US Congressional elections: Why the House vote is difficult to predict

The U.S. Federal system divides power in many ways but few are more important than the geographic isolation of the 435 representative districts that enable minority political views to find a home within a majority dominated state.  No large U.S. state is political monoculture. Even the most Democratic or Republican state at the Senate or Presidential level has Representatives from the opposition. 

The American electorate varies greatly from urban centers through the suburbs out to large thinly populated rural areas. In general rural voters elect Republicans and urban voters Democrats. These areas are each party's geographic base.

But there are a group of districts, often in the suburbs and at the juncture of rural and urban influence that vary election to election.  In so-called wave elections, where one party dominates such as the 2010 House contest, where the Republicans won 63 House seats these swing districts are predicted by the national returns. But in close elections each of these districts is a contest to itself, where the qualities and history of the individual candidates are as or more important than the national politics.

California is solidly and routinely Democratic at the national level. It has not voted for a Republican presidential candidate in a generation, since George H.W. Bush won in 1988. It has not elected a Republican Senator since 1988.  Yet of California’s 53 Representatives 14 are Republican. Texas has not voted for a Democratic President since Jimmy Carter in 1976. It last elected Democratic Senator in 1988.  It has 11 Democratic Representatives out of 36. 

All indications point to solid returns in each party's geographic base. Normally Republican and Democratic districts will likely vote their history. The House will be lost or won in the middle districts where America’s nearly divided politics will play out.

RealClearPolitics (RCP) lists 205 seats leaning or solidly Democratic and 199 for the Republicans. In this accounting the Democrats need just 13 more seats to reach 218 and the majority.

RCP has 31 House seats as undecided, 30 currently held by Republicans and 1 by a Democrat.  Of these 30 Republican seats 16 voted for Clinton in the 2016 election, 14 for Trump. The one Democratic seat voted for Clinton. If those 17 seats keep their Presidential loyalties Democrats have a majority of 4. Even if they win none of the remaining 14 Republican seats.

This would appear to be almost a lock and thus the 75 percent to 85 percent chance given by many analysts for a Democratic House.

There are three problems with the appearance of security in this analysis.

The first is that in many of these districts very little polling has been done. Unlike the Senatorial races which often, have 5 or 10 polls to average, many of these districts are based on one or two polls or none.

Second many of the polls that are available are badly out of date. The contentious Senate hearing for Supreme Court Justice Brett Kavanaugh and the violent Democratic campaign against his nomination, which concluded with the vote on October 6^th, seems to have effected a large Republican backlash. Several Senate races have moved toward the Republican since that event. There is no reason to think that the same movement would not have happened in many of smaller districts.  Put simply any poll prior to October 6^th has little value unless confirmed by subsequent numbers.

Third is the reliance on national polls to provide insight into local races. It is not the national score that matters in the House races but the local opinion. For instance, President Trump’s approval rating is 44.2 percent nationally. But that is skewed by large majorities in population centers like California and New York. It does not have much relevance for voters in suburban or rural areas where his approval is often very different.

The same is true for the generic vote, which asks which party the voter would rather have in control of the Congress and Democrats lead by 7.7 percent. The national result may bear little on the local opinion and it is the local judgement which elects Representatives.

And finally, these caveats do not cover the problem pollsters appear have in getting an accurate reading of the President’s support in places where it is strongest.

Into the final two weeks Democrats and Republicans, as portrayed in their campaign rhetoric, have agreed on one thing. The election is indeed about President Trump.

On November 6^th a surprisingly few people in a small number of districts scattered around the country will decide which view of the President controls Congress.   

U.S. Congressional election update: Is geography destiny?

U.S. Congressional election: Can the Republican economy hold off the Democrats?