Wednesday, November 10, 2010
Monday, November 01, 2010
NPR spreads a falsehood
On Sunday's "On The Media", NPR's Brooke Gladstone interviewed Stephen Dubner, co-author of Freakonomics. Dubner recounts a story from the book:
[Levitt] began to notice a pattern, which was that there were quite a few instances in which the two same candidates ran against each other repeatedly.
And what Levitt found was that if you looked at the Congressional races where the same two candidates ran each other repeatedly, well, the appeal of the candidate presumably didn't change a whole lot – you'd have to control for incumbency and things like that – but what often did change was the amount of money that was spent.
And what that allowed him to do was to try to isolate the causal effect of the money itself. How important is it to spend a lot more money? And the result was ... you can double the amount you spend and raise your share of the vote by about one percent.
Similarly, if you’re winning an election and you cut your spending in half, you'll lose only about one percent of the vote.
So, the original study showed that the amount of money spent in a second (or subsequent) match-up between two candidates did not substantially change the outcome from that which occurred in the first election. In other words, campaign spending has limited effectiveness in rematches between two candidates. The study says nothing about the role of campaign spending in the original matchup -- it can't, for those are the baseline used to measure the effect of spending on the rematches.
This result seems plausible -- there exists what we know as "brand loyalty", and, having once selected one candidate from a pair, it is reasonable to expect voters to have some brand loyalty that keeps them voting for the same person election after election given the same match up. So, to the degree the composition of the voting public is static, we would expect campaign spending to have limited ability to sway voters from the choices they had already made between two candidates.
However, in the first election between the pair, a significant number of voters will not have made up their minds, and thus be more open to influence from many sources--including campaign ads. So, there is reason to believe that campaign spending can be important in the initial campaign between a particular pair of candidates (e.g.: Angle v. Reid; McMahon v. Blumenthal; O'Donnell v. Coons, etc.), and Levitt's study, of course, says nothing on the matter.
And so, Dubner sliped one past Ms Gladstone when he then turned around and made the far broader -- and unjustified -- claim that: [I]f you look over the long stretch, you just don't find a causal relationship between spending and electoral outcome. Dubner knows nothing of the sort. Most elections are not rematches, and Levitt's study gives no insight into the role of campaign spending on initial matches.
Am I asking too much to expect the reporters to catch such errors? Your thoughts are appreciated!
Labels: Jim's Posts