longer-term effort to show Republicans in Group 4 that their views of the two parties are incorrect (e.g., a person who views the Republican party as fiscally responsible may be susceptible to data showing that the largest deficits have been in Republican administrations). 

In both the primary and general election, people in group 3 are after information about a few issues.  The problem, then, is telling which people in group 3 may be view our candidate’s positions favorably, or the opponent’s views unfavorably.  Fortunately, many single issue or few-issue voters may belong to organizations associated with those views.  Clearly, people who vote on one or a few issues feel strongly about those issues, and these strong feelings are unlikely to surface only at election time.  We can identify these people by memberships in organizations.

We can then turn to people in group 1.  These people have a tremendous amount of information to process, and, like all people, they have limited means with which to do so. Not only are we all limited in terms of how much time we can devote to finding out candidate’s positions, we have limited ability to hold those positions in both short and long-term memory.  Even in the general election, when there are usually only two serious candidates, each may take a position on 50 issues, and each position may be complex.  Perhaps a position takes one page of text to adequately express.  We then have 100 pages of text to evaluate. This is very hard.  How do we do it?  We rely on transferring information from short term memory to long term memory.  We also rely on a number of heuristics.  More about these later.

Given these four types of decision strategy, it is critical for campaigners to identify who employs which type of strategy. One method for doing so has already been touched on: We can find group 3 people by membership in organizations.  It may also be possible to find group 1 people this way: We Kossacks, for example, are likely to be highly partisan, but we are also members of group 1.  We are very interested in politics.  (As an aside: People in group 1 can be, and often are, highly partisan.  They differ from people in group 2 in that they, nevertheless, gather a lot of information).  I cannot see ways to easily identify people in group 2 or group 4.

People with different characteristics have different preferences as to *amount* of information to process.  In particular, more educated, more politically sophisticated, and younger people prefer and process more information than those who are less educated, less sophisticated, and older.  This could play a role in pitching our message to different audiences.  In addition, different campaigns lead to different decision strategies.  When there are more candidates in the race, and when they are less ideologically distinct, people are able to process less information about them.  This is important to keep in mind when planning primary vs. general election campaigns. 

Evaluation of candidates is based on two general types of information processing: On-line and memory-based.  On-line refers to a sort of running tally of good points vs. bad points about a candidate, without remembering exactly what those good and bad points were.  Memory based refers to actually remembering things about the candidates.  People in groups 1 and 2 rely on memory, while those in groups 3 and 4 can rely on on-line processes.  But both are very important, even when the other is controlled for.  And both are more important in primaries than in general elections, when political party is the most important factor.

As noted above, political heuristics are often used to simplify the process of making a choice. A heuristic is a sort of cognitive shortcut, or a rule of thumb.  The general study of heuristics was pioneered by Kahnemann and Tversky.  In elections, Lau and Redlawsk consider several types of heuristic: Group endorsements, partisan schemata, person stereotypes, and candidate viability.  Group endorsements refers to the ratings made by various groups (e.g. the ACLU) of the various people.  Partisan schemata is the use of stereotypical images of the Democratic and Republican party (and also of minor parties), person stereotypes are those based on image of the candidates as people (e.g. ‘he seems likable’) and viability is a judgment of whether the person ‘could win’.  Everyone, the authors found, uses these heuristics to lesser or greater degree, and heuristic use is not strongly related to voter characteristics or to campaign characteristics.

In summary, this book provides a lot of information that will be of use to people who run campaigns, if they are willing to dig a little.