Diversionary Theory and Voting

by Professor Martini

Somewhere in your IR classes, you have certainly heard of Diversionary Theory.  This theory is plays an important part in thinking about the relationships between leaders and the public.  Essentially, the idea is that if the domestic situation is not looking too good, then the leader may try to “divert” public attention away from this bad domestic situation and focus citizens on some sort of perceived foreign threat.  When focusing on a foreign threat, the people may then be more likely to “rally around the leader” and this should lead to better evaluations of the leader.  This is especially important for leaders getting close to an election (i.e., October Surprise), but it isn’t constrained to only being around elections (many have argued that this is what Putin has done in Russia for quite some time).

This theory makes a lot of sense.  If a leader wants to stay in office (and we assume that leaders do) and the domestic situation looks grim, then we should expect leaders to try to focus public attention to foreign issues (where leaders tend to be evaluated more positively).  The problem is that we have very mixed empirical support for this notion.  While the theory makes sense, and there is some empirical evidence to justify it, there is also evidence disproving that diversionary theory is done by leaders or that it even works.

Given this need for further clarification, Sam Schutt and I decided to explore this theory at a micro-level (individual level of analysis).   Our basic theory is that if diversionary theory holds at the individual level, then we should see a relationship between individuals focusing on a foreign policy event and the lack of correctness of their vote choice.  To put it another way, if diversionary theory works, individuals should be less likely to vote for the “correct” candidate if they are focusing on foreign policy events compared to other events.  Again, diversionary theory is not constrained to only electoral periods but our research focus is on this context.

Before I go any further, let me explain what I mean by a “correct” vote.  Here we draw from some fascinating work by Richard Lau and David Redlawsk (1997, 2006) who ran experiments in the 1990s and 2000s on voting behavior.  Individuals are never going to have full information when making a vote choice, but Lau and Redlawsk were interested in understanding how well individuals selected the correct candidate for them with the limited information voters possess (this is in comparison to if voters had full information about the candidates).  If voters selected the candidate that they would have with full information, then the idea is that they voted correctly.  Through Lau and Redlawsk’s research they were able to identify what factors went into the calculation for voting correctly and how well people typically did.  Correct voting ranged between 70 and 80% (not bad with the limited information we typically use) and some of the factors used in calculating correct voting were things like issue stances of candidates vs the respondent, group endorsements and candidate specific factors.

To test our theory, we used survey data from the 2004 American National Election Study.  We use the 2004 ANES because of the focus on foreign policy in the 2004 election and the different foreign policy focuses of the candidates (Bush with terrorism and Kerry with Iraq).  We use the correct voting measure from Lau and Redlawsk to make our dependent variable – did the individual vote correctly.  We then look at what individual’s labeled as the most important issue facing the country.  We are interested in a foreign policy issue and specifically a focus on terrorism since that is what Bush pushed heavily on.  Fortunately, about 40% of the individuals in the ANES survey stated that terrorism was the most important issue.  This variable is our primary independent variable.

Sam and I ran a number of bivariate and multivariate models to test this relationship.  We don’t have space to share them all here but let me quickly summarize our findings.  First of all, we see that in general, people who believed terrorism was the most important issue in the 2004 election were significantly less likely to vote correctly.  This lends additional support to the notion that focusing on foreign events can negatively impact voter calculations.  Secondly, and most importantly, we see that this impact on voting correctly really is conditional on who the ideal candidate was for each respondent.  In our second model we broke out only those who should have voted for Kerry and only those who should have voted for Bush and saw pretty sizable differences between the groups.  For those individuals who should have voted for Kerry, they voted correctly 71% of the time if any issue other than terrorism was the most important issue.  This drops to 46% if terrorism is the most important issue – a substantial decrease in correct voting.  In contrast, for those who should have voted for Bush, the probability of correct voting actually goes up about 5 points if terrorism becomes the most important issue.

So in the end, what is this telling us?  Most importantly, I think this research gives additional support to the idea that diversionary theory can work.  People’s ability to vote correctly really did seem to be impacted by a focus on a foreign policy issue.  Many who should have voted for Kerry did not do so if their focus was on a foreign policy issue owned by Bush.  This supports a lot of research out there but also gives us a more refined understanding of how this works because we are able to provide an individual level exploration of diversionary theory.

*** If you want to learn more about public opinion and foreign policy, Professor Martini is teaching a class on Public Opinion on War and Foreign Policy