CNN’s Peter Hamby reviewed the new campaign book “Double Down” and concludes with this shot at the “Nate Silver wing of the internet”:
Campaigns, though, aren’t just about number-crunching and statistical analysis. Candidates matter. Voters tell pollsters that they make their choices based on issues such as education, health care, taxes and the economy — and they do. But they also care about temperament, empathy, strength, reason, trust and the human side of these strange and wily people who think they’re up to the task of running the country. And as Halperin and Heilemann understand, so do readers.
It’s an odd statement in many respects. For one, Silver also thinks candidates and issues matter. For another, as Dan Hopkins pointed out, understanding issues and temperament also takes numbers. Yes, if you want inside stories that illuminate the candidates’ temperaments, read “Double Down.” But if you want to know how much their temperaments mattered to voters, you need numbers. So here are some numbers, courtesy of Lynn Vavreck’s and my book about the election, “The Gamble.” What we show is that Romney’s deficits in terms of favorability and empathy weren’t enough to cost him the race.
Here are two key things to know about how voters perceived Mitt Romney and Barack Obama. First, although Romney was viewed less favorably than Obama early on — less favorably than any presidential nominee dating back to Reagan, in fact — he closed almost all of that gap by Election Day. Here is a graph from “The Gamble”:
Pollster shows the same thing. The main drivers of this trend were Republicans, who came to like Romney quite a bit more as the campaign went on. This illustrates a decades-old political science finding: campaigns tend to rally the party faithful.
The second key thing: Romney trailed Obama in perceptions of empathy. Beginning in January 2012, we asked YouGov respondents how well various phrases described the candidates: cares about the poor, cares about the middle class, cares about the wealthy, and cares about the poor. The majority believed that “cares about the wealthy” described Romney better than Obama, but thought that the other phrase described Obama better:
Perhaps most striking is how stable these perceptions were. Obama was perceived as more empathetic than Romney even in January 2012, and this did not change much during the campaign — even after the alleged bombshell that was the “47% video.” Moreover, Romney’s disadvantage on this dimension was one that every Republican candidate has had since the 1980 election. The bigger problem for Romney may have been his party, not his personal wealth or tenure at Bain Capital.
At the end of the day, did Romney’s disadvantages in favorability and likability actually cost him the race? Our analysis suggests that they did not. We examined the connection between how voters perceived the candidates and how they ultimately voted. Then we asked a simple hypothetical: What if Romney’s disadvantages had disappeared? What might have happened in a world in which voters liked Romney as much as Obama, and thought he cared about “people like them” just as much as Obama did? Here’s our finding:
If we adjust Romney’s favorability to eliminate his disadvantage, what happens? Very little. Obama loses only a tenth of a point of vote share. If we adjust both Romney’s favorability and empathy to eliminate his larger empathy disadvantage, what happens? Obama is estimated to lose almost 1 point in this simulation. This would have been enough to tighten the race—theoretically, shrinking Obama’s winning margin from nearly 4 percentage points to just about 2 points—but would not have tipped the race clearly in Romney’s favor.
And note that even these estimates may be overstating the impact of favorability and empathy, since we can only partially confront the obvious chicken-and-egg question: Did people decide how to vote based on favorability or empathy, or did they decide how to vote based on other factors and then arrange their perceptions of favorability and empathy accordingly? We find some evidence of the latter, actually, suggesting that the role of temperament was even less consequential. This is consistent with I wrote months before Election Day and with Jon Bernstein’s comments Monday.
All of this amplifies a reason we wrote “The Gamble.” Journalists who write books about campaigns are great at illuminating who the candidates were and telling us why the candidates made the decisions they did. But those books are not as good at telling us whether any of that mattered to voters. For that, you need social science, and you need numbers. As Bill James once said, the alternative to good statistics isn’t no statistics. It’s just bad statistics.