In voting, demographics matter more than ever

Political scientist Thomas Holbrook has been studying presidential campaigns and voter behavior for more that 25 years. His forthcoming book, “Altered States: Changing Populations, Changing Parties, and the Transformation of the American Political Landscape” (Oxford University Press), looks at how voting patterns have changed from state to state. His research interests are rooted in understanding how factors such as the national economy influence election results.

Is this election particularly challenging in terms of your work in forecasting?

The Republican field is so different from what we’ve seen before, both in terms of numbers and the type of candidates. Just the size of the field on that side makes things difficult. For one thing, it’s simply harder to poll. Moreover, two of the candidates, Donald Trump and Ben Carson, haven’t run for office before.

The Democratic side is fairly standard – a conventional candidate with a challenger who’s significantly farther to the left. So that’s easier to think about than the Republican side.

Do you see any departures from how candidates have campaigned in the past?

One big difference is that Donald Trump has hardly done any advertising at all. Of course, he hasn’t had to, he’s got the media advertising: Trump speaks, it leads.

But in terms of what’s happening under the hood, most campaigns are continuing the trend of micro-targeting, using big data to try to find voters, match them with candidates, assess the probability of turnout. In the general election, that’ll be even more significant, since the data collected by all the campaigns will be compiled for use by the nominees.

Has big money changed politics?

A lot of the money appears to be wasted. It’s spent on advertising, and the impacts of advertising are ambiguous.

Can you elaborate on that?

I think a big change is that parties have started to realize that advertising is fine, but you really have to get voters out, and you have to get your voters out. If you go back to 1998, when the Democrats picked up congressional seats even though the president’s party normally loses seats in midterm elections – they really targeted voters. In 2002, the Republicans picked up the idea.

In recent years, voter turnout has gone up. Some may say that it’s due to increased polarization, but I suspect this emphasis on mobilization has a lot to do with it.

Your recent work has been a state-by-state inquiry into changing voting patterns.

Some states have become more Republican, others more Democratic. Some have become more competitive.

There are a couple of explanations for these changes. There are standard demographic assumptions. For example, that Nevada has become more Democratic because there are more Latinos in Nevada than there used to be.

But the thesis I’m pursuing is that – again, using Nevada as an example – it’s not just that there are more Latinos, but being Latino now matters more than it used to. In other words, Latinos have become more likely to vote Democratic, so you wouldn’t need more Latinos to have the observed change.

Another example: Increases in religiosity don’t seem to have been significant, but religiosity itself matters a lot. So, states that have greater concentrations of religious congregations have become more Republican, though the number of congregations hasn’t necessarily changed relative to other states.

What that means is that the relationships between demographics and voting patterns have intensified in some cases, which can be helpful to both Democrats and Republicans, and this varies from state to state.

Can you give some other examples?

A few things really stand out. Increases in highly educated voters – in places like Virginia, Vermont and North Carolina – that really helps Democrats. An increase in foreign-born voters also benefits Democrats.

Thirty years ago, people in professional occupations were a Republican voting group. Now, they lean Democratic. They seem to be fine with economic conservatism, but are turned off by cultural politics.

Has this had an impact on the Electoral College?

This hasn’t led, as some people say, to a Democratic lock on the Electoral College. But we’ve moved from a place where Republicans were pretty comfortable with Electoral College outcomes. In the ‘70s and ‘80s, a 51 percent popular vote share would likely have translated into a comfortable Electoral College win for the Republicans. Now, that same margin would give them only a slim lead in the Electoral College, and Democrats have a slight Electoral College edge in a 50-50 popular vote.

That’s all because of this movement in the states.

A lot of your work is concerned with predicting electoral outcomes. How does your work differ from that of analysts like New York Times blogger Nate Silver?

Most academics are trying to develop a model they can use the make a prediction. So for the general election, they’ll make a prediction and stick with it – to test their model. Whereas a lot of the models you see in the media are based on polls, which reflects what’s happening in the specific campaigns. So those predictions will change from day to day.

Models that are very poll-based are accurate, but if you want to know what has shaped the election, you have to move back from that a bit and think about the structural factors: the state of the nation, the general pulse of public opinion, the state of the economy, presidential approval, things like that.


Holbrook is the Wilder Crane Professor of Government in the College of Letters & Science.


Top Stories