« Twitter roundup | Main | Twitter roundup »

September 01, 2010


Fair enough, as far as it goes. Yes, the economy probably does play the largest role in determining elections--as perhaps it should, since politicians (especially Democratic politicians) persist in promising that they'll improve the economy and the lot of the middle class, and since they unfailingly take credit for a good economy even if they had little to do with bringing it about.

Brendan uses all the right qualifiers--"largely driven", "often is a reflection"--but these tend to obscure the fact that the state of the economy accounts for only something like 60% of the variability in election results. That still leaves a lot of variability, some of which is almost certainly accounted for by tactics, and some of which is surely accounted for by policy. And thank goodness for that. It would be awful if politicians could enact any policies they liked, no matter how unpopular those policies might be, and suffer no electoral consequences.

While I think that Brendan make a good point - there is no doubt many pundits are just making stuff up - I have my doubts about how robust any studies might be that BN places so much faith in re: the economy and mid-terms.

Is there enough data to even measure how much of a factor the economy has (re: Rob's 60% above) with any margin of error that isn’t enormous? I mean there are only 10 mid-term data points in the past 20 years, and any of these could have been influenced by extraordinary factors beside the economy, so how can you measure what is "normal"?

How do they measure "the economy" and what factors might be different now vs. the past.

And how do these models deal with the tails of the distributions, which is where the surprises are? ( 2008 financial melt-down!)

If they are indeed reliable, I am curious what Poli-sci models of expected congressional change of hands due to economy alone might predict vs. what actually occurs in November. Any excess might be attributed to other factors such as tactics, policy, etc and we would know whether it is the pundits many theories or Brendan's many warnings that should carry more weight.

Marty - see the forecasting articles for the '04, '06 and '08 elections here: http://www.apsanet.org/content_13020.cfm They're not perfect, but they generally do well, especially in comparison to tactics-based storytelling (which can't be quantified and thus has no predictive power).

ISTM that few, if any, conservatives are blaming Obama's drop in popularity to tactics. Conservatives mostly blame his actual job performance.

@Marty, you can also check out my way cool blog. I wish I had more time to dedicate to it.


The comments to this entry are closed.