Forecasting Elections

September 24, 2006

            During an election season many people ask me who is going to win. When I say ‘I don’t know’ they always seem disappointed. They must believe that political science professors have a crystal ball to predict elections.

            We have no crystal balls. Predicting human behavior is tricky business. Do not believe any pundits right now who say that they can predict the upcoming midterm election results.

But we do know historical trends and we can point to certain variables that decently explain past elections. So we can tell you what is likely to happen if this election is like previous elections.

Of course, the future might not be like the past – human behavior is more complicated than natural phenomena like gravity. If you let go of a ball, it always drops to the ground. But people can change their voting behavior. This is why our theories of politics are always more tentative and partial than our theories in physics or biology.

            So who is going to win? Will the Republicans hold on to Congress? Or will the Democrats gain majority status for the first time since the 1994 midterm elections? Well, I don’t know. But what we do know suggests that – if this election is like previous elections – the Democrats should pick up seats and have a chance to take over Congress.

            Two historical trends about midterm elections help the Democrats. One is that turnout is much lower than in general elections. Instead of the pathetic 50% turnout we average in general elections we will have an abysmal turnout between 25-35%. People need a real reason to vote, and polling evidence suggests that the war in Iraq is motivating Democrats to vote in November more than Republicans.

Another historical trend is that the party of the president usually loses seats in midterm elections. This trend obviously helps the Democrats. Only three times since the Civil War has the president’s party gained seats in a midterm election. However, two of those three were the last two midterm elections. Republicans gained seats in 2002, and the Democrats picked up seats in the 1998 midterm elections.

            Two variables that generally predict midterm election results are presidential approval ratings and the performance of the economy. Low approval ratings for President Bush are driving Republican anxieties and Democratic hopes. In the past two midterm elections, presidents have had approval ratings in the 60s: Bush in the first national election after 9/11 in 2002 and Clinton after Republicans overreached by impeaching him in 1998.

            How the state of the economy will influence the election depends on which statistics you want to emphasize. Unemployment, inflation, and the stock market all suggest that the economy is doing well. But our deficits (budget, trade, personal savings, etc.), poverty and inequality rates, and stagnant median income all suggest that the economy is doing much better for the wealthy than for everyone else.

            The best economic predictors of elections are not these hard statistics but citizen beliefs about the economy. In politics perception is reality. And most polling suggests that citizens are both worried about the economy and believe that the Democrats would do a better job with the economy.

            All of these trends favor the Democrats, and this is why the Republicans only want to talk about terrorism. National security is their best issue. Fear is their best political weapon. They essentially have to argue that this is not a normal election. We should not act like we usually do in midterm elections under these conditions. Our very survival requires that we change our behavior.

            Republicans have two things going for them. One is that fear is a very powerful motivator of human behavior. This is why politicians run negative attack ads: they work. The other is that our elections are rigged in favor of incumbents. More than 90% of House races are not close because the district lines are drawn to maximize party advantage.

The elections will be therefore decided by about 30 House races and a handful of close Senate races (including Missouri, Tennessee, and Ohio – three states that almost always vote for the winning party in presidential elections). The Republicans have the fund-raising advantage in nearly all of those elections. The Democrats will have to run the table in those close elections to take over Congress.

So please do not ask me who is going to win. All I know is that the Democrats hope the future is like the past and this will be a typical midterm election. And the Republicans hope that the politics of 9/11 means that the future will not be like the past. We will find out in November.