Let’s look at seats held by Democrats first. The following graph compares the percent of votes received by the Republican to the estimated presidential approval rating in a given district for districts held by Democrats. Estimates of Bush’s approval ratings were calculated as discussed in this diary, and have an error of approximately plus or minus five percentage points.
f all those and only those who approved of Bush voted for Republicans, and all approval estimates were correct, all the dots would fall along the solid line, which marks a 1:1 ratio. The dashed line is there to give an approximation of the highest possible approval rating, given the error involved in calculating the approval rating.
Many points fall below the solid line. This means many people who approved of Bush voted for their incumbent Democratic representative anyway, showing the power of incumbency. This power of incumbency results in a lack of qualified opponents and/or increased loyalty to the current representative.
Those who do not approve of Bush, did not vote for a Republican when a Democrat was an incumbent. The same is true for open seats held by Democrats in the previous Congress. The lines are therefore an estimate of the maximum support possible for Republicans in Democratic-held districts.
There are three districts that are above the dashed line and do not follow these rules very well – IN-7 and IA-3, both with incumbents with severe recent illnesses, and WI-2, which has an incumbent who is openly gay.
If similar dynamics are in play for the 2008 elections – record low approval ratings for Bush, and a strong connection between Bush and the Republican party, we would expect to keep all the seats currently held by Democrats in the House, except in unusual circumstances.
Republicans
Here’s a similar graph for Republicans. I’ve added a new line that shows the vote Bush had in 2004 in each district. Incumbent Republicans ought to do at least as well as Bush did in 2004. That a large number did not is telling.
Again, if all those and only those who approved of Bush voted for Republicans, and all approval estimates were correct, all the dots would fall along the solid line, which marks a 1:1 ratio. The dashed line is there to give an approximation of the lowest possible approval rating, given the error in estimating the approval rating.
Many points fall above the thin solid line. This means many people who disapproved of Bush voted for their incumbent Republican representative anyway, showing the power of incumbency on the Republican side. This power of incumbency was lessened, however, as we might also expect the baseline for local Republican support to be not the approval of Bush this year, but the vote for Bush in 2004. The thick purple line shows Bush’s 2004 support. As we can see, Republican support was not typically much above this line.
Those who did approve of Bush, did vote for a Republican when a Republican was an incumbent. The same is true for open seats held by Republicans in the previous Congress. The black lines therefore are an estimate of the minimum support possible for Republicans in Republican-held districts. Competitive districts generally fell between this minimum and the 2004 support for Bush.
There are three districts that are below the dashed line and do not follow these rules very well – OH-18 and TX-22, both with incumbents who did not run for reelection because of corruption charges, and IN-8, the ‘bloody eighth.’ WY-AL is also marked; ID-1 is the dot right above it.
If similar dynamics are in play for the 2008 elections, we would expect every Republican seat to be potentially competitive, depending on recruitment and funding on the Democratic side (and of course other factors). Remember, current Bush approval levels are lower than those used to make these graphs, so the minimum Republican support line has fallen.
All the data for the 2006 House elections plotted together can be seen here.
(Note: You can view every article as one long page if you sign up as an Advocate Member, or higher).
