I did some looking around at DC's Political Report, to do some research on the races for this fall. They have a nice color-coded map for each category of races, but what I found especially interesting is the choice of information options. In my case, I want to make up my own mind, and so I prefer raw data if at all possible. DCPR provides its own information, but also links to many other sources. One thing I keep finding despite this wealth of information, whether from DCPR, Cook Political Report, Real Clear Politics, or any of the other services out there, is the lack of definitive data on the House races. Yes, a lot of these services will sell you a subscription which opens the polling they have on those races, but 'caveat emptor' folks, the polling is a bit less than trustworthy as a barometer, for all the reasons any poll should be taken with a measure of salt. In the first place, the respondent pool is often not identified to any great degree, the size is too small for satisfactory accuracy, the methodology is not transparent, and the polls are not taken often enough to establish a realistic pattern.

So, without the polls, what is there to ride on? Well, there's news and there's news. The first news is History, which is both bad and good. Incumbents tend to get re-elected, unless they trip badly in the public view. This is both bad and good, because it means that the arrogance of your average elected official tends to only grow with time and re-election, but it also suggests that the present Republican majority in Congress (231 Republicans, 201 Democrats, 1 Independent, 2 Ghosts) is more likely to continue than the press contends. Since every House seat is to be decided this fall, the Conventional Wisdom suggests that unhappiness with the Republicans could lead to a change in leadership, but a closer look does not support this contention. While the possibility of a Democrat takeover is possible, the chances of it happening is a bit less than they claim.

You can figure this out by simply scanning headlines. Basically, two things make for change in the House; scandal and retirement. Retirements affect individual seats, so we're really looking then on whether there is something to make the voters mad enough to make the switch to the other party. For that, you need at least two components - people have to be mad enough at the governing party to kick them out, and the other party has to offer some reason to believe they would be better. The 1994 'revolution' in the House was built on both factors, and so was the 1930 change. Most of the changes, especially in the House, have otherwise been gradual, and for the Republicans to lose control, they need to lose 15 seats of the 231 they hold, and they must lose them to Democrats. That is, 15 out of 231 and assuming no incumbent Democrat loses. If, say, 3% of the Democrats lose their seats, that's 6 seats so the Republicans would then need to lose 21 of their 231 for the Donks to gain control, or 9% of their position, three times the Democrats' loss rate. While President Bush has recovered steadily in Approval since May, the same is not true for Congress, and Democrats do not enjoy a significant position of confidence, relative to Republicans.