It’s a little unfair to pick on this book in particular. The book’s main problem is endemic in much journalism about political campaigns: a belief that campaigns matter, but an unwillingness to look hard for actual evidence. I am going to examine five events, all of which Balz and Johnson (and, in one case, David Axelrod) believe were important. This posts examines the first two.
First, at the Oct. 30, 2007, debate among the Democratic, candidates Hillary Clinton’s answer to the question about driver’s licenses for illegal immigrants. Balz and Johnson write:
bq. Hillary Clinton’s stumble over immigration proved to be highly damaging to her campaign. (p. 99)
See also Balz’s blog post at the time.
The second is Obama’s Nov.10, 2007, speech at the Jefferson-Jackson dinner in Iowa:
bq. The next morning, David Yepsen, the most influential political columnist in the state,wrote, ‘Should he come from behind to win the Iowa caucuses, Sunday’s dinner will be remembered as one of the turning points in his campaign here.’ Obama’s speech created an immediate surge of energy for his campaign, and a growing sense that caucus night in Iowa could become a climactic showdown.” (p.120)
The first graph below is from all national pre-primary polls. The second is from all Iowa pre-primary polls — where Obama’s speech might be expected to matter more. The graphs are pretty self-explanatory. The key is the vertical lines demarcating the events.
In the national polls, it sure looks like something happened after these two events: Clinton’s poll numbers dropped and Obama’s increased. But were these events responsible? Probably not. For one, the timing wasn’t really right. Clinton’s “gaffe” was on October 30. Her standing in the polls, according to the smoothed trendline, was 44%. A week later, it was…44%. Two weeks later it was…44%. Here are some apples-to-apples comparisons from before and after. Zogby, October 26: 38% for Clinton; November 16: 38%. Rasmussen was in the field during the debate. They put Clinton at 42%. A week later, they had her at…42%. Where is the “damage to her campaign”?
Let’s consider Obama’s speech in light of the graph above and this one with the Iowa polls:
In the graph of the national polls, Obama’s numbers trend upward in the weeks after the speech, but it’s very difficult to peg that to the speech itself. According to the trendline, Obama was at 23% on the day of the speech and at 24% two weeks later. The same apples-to-apples comparisons don’t suggest much of anything. The Economist/YouGov did 4 polls in November, one before the speech and three after. In these polls, Obama’s numbers were 24, 23, 18, 25. No evidence of any “speech bump” there.
But it’s Iowa where the speech should matter most. There, the evidence is even more equivocal. Obama’s numbers were improving in Iowa in a pretty linear fashion throughout 2007. After the speech, they improved at a slightly faster rate, but it was still a pretty slow increase over a long period of time. His numbers on the eve of the Iowa caucus are only 3 points higher than his numbers right before the speech. Some specific polls tell a similar story: Zogby showed no Obama bounce between its early and late November polls (actually, a 1 point decrease). Strategic Vision was in the field right around the speech and then two weeks later. They showed a 2-point increase, well within the margin of error. In short, there appears to be no “turning point” here.
Both the national polls and Iowa polls found Obama gaining ground as 2007 came to a close. But he was doing so gradually, not in sudden leaps after any barnburner speeches. Similarly, Clinton’s declining standing in the national polls did not begin until well after her purported stumble. And even if you think I’m wrong, and these events somehow mattered, the candidate’s poll numbers changed by only a few percentage points — changes that seem much smaller than phrases like “highly damaging” and “immediate surge” would suggest. Whatever effect these events had was much larger in the minds of journalists than the minds of voters.