We agree with much of Gelman and Sides’s characterization of the 2008 election: fundamentals matter, and the economy is fundamental. Nonetheless we have a few cavils, and we think some claims can be pushed further than they dare. In particular, they place more faith in forecasting models based on economic performance than we do, even as they understate how puzzling it is that the September 2008 economic meltdown failed to alter the partisan landscape. And there is much to say about Sarah Palin as a campaign phenomenon.

Gelman and Sides rightly criticize the tendency of pundits to credit single campaign moments with determining the outcomes of elections. With an unpopular administration, war in Iraq, and an anemic economy (even before September), a Republican victory was improbable from the start. From this perspective, the Obama campaign was simply a series of reminders to voters about whom to blame for their plight. (Oddly, campaigns may take on a special significance when one side fails to assign appropriate blame or credit. Al Gore’s loss in 2000 arguably resulted from his refusal to explicitly link himself to the Clinton administration’s economic accomplishments.)

On these grounds, what would the predictions have been for 2008 vote shares? The forecasts posted in the October edition of the American Political Science Association’s PS: Political Science and Politics were uncannily precise. This apparent precision is misleading, however. The median forecast was right on the money, but most of the others were well off target. Moreover, in 2004, the range of predictions barely captured the actual result. In 2000 the range was off completely. The quality of the predictions from individual vote-prediction models shifts from year to year, and different models take turns as the best predictor. The recent record suggests that the accuracy of the 2008 median forecast was a happy accident.

What is more, the predictions performed so spectacularly by ignoring the meltdown that began in mid-September. A few of the forecasters took this to heart and sought to explain why, contrary to intuition, their predictions held up. One explanation was that Barack Obama’s race prevented an even bigger Democratic landslide. But Gelman and Sides train an acidly critical—and persuasive—lens on this claim. Another argument is that the McCain-Palin ticket was actually quite strong but was deprived of a narrow victory by the meltdown. We pour cold water on this proposition ourselves.

Both of these arguments founder on a pattern that Gelman and Sides identify: little in-campaign evidence indicates that the financial crisis triggered by Lehman Brothers’ September 15 collapse was a driving force behind Obama’s victory. This point deserves more attention. It is clear from the daily tracking in the National Annenberg Election Survey (NAES) that the crisis made the economy far and away the most important and troubling issue for voters. But the effect of national economic perceptions on voters’ evaluations of candidates and vote intentions actually shrank after September 15. The election seemed to draw attention away from economic troubles rather than toward them. The NAES also confirms, as Gelman and Sides note, that the Democratic share of vote intentions turned up before the collapse of Lehman Brothers.

The failure of the economy to move the campaign is a puzzle in its own right. Election after election shows that perceptions of the economy are a crucial force shaping vote preference. Why then did events as terrifying as the meltdown not translate into electoral decisions? And if the meltdown did not drive the dynamics of the campaign, what did?

We think the answer to this question starts with Sarah Palin, and we base this judgment on the level of detail made possible by the NAES. John McCain’s August 29 announcement of Palin as his running mate surprised the Republican establishment, the media, and especially voters. She made a strong first impression: she enjoyed high approval ratings after her acceptance speech, and the percentage of voters saying that they intended to vote Republican skyrocketed. But within days of the speech, her ratings began a precipitous slide from which she—and the McCain campaign—never recovered. Throughout the rest of the campaign, vote intentions were closely tied to Palin’s approval ratings: each major Palin approval drop was followed, within a day or two, by a drop in McCain vote intention. No other factor moved McCain support with such precision. Comparison of the correlation between running mate approval ratings and vote intentions from 2000 and 2004 confirms Palin’s peculiar importance in 2008.

This finding also presents a puzzle. We normally discount vice-presidential picks as factors in elections, and so we should. What was it about Sarah Palin or about 2008 that made this choice so important? We suggest two places to start. First, John McCain’s age and medical history may have played a role, as there was a reasonably high probability of Palin becoming President before 2012. Second, borrowing powerful themes from Hillary Clinton’s primary campaign, McCain sought to emphasize Obama’s lack of executive experience, but by choosing Palin, he may have undermined that argument at a stroke. Is emphasizing Palin’s role in the unraveling of McCain’s post-convention advantage the same as saying she cost him the election? This seems a stretch. One could as easily say that she gave him a significant but unsustainable boost and that the rest of the campaign consisted mainly of shifting focus away from this distraction. Even so, bringing Sarah Palin to the foreground allows us to consider factors that attending only to fundamentals forces us to neglect: rhetoric, perceptions of candidates’ fitness for office, and the contingency of things.