The House Elections

People are always asking me, Which party is going to win the House elections? The answer is, I don't know. We have very many more publicly released polls than we used to have, and therefore more basis for making estimates. But they're still just estimates, subject to error. I've had a pretty good, but far from perfect, record of predicting election results in the past.

In any case, I decided to address the issue today, by looking at individual races, those with polling information and those without, and making my own best estimate of who would win if the election were held today. I used the Real Clear Politics list of 50 House races in play. And I decided to take the plunge of classifying each of the 45 seats in the list that are currently held by Republicans as sure Democratic, lean Democratic, lean Republican, or sure Republican. I didn't include a tossup category, which pretty much guarantees that some seats I place in the Republican column will end up Democratic and some I place in the Democratic column will end up Republican. I've consulted the polls but haven't treated them mechanically: Some pollsters I believe more than others, and I take some account of possible trends away from or toward an incumbent. Anyway, here goes:

There are five Democratic seats I'd rank as either sure Democratic (GA 8, VT 1, IA 3) or lean Democratic (IL 8, GA 12).

I will readily concede that many smart observers would put many of these races in different columns. An optimistic Democrat or a pessimistic Republican would probably switch several races from the Lean Repub to the Lean Demo column.

My predictions would produce an almost evenly divided House: 219 Democrats, a net gain of 16, and 216 Republicans. Such a result would raise the question of whether Mississippi Democrat Gene Taylor, who declined to vote for Nancy Pelosi for speaker in this Congress, would do so again, and whether another Democrat might do so—which could produce a Republican majority for speaker. My predictions also suggest, correctly, that I do not see this, at least yet, as a "wave" election. In a "wave" election, the winning party—Democrats in 1974, Republicans in 1994—win about half the districts they seriously contest, while the losing party wins about 10 percent of those they seriously contest (since the Republicans seem to be seriously contesting only five seats, this would give them at best one offsetting gain). If you count all these 45 Republican seats as seriously contested, this would mean that Democrats would gain only 36 percent of them. A "wave" result, which some are forecasting, would give Democrats a net gain of 22 or 23 seats, enough for a 225-210 or 226-209 majority.

One reason I do not see this election as a "wave" is that I think Republicans have a superior turnout program. The samples in most recent polls show a Democratic advantage in party identification—quite different from the 2004 exit poll that showed party identification at 37 percent Republican and 37 percent Democratic. I think there probably has been some shift in party ID since November 2004, but I doubt that it's as great as those polls suggest. In any case, polls are not good at predicting turnout. Some but not all polls show Democrats to be more "interested" or "certain to vote" or "motivated." But responses to those questions have not done a good job at projecting turnout in the past, including November 2004. To get a really good idea of turnout, I think we have to wait for elections—or, rather, for the vote to be counted.

By the way, if Democrats do end up with a majority of the magnitude of 219-216, we probably won't know it on election night. There will be some races too close to call, others where the absentee votes remain to be counted, and, as John Fund has suggested, others where the result will be litigated. Back in the old days, when the Congress elected in November of the even-numbered year did not begin its first regular session until December of the next odd-numbered year, it was sometimes unclear for months which party would control the House. I append the following paragraph from my 1990 book Our Country.

Today 1930 is remembered as the year in which Democrats began to enjoy their only briefly (in 1946 and 1952) control of the House. But this future was not apparent the morning after the 1930 election. Speaker [Nicholas] Longworth, who used to give Minority Leader [John Nance] Garner a ride to work in the speaker's limousine, wired Garner immediately after the election, "Whose car is it?" The answer was not clear even a year later. As it happened, the normal attrition of deaths, combined with a collapse of confidence in the economy, finally produced a Democratic majority. In July 1931, Congressman Bird J. Vincent died while sailing on the Henderson from Hawaii to the mainland. He represented the old lumber mill town of Saginaw, Michigan, the sugar beets and navy bean farmlands and a heavily Republican district that included Owosso, the hometown of future Republican presidential nominee Thomas E. Dewey. The district had been represented for many years by Joseph Fordney, who rose to chair the House Ways and Means Committee, and it had given Vincent 67% of the vote against Democrat Michael Hart in November 1930. But on November 3, 1931, Hart was elected to the vacancy. That gave the Democrats 217 seats. Three days later in San Antonio came the death of long-ailing Henry Wurzbach, the Republican congressman from the 14th district of Texas, which stretched from the hill country north of San Antonio through Texas German farming counties to the hot level farmlands along the Gulf of Mexico. Democratic Governor Ross Sterling, hoping to capture the speaker's office for his fellow Texas Democrat Garner, called a quick special election, and on November 24 the district elected Democrat Richard Kleberg, a member of the family that owned the giant King Ranch. (Kleberg hired as his chief assistant a twenty-three-year-old teacher from the northern end of the district named Lyndon Johnson.) There were still two other vacancies in the House, in historically Republican seats in New Jersey and New Hampshire where Democrats had gotten only 34% and 44% in 1930. Democrats won the special elections for both seats as well, in December 1931 and January 1932, and those victories together with their earlier win in Michigan were the first signs that they could win elections in the Republicans' Yankee heartland.

Presumably we won't have to wait 13 months this time to learn which party will control the next House.

UPDATE: I have been informed that Gene Taylor in 2005 said he would vote for Pelosi as speaker. But I wouldn't be surprised if Republicans sounded him out if it turns out that his vote would make the difference in who organizes the House.