Will Democrats win control of the House in November? It’s a question lots of people have been asking in Washington and around the country these days. It seems possible, certainly. Democrats only have to make a net gain of 15 seats to win a majority. But it’s also true that, with the single and large exception of 1994, neither party has made a net gain of more than 10 House seats over the last 20 years.

I think there are two plausible hypotheses about how House elections work. If Hypothesis One applies, Democrats have a good chance at gaining a majority. If Hypothesis Two applies, they don’t.

Hypothesis One sees House elections as a referendum on the president and his party. If the president’s job rating is above 50 percent, his party tends to suffer only narrow losses or even, as in 1934 and 1998 – and almost in 1962 – makes gains. If the president’s job rating is significantly under 50 percent, his party tends to lose lots of seats.

The theory is plausible, and fits many election results over the years. Presidents Roosevelt, Kennedy and Clinton had high job ratings in 1934, 1962 and 1998. President Johnson in 1966, President Nixon in 1974, President Reagan in 1982 and President Clinton in 1994 had job ratings under 50 percent, and their parties all lost many seats in those years.

Currently, George W. Bush’s job rating is hovering around 38 percent. Under Hypothesis One, Republicans should lose lots of House seats – quite possibly more than the magic number of 15.

Hypothesis One was developed by political scientists and psephologists over many years. Hypothesis Two is one I developed myself, and it’s based only on the elections of the last 10 years. In the five House elections from 1996 to 2004, there has been very little variation in the popular vote percentages for both parties. The Republican percentage of the popular vote for the House has fluctuated between 49 and 51 percent, the Democratic percentage between 46 and 48.5 percent.

This has been true despite great differences in the job ratings of the parties’ leading figures. Republicans won pluralities of the popular vote for the House in 1996 and 1998,when Bill Clinton’s job rating was high and the favorability ratings of the highly visible Newt Gingrich were very low. Clinton’s job rating was high in 2000, too, but Republicans still won the popular vote 49 percent to 48 percent. In 2002, when George W. Bush’s job rating was up around 70 percent, Republicans won 51 percent of the popular vote for the House. In 2004, when his job rating was around 50 percent, Republicans won 50 percent.

These numbers seem inconsistent with Hypothesis One. How to explain them? We have a highly polarized politics that divides us along cultural lines. Those cultural divisions tend to be more important to voters than their ratings of presidents’ and parties’ performance. The polarization is exacerbated by the fact that Bill Clinton and George W. Bush both happen to have personal characteristics – I don’t have to spell them out, do I? – that people on the other side of the cultural divide absolutely loathe.

The slight uptick in Republican percentages in 2002 and 2004 can be explained by higher Republican turnout. Looking ahead to next November, there is reason to believe that the Republican base is turned off – by high spending, by immigration – and may not turn out as heavily. But if so, how much difference will that make?

Polls are not good predictors of turnout – only elections are. Last week, we had a special election in the 50th district of California, whose Republican congressman resigned in disgrace and went to prison. In 2004, the 50th district voted 55 percent for George W. Bush and 44 percent for John Kerry. Last week, the district voted 53 percent for Republicans (there were 14 candidates, the winner among whom goes on to a June 6 runoff) and 45 percent for Democrats. There were only two of them, and the leader, Francine Busby, got 44 percent of the vote – the same percentage as Kerry. That may be 1 percent higher when the last absentees are counted.

These results are inconsistent with Hypothesis One. They’re consistent with Hypothesis Two. Republican turnout was down more than Democratic turnout, but only very slightly. Of course, things may change by November. But it looks like Hypothesis Two is still in force, and if so, Democrats will have a hard time winning control of the House.