All the talk about how Obama will win or lose based on the economy got me thinking:
Would it be possible, at any time in 2012, to simply look at one or a few or some economic numbers (e.g. unemployment, wages, whatever), and say, with any confidence, from that/those numbers alone, who’ll win the Presidential election?
I’d say that for certain past elections that would work, but with two wars going on, active terrorism, and what’s happening in Libya and Egypt and Tunisia, there’s too many things that could have an effect on the election.
Assuming you could hold every other number the same, sure. But you can’t, so no.
The Bread and Peace Model does fairly well by taking into account only wage growth and foreign troop casulties.
I suspect its still to symplistic, but I think the general idea is right. Americans vote primarily based on economic performance, most of the stuff talking heads spend their time talking about end up being irrelevant. For example, I seriously doubt the 2012 election will turn on Libya, Egypt, Tunisia, Terrorism or even Iraq/Afghanistan as Lamar suggests. At best, those will be second or third order effects compared to how voters feel their personal economic well-being is.
I’m still interested in thoughts on this question, especially given that no one seems confident that the economy will be noticeably better (and may even get a little/lot worse) by election time.
Could the Republicans nominate anyone and win if particular numbers are just “right”?
Ooh, just found a timely blog post Nate Silver explaining why using at least one single number (in this case, unemployment figures) is hazardous at best. He also gets a little into my basic question, I believe.
ETA: I also note that he mentions that, of course, unemployment DOES have some impact on reelection chances, but it may be only insofar as it’s a reflection of having/not having confidence in the future, and that the country is on the right track. But even THEN, it’s STILL complicated.
I think any number, even unemployment, is relative. The trend is probably much more important. Remember, Reagan got re-elected in '84 with over 7% unemployment but the trend was down.
Yes, and Silver’s article brings that up as a possible factor/vector.
Updating, because Silver spent some time today researching and analyzing this very model.
All in all, it appears to me that he is saying the answer to my OP question is “no.”
I saw a study once (yes, I know, don’t shoot me) that showed a strong correlation between incumbents’ winning and the direction of change in unemployment (not its actual value) during the primary season. A high number that decreases during the period when the most people are forming their impressions and making their decisions will give them more confidence in the incumbent, but a deteriorating economic situation, even if it’s still pretty good, will induce them to consider alternatives.
Yes; Silver actually analyzes that in my link in post 6. To quote him, “But it, too, is not always reliable.”
No, he analyzes the total change, direction and amount. It’s a related but different question to just ask “was the change positive or negative?”, and see if there’s a correlation there.
Of course, there’s also a risk of looking at many different comparisons: If you run 20 different comparisons, then you’re apt to find one that shows a 95% confidence level of being statistically significant, even if they’re all actually junk. It’s best to decide a priori which (few) comparisons you’re going to check, and if those don’t work out, to just end the work there.
Silver, from that link (underlining mine):
The study Elvis mentions (which I remember too, though I can’t put my hands on it either) was about the change in unemployment during the election year alone.
The thesis being advanced was that while, overall, the economy fared more poorly under GOP than Dem management, the GOP had a much higher incidence of getting the numbers moving the right way during election years, and had benefited substantially from that trend.