I appreciate Nate Silver’s characteristically thoughtful reply to my NRO critique of a post he made on the interplay between what the generic ballot says about the national House vote and the number of districts the GOP might be expected to gain on Election Day. Since he has agreed with the major thrust of my critique — that an individual candidate’s name ID greatly effects the result in any district — I presume he will not disagree with my subsidiary point, that we should take current polls showing better-known Democratic incumbents leading their GOP challengers at this point in the campaign as not indicating the final result.
Generic ballot results do not predict the size of a final victory with any degree of linearity. Thus, a simplistic model (unlike Nate’s) that relies purely on a generic ballot average to predict how many seats the GOP will gain is intrinsically flawed. One can see this by reference to a recent election Nate and I both followed, Australia’s. On Election Day, the conservative Liberal-National Coalition gained 2.6 percent on their total in the last election, and a whopping 4.9 percent in Australia’s most populous state, New South Wales. Applying that swing to the individual House seats in NSW would have led you to predict a Coalition victory. Instead, Labor prime minister Julia Gillard hung on because the Coalition did not get that large swing in four key NSW marginal seats. The Labor incumbents ran well ahead of their national party, saving their party from ignominious defeat.
So, should American conservatives begin to worry the expected wave will not materialize? I think not, based on the limited data we now possess.
Look at any of the leading prognosticators’ list of House seats at play. For ease of comparison, let’s use the nice aggregate list at KeyHouseRaces.com. Compare these to demographic data available from the Census Bureau, and you’ll find that a disproportionate number of these districts are populated by whites without a college degree. These folks are much more affected by the recession than others and have higher unemployment rates (though not as high as that of blacks and Hispanics). We know from Gallup data that President Obama is least popular among people without college degrees (42 percent) and whites (37 percent). Since he remains highly popular among African Americans (88 percent), who are likelier to be without a college degree, and among post-graduates (54 percent), who are likelier to be white, we can infer that the president’s standing among whites without a college degree is stunningly low. And unless that changes, it should lead to a higher swing in congressional districts heavily populated by this demographic group, leading to Republican gains.
One can see this trend at work anecdotally. People have been surprised by the rise of IL-17 (Phil Hare) on prognosticators’ target lists. This district was carried by Obama with 57 percent and by John Kerry with 51 percent. But it is 85 percent white and only 17.5 percent of the population has a college degree. Just yesterday a poll was released showing Gene Taylor (MS-4) leading his challenger by only 44 percent to 41 percent. This race had not yet appeared on any person’s prognostication list, but it should have. Seventy-one percent of the population is white, only 18 percent have a college degree, and it’s Trent Lott’s former CD to boot. This is prime demographic territory for a GOP surge.
I could go on ad nauseum, but let me conclude with two observations. First, there’s a long five weeks to go, and we know that individual campaigns and national dynamics can change. Many Democratic incumbents could outrun their national party, à la the Australian example. If the generic-ballot totals slide closer to 50-50, the Democrats could keep the House; if we continue to see GOP generic-ballot leads of four or more, expect a disproportionate effect in white working-class districts and, consequently, larger-than-expected GOP gains.
— Henry Olsen is director of the National Research Initiative at the American Enterprise Institute.