Many of you may have seen yesterday’s New York Times article here analyzing the 2008 Democratic vote in House races. The Times .would have you believe that because many Democratic congressional candidates received more votes than Obama did, this indicates a stronger pool of Democratic candidates who should be less vulnerable to Republican attempts to capture these Democratic House seats in two years. (They also acknowledge that it might make it more difficult for Obama to count on House Democrats’ support, but this point is downplayed.) As with any NY Times articles that go beyond reporting the news to interpreting it, you should view this argument with skepticism, for several reasons. Let me address them here.
In theory, comparing how well a presidential candidate does in each congressional district is a useful tool for assessing presidential “coattails”. A president who runs ahead of the congressional member in her district can be expected to have greater leverage with that member once in office. He can tell her, “Your constituents like me better than you, so if you value your seat you’ll support my policies.” However, despite its claim that it is assessing the relative strength of House Democrats versus Obama, the Times article does not actually provide this data. If you go to the source of the data (see here) referenced in the NY Times article, you will see that the data actually compares how well Obama did in the entire state to how well the Democratic candidate did in their single congressional district. This is an apples-to-oranges comparison that is not a very useful measure of Obama’s “coattails”. What we would really like to see is how well Obama did in each congressional district.
But even if the state-to-district comparison was meaningful, the interpretation the NY Times give is suspect. Rather than a sign of Democratic congressional strength, we might equally view it as a sign of Obama’s electoral weakness – in the closest House races, he ran behind the winning candidate. That is, he lacks substantial coattails. Indeed, that is how political scientists usually interpret instances in which the president receives fewer votes than does his party’s candidate in the congressional district. Because the member of Congress is more popular than the president, she feels less obligated to follow his lead.
In any event, in the absence of additional information, such as the location, previous results, and margin of victory for each of these “competitive” races, we cannot tell very much about the likelihood that these candidates will win reelection in the 2010 midterm elections. Historically, of course, the president’s party tends to lose seats in the first midterm election after the election of a new president. The recent exception was 2002, when George W. Bush’s Republican Party gained 7 House seats. Prior to that time, however, every newly elected president dating back to Harry Truman lost seats in their first midterm election – indeed, the average loss is 24 House seats!
In short, we cannot be sure that because Obama ran behind House congressional Democrats, those Democrats are a stronger than usual pool of candidates. To take an extreme historical comparison, Bill Clinton ran behind every single House Democrat in 1992. By the NY Times’ reasoning, that would suggest that 1992 saw a boatload of strong Democrats elected to office. In fact, the Democrats lost 52 seats and control of the House in the Republican wave of 1994.
There is an additional complication. House races are not simply a reflection of national forces – they usually turn much more on local issues and specific House candidate qualities. Indeed, political scientists have tried statistically to separate the national and local component of the congressional vote in previous elections, and we find that local factors are almost always a greater influence than national factors in explaining the congressional vote. (For the statistically interested, what we do is regress the current House district vote on the previous House district congressional and presidential votes, and use the coefficient for the presidential vote as a proxy for “national” trends.) I am currently writing an article that assesses the relative importance of national and local forces in the 2008 congressional races, and I will report the results here when I calculate them. Suffice to say, however, if history is any guide, local factors will be of greater importance than any national tide in explaining the outcome of most congressional races.
Without additional data, then, and a more sophisticated effort to tease out the interplay of local versus national forces on the 2008 House races, it is difficult to assess whether Democrats have in fact elected a stronger pool of candidates than in previous years. But, contrary to what the Times suggests, we shouldn’t assume based on this data that Democrats are better positioned to buck the historical trend in which the president’s party loses House seats in the first midterm election. And we cannot yet assess the length of Obama’s coattails, if any, in the 2008 election. When I get that data, you’ll be the first to know.