Representative James Lankford’s 23-point victory margin in last week’s Oklahoma Senate primary caught most pundits off guard. He had been opposed by leading “tea party” groups and leaders such as SCF, FreedomWorks, Ted Cruz, and Sarah Palin. He did not, however, get outside support from “establishment” groups like American Crossroads and the U.S. Chamber of Commerce. Neither fish nor fowl in this year’s “tea party versus establishment” trope, pundits quickly decided that Lankford must have been helped by his long association with a large Southern Baptist summer camp (Southern Baptists are easily the largest single denomination in the state) and by kind words from outgoing senator Tom Coburn. (see here, here, and here).
House majority leader Eric Cantor’s defeat also caught pundits off guard. Here was a “tea party versus establishment” race, but one in which no major tea-party group or politician supported the challenger. How could Cantor have lost to a political nobody? Pundits leaped into action, however, contending that Cantor lost because of a poor campaign, his support for some immigration reform, and his opponent’s endorsement by radio personality Laura Ingraham, among other explanations.
So what do these reactions to two very different races have in common? One simple fact: They have no data to back them up.
In the absence of lots of polling data, most pundits are rudderless when navigating the seas of post-election analysis. And when the polling is wildly off, as it was in the Cantor race, they find it even harder to explain what happened before their very eyes.
That wouldn’t happen if they just listened to Yogi Berra. “You can observe a lot just by watching,” the Yankee great said, and so it is with elections. Election results themselves can often give strong clues as to why a campaign turned out the way it did.
Take the Oklahoma Senate race, for example. The map of the results shows a very clear regional divide between the center of the state, which went heavily for Lankford, and the rest of the state, which either went for Shannon or was closely contested (Lankford is in blue, Shannon is in red):
If the “Baptists for Lankford” theory is true, we should expect Lankford’s support should rise as the proportion of Southern Baptists rise. And, thanks to the data available from the Association of Religion Data Archives, we can calculate exactly which counties fit that bill in the Sooner State.
Unfortunately for the pundits, the data do not support the “Baptists for Lankford” hypothesis. Baptists are strongest in Oklahoma’s east and south, but this is where Shannon was strongest. Moreover, Lankford received his strongest support in and around Oklahoma City, areas where Baptists comprise less than one-quarter of the population.
The Coburn explanation doesn’t bear up under scrutiny either. If Coburn’s de facto endorsement mattered, it should have mattered more in the area of the state that Coburn knew best, the eastern second congressional district where he first won election. Yet this was Shannon’s strongest CD and Lankford’s weakest.
So, why did Lankford win? The answer lies in comparing the map of the results:
With these three maps:
First, Lankford’s congressional district, the fifth district, based in Oklahoma City:
The three counties in this seat were also Lankford’s three best in the state. He received over 72 percent of the vote here, and that congressional district cast about 25 percent of the total statewide vote.
The second map shows the Oklahoma City metropolitan area:
Voters in these counties would have seen or read about Lankford in the prior four years in local newspapers or on TV news. They also would have seen his ads from prior races. The six counties in this area not in Lankford’s district were his six next best counties. He averaged over 64 percent of the vote here, and these counties cast another 16 percent of the statewide vote.
The third map shows Oklahoma’s media markets:
Focus on the large market in the middle, the OKC market, the one where voters would have seen Lankford’s ads and TV news appearances, and look again at the results map:
Lankford carried every county that’s in this market and outside of the OKC metro area often by margins of 20 points or more. Indeed, his support virtually stopped at the media-market line. Outside of the OKC market, he only carried four counties with the 50 percent or more needed to avoid a runoff: Tulsa (Oklahoma’s second big city, and one with a relatively low Baptist percentage); Washington (northern Tulsa suburb); Jackson (on the border of the OKC market); and Choctaw (Shannon is half Chickasaw and the Choctaw-Chickasaw rivalry is deep and long-standing).
These data suggest that Lankford won primarily because of regional loyalty. Simply put, he started from the largest base imaginable, people knew and liked him, and Shannon’s millions of dollars or negative ads could not change people’s opinions.
The same effect was also noticeable for Shannon. He carried his regional base with over 70 percent, but Comanche County (in the Lawton/Wichita Falls media market) cast only 2 percent of the statewide vote. The tiny Lawton media market overall went for Shannon, but the other counties in it cast less 2 percent of the total vote. Shannon, despite support from national groups, was fighting uphill from the minute he entered the race.
This regional effect has been observed in virtually every contested race this cycle. Indeed, it showed up even in the Oregon Senate primary between “establishment” Monica Wehby and “tea partier” Jason Conger. Conger represented a state-house district in Bend, the largest city in Deschutes County. Despite being unable to counter Wehby’s statewide TV buys, he carried Deschutes by 36 points.
So, what does this have to do with Eric Cantor? Simply put, the data show that he had no regional base. He not only failed to carry his home county of Henrico, but he lost most of the precincts in Henrico. Indeed, in Henrico’s 66 precincts, Cantor carried only three with over 60 percent. Jason Conger did better by far in his home county than Eric Cantor did in his.
Even though he was a twelve-year incumbent, even though he was the only candidate on the airwaves, even though he had all the advantages that incumbency can offer, when push came to shove, Cantor had no residual goodwill to fall back on.
The data from Oklahoma, Oregon, Mississippi, Georgia, and other contested races this cycle show that an incumbent who is known and liked can withstand vicious, well-funded attacks. The fact that Cantor could not withstand a campaign that had little money and little ability to communicate its message suggests that his goodwill had evaporated long before Dave Brat ever thought about throwing his hat into the ring.