An Anti-Democratic Year

An Anti-Democratic Year

By Sean Trende - November 4, 2010

Earlier this year, I posited three outcomes for this election. In order from smallest potential Republican gains to largest, they were: an anti-incumbent year, an anti-liberal year, or an anti-Democratic year. I explained:

Although 1994 and 2006 are generally lumped together as wave elections, they were actually two very distinct phenomena. As I noted above, many Democrats in Republican-leaning districts thrived in 1994.

But this was not the case for Republicans in 2006. Heading into those midterms, there were 27 Republicans who represented Democratic-PVI districts. After the midterms, there were 14 (today there are 8, using the 2004 PVIs). That's more than half of the Republicans in Democratic-leaning districts losing, versus about a third of analogous Democrats losing in 1994.

The most interesting thing about 2006 is that even the Republicans who survived saw a marked decline in their performances. The only Republicans representing Democratic districts who didn't have a close call in 2006 were Mike Castle (DE-AL), Bill Young (FL-10), Frank LoBiondo (NJ-02), Jim Saxton (NJ-03), and Peter King (NY-03); even they saw decreases in their performances from 2004. Again, this contrasts with 1994, when half of the Democrats in Republican-leaning districts who opposed Clinton's agenda saw their percentages increase from 1992, despite the substantial national shift against Democrats.

It is obvious that this is not a generalized anti-incumbent year; only two Republican House incumbents lost, both of whom had been elected in fluky circumstances. Nor was it simply Democrats who had supported the President's agenda who bore the brunt of the wave. Instead, it is apparent that we experienced a substantial anti-Democratic wave last night.

Democrats saw their margins fall across the board. The only Democrat in a swing- or Republican-leaning district (D+5 or less) who saw his or her margins increase over their 2008 numbers was Maine's Chellie Pingree. Everyone else saw their margins fall. Liberal stalwarts like Alan Grayson and Carol Shea-Porter lost, but so too did highly conservative Democrats like Gene Taylor and Bobby Bright.

We can see how different this year is from 1994 by running a simple regression analysis. Last year, I explained that in 1994, voting for controversial legislation made a big difference in whether or not a Democrat won or lost. In Republican-leaning districts, voting for the Clinton budget or the assault weapons ban cost Democrats approximately four points for each vote off of their 1992 margin.

I ran a similar regression analysis this year. This election was a bigger wave, so instead of testing only those Democrats in Republican-leaning districts, I expanded the analysis to include all Democratic incumbents in D+5 districts or better who had major-party opposition in 2008. My dependent variable was the change in the incumbents' vote share from 2008 to 2010. In other words, I'm trying to explain what caused each individual Democrats' vote share to drop from 2008 to 2010.

I tested nine variables: (1) The incumbents' share of the vote in 2008 (if an incumbent was at 80 percent, there's more room to fall than if they were at 51 percent); (2) PVI (roughly how Republican or Democratic a district is at the Presidential level); (3) African-American percentage (to test the effect of the Obama '08 surge); (4) a flip from No-to-Yes on health care; (5) a yes vote on health care either time it came up; (6) a yes vote on cap-and-trade; (7) a yes vote on the stimulus; (8) an improvement in challenger experience (measured on a five-point scale); and (9) the effect of each dollar in additional funds raised by the challenger.

Overall, this model explains 65 percent in the variance in an individual incumbent's vote share from 2008 to 2010. In other words, if one incumbent saw his vote share drop 10 points from '08, you can generally ascribe most of that to those nine variables alone. All of the variables point the way we might suspect, except for African American vote percentage, which appears to have improved an incumbents' vote share from 2008 to 2010. This is something of the opposite of what we might expect in a midterm year.

But what is interesting is that the effects of individual votes this year are not particularly strong. This is in stark contrast to 1994. This year a vote for the stimulus had no significant effect on an incumbents' vote share (not entirely surprising, given that almost everyone voted for it). Cap-and-trade cost an incumbent a point, but we're only about 63 percent sure of that. And a vote for the health care bill only cost an incumbent Democrat in a swing- or Republican-leaning district 1.3 points (we're 73 percent sure of that). Again, these types of votes cost an incumbent about 4 points in 1994, and were strongly significant.

What variables were significant? Three. For every point in the two-party vote an incumbent received in 2008, we're almost positive (t=10.5) he lost about .68 points this year, suggesting a substantial reversion-to-the-mean effect after two wave elections. And, interestingly, while a vote for the health care bill didn't cost an incumbent all that much - in fact, we're only about 77 percent sure it cost an incumbent at all, a flip from "no" to "yes" on the health care bill cost those incumbents 5.5 points.

But the most interesting results was that, for every point in a district that John McCain won in 2008, the Democratic candidate lost about a half-point off of his 2010 vote share (t=4.9). This variable wasn't even significant in my 1994 analysis.

This suggests that, much as 2008 was the year that Democratic-leaning districts determined that they could no longer afford to send Republicans to Congress, no matter how moderate, 2010 is the year that the Republican-leaning districts made the inverse decision. It didn't matter whether the challenger had experience, or if he raised a lot of money. Neither of those variables is statistically significant.

This doesn't mean that this election was all about the economy. Far from it. As Jon Chait pointed out, a statistical model based only on economic/structural facts suggested that Democrats should lose only about 45 seats. This is only one model and it suggested the worst-case scenario for Democrats; other "structural" models put the number in the 20s and 30s. In other words, the Democrats' losses were about 20 to 40 seats in excess of what we would expect from the effects of the economy and back-to-back wave elections. Remember, Reagan's Republicans encountered similar economic headwinds in 1982, and they lost 26 seats, a baker's dozen of which can be directly chalked up to the intervening redistricting.

Rather, it suggests that voters in swing- and Republican-leaning districts decided that they disliked the Democrats so much that it didn't matter whether a candidate supported the President's health care bill or the stimulus. They voted against them anyway.

This year could have been worse for Democrats, but the preparations that they did for a year like 1994 probably prevented them from enduring a worse wipeout. The opposition research they performed and work they did shining the spotlight on their opponents' flaws - things they didn't do a good job on in 1994 - shows up in the results.

The most high profile races: Senate and gubernatorial races, were bad for Democrats, but not catastrophic. They even overperformed to some degree. In House races, it was a bloodbath, but a number of Democrats with flawed challengers eked out a win.

But in the state House and Senate races, where it is very hard to shine the spotlight on the challenger, the races were a complete debacle for Democrats, who are now at their lowest levels since the 1920s. If this had been repeated at the federal level, Democrats would have probably lost an additional 20 House seats and 2-3 Senate seats. Of course, with the power over redistricting Republicans have gained, at least at the House level, Democrats might have merely put their day of reckoning off for two years.

Sean Trende is senior elections analyst for RealClearPolitics. He is a co-author of the 2014 Almanac of American Politics and author of The Lost Majority. He can be reached at Follow him on Twitter @SeanTrende.

The Tired Party
Rich Lowry · November 7, 2014

Sean Trende

Author Archive

Follow Real Clear Politics

Latest On Twitter