Campaigns matter. As someone who writes about racial politics and has modeled the role that racial factors have played in presidential campaigns, I am well aware that many of my astute, quantitatively oriented colleagues (including an early mentor or two) have impressive models for predicting the outcomes of presidential elections. These models, as Gelman and Sides suggest, have an excellent record of forecasting not only who will win, but also the winning candidate’s share of the popular vote. Indeed, the models-based state-by-state forecasts have often accurately predicted electoral college totals as well.

So why am I claiming that 2008 is different—that the models, if not wrong, are missing the essential feature of the historic 2008 presidential contest? First, let me restate a counterfactual that both scholars and pundits, including thoughtful Republican analysts, provided for us: in most any other year, including 2004, McCain would have won this election. There were factors in the 2008 election that made it so different that it should not be understood in the same statistical universe as previous elections, and, thus, we cannot be at all certain that models are accurate in telling us why Obama won.

Did Obama’s race make the 2008 election different? Maybe. There are certainly historical reasons to think that an African-American candidate for president would change the electoral calculus of enough voters to invalidate our models. In 1983 a Republican, Bernard Epton, almost was elected mayor of Chicago, losing only by the narrowest of margins. His election would have been earthshaking—there has not been a viable Republican candidate for Chicago mayor in most of our lifetimes. The reason he came so close was a massive defection from the Democratic party because its candidate, Harold Washington, was black. Washington would go on to be the first black mayor of Chicago. In the fall of last year, a Ford Foundation-funded study directed by my colleague Cathy J. Cohen found that 30 percent of white respondents reported knowing someone who would not vote for Barack Obama because he was black. Neither piece of evidence that race plays a large role in elections, however, is decisive. It is certainly possible that a substantial number of those who would not have voted for Obama because he was black would not have voted for any Democratic candidate. Not to put too fine of a point on it, we do not know how many likely Democratic voters defected due to Obama’s race. Similarly, it is possible that Chicago’s white voters are substantially less apt to vote for a black candidate for high office than their compatriots around the country. Possible, but not likely.

More plausible is that white attitudes have substantially changed in the quarter century since Washington’s victory and, therefore, the Chicago pattern does not extend to this era. Further, it is also the case that voters hold the president responsible for the economy far more than they do local officials. Consequently, the state of the economy is far more important in presidential than in mayoral elections.

The decisive factor in the 2008 election was the change in the composition of the electorate. In 2008 there were two million more African American voters, 600,000 more Asian Americans, and about the same number of Latino voters as in 2004. The 2008 electorate was nearly 25 percent non-white; it was only 20 percent non-white as recently as 2004. In 1976 the non-white percentage was only 10 percent. Thus, the same background conditions would have produced a McCain win if the composition of the electorate had been the same as it had been in earlier elections. In those earlier—but recent—years, McCain would have won despite his party’s unpopular incumbent, an increasingly unpopular war, and a deeply troubled economy heading for a wreck the dimensions of which we have not seen since the Great Depression.

Which explanation of the electoral results is correct? The answer has important normative implications. If campaigns really do not matter, if destiny is written in the economy and other fundamentals, we have very few reasons to make instrumental arguments for increased political participation, particularly on the part of those who have been deeply pessimistic about their prospects in this nation. The last point is not based on speculation. Along with two colleagues, I directed a study shortly after the Katrina disaster in which 80 percent of African-American respondents reported that they either did not expect blacks to achieve racial equality in the United States in their lifetimes, or ever. When I first asked this question in a 1993 survey, approximately two thirds of black respondents were deeply pessimistic about the prospects of achieving racial equality. The percentage rose steadily throughout this decade, with no motion in the more optimistic direction, and reached its height in those days following Katrina.

When we asked the same question in the fall of 2008, shortly before the election, the number had falled to 47 percent. Blacks had never been so optimistic in the last two decades (although our study also shows blacks are still far more pessimistic than any other racial or ethnic group). This optimism translated into a very sharp increase in black voter turnout as many African Americans believed for the first time (at least in recent years) that they could collectively make a difference. Ninety-five percent of African-American voters cast ballots for Obama.

If campaigns do not matter, there is less reason for these newly somewhat optimistic citizens to think they have a stake in the electoral system. Of course, normative concerns do not dictate the evidence or how we interpret it. But I believe that in this case the empirical evidence supports a normative story to the effect that, at least under some conditions, political mobilization and organization matter and can change the direction the nation takes.