This is an outgrowth of a
diary written a couple weeks ago by Superribbie, one of our in-house election experts. That diary looked at the correlation between a president's approval rating and his party's performance in midterm elections. This diary extends the analysis a bit further.
Update [2006-5-10 21:41:38 by dmsilev]: Added a reference line to the first graph showing the boundary between (D) and (R) control of the House. If we believe these graphs, Bush will need to be in the high 40s or low 50s for the GOP to keep the House.
See the original diary for the actual numbers of approval vs. seats won/lost in the midterm House elections. I've summarized those results in this graph:
The large points are midterms from 1960 onwards; smaller points are earlier ones where approval-rating polling was less accurate (1974 uses an average of Nixon's and Ford's approvals) As you can see from the solid red line, there is a very strong correlation between a President's approval rating and how well his party does in the midterms. If you fit over all data, including the earlier ones, there's still a corrrelation, shown by the dotted line, but it isn't as good. Since Bush is currently exploring new depths for a midterm President, I've extended the trend line down to Cheney-esque territory.
So, where does that leave Bush and the GOP? Well, we know what his approval rating is now, but where will it be in 6 months? There's no way to know for sure, of course, but we can do the next best thing, which is make a guess. To do this, I've looked at the most Bush-friendly pollster out there, which is Rasmussen, for the last few months:
What we see for the data since February is a roughly linear drop in approval and increase in disapproval. Both numbers are changing at a rate of a bit over 2 points per month. If we make the (totally unjustified) assumption that these trends will continue, we see from the graphs that by early fall, he'll have an approval rating in Rasmussen of about 34% (or about 28ish percent in most every other poll). By the election, he'll be around 30%.
So, let's connect the dots. We're piling extrapolation on top of extrapolation here, but the naive interpretation of the data is that the President can be expected to have an approval around 30%, and such a President can expect to see his party lose
70 seats
in the midterm elections.
Now, I don't for a minute believe that we're going to gain 70ish seats in the House. Gerrymandering will see to that. But, the take-home lesson of these plots is that Presidential approval matters, and the GOP is laboring under a really heavy structural disadvantage going into November. The President is the public face of his party, and this President is really really unpopular.
One other thing: To keep control of the House, the Republicans have to limit the bleeding to be less than 15 seats. Historically, that means a President with an approval rating in the high 50s. It's been a long time since Bush was anywhere near that level.
-dms