The past few weeks haven’t produced much good news for Democrats’ hopes of retaining the Senate. While their position is far from catastrophic — the Senate playing field is broad this year, and the outcome of many races is uncertain — Democrats’ chances of keeping the Senate were down to 35 percent as of the FiveThirtyEight forecast late Wednesday afternoon.
Part of the problem is that Democrats are almost entirely playing defense, with few prospects to pick up Republican-held seats. Georgia, where we have the Democrat Michelle Nunn’s odds at about 30 percent, looks like their best opportunity. It’s also too early to foreclose the possibility of Democrats winning Kentucky, but Republican Minority Leader Mitch McConnell has not trailed in a nonpartisan poll since May, and two new surveys Wednesday put him ahead.
Kansas, however, had become an under-the-radar opportunity for Democrats. The Republican incumbent there, Pat Roberts, barely survived his primary and has extremely low approval ratings. Several recent polls had put the race in single digits between Roberts and his Democratic opponent, Chad Taylor, with the independent candidate Greg Orman getting about 20 percent of the vote. As of Wednesday, the FiveThirtyEight forecast gave Roberts an 80 percent chance of winning. That’s not bad, but it’s not any better than McConnell, who also has about an 80 percent chance of holding on in a race that has gotten far more attention.
Late Wednesday afternoon, however, Taylor announced his withdrawal from the race, setting up a contest between Orman and Roberts. (There is also a Libertarian candidate, Randall Baston, on the ballot.)
Why would Taylor leave the race right when polls showed it tightening?
Perhaps because he and Orman share a lot in common philosophically. Based on the ideological ratings we track (more background on those here), both Taylor and Orman rate as the equivalent of moderate Democrats. Orman, in fact, ran as a Democratic candidate for the Senate in 2008, although he withdrew from the race during the primary.
But Orman had raised more money than Taylor — about $625,000 in individual contributions to Taylor’s $120,000 as of July 13 — and probably had more momentum, having recently received endorsements from a bipartisan group of legislators.
There was also a recent survey, from Public Policy Polling (PPP), which showed Orman ahead of Roberts 43-33 in a potential two-way race. The same poll had shown Taylor trailing Roberts by 4 percentage points in the event Orman dropped out.
If the PPP survey is accurate, this is a huge problem for Republicans. Suddenly, they’re behind in a race against a former Democrat who might caucus with the Democratic Party should he make it to the Senate.
There is ample reason to be skeptical, however. Although our model largely relies on polls, it also evaluates other factors on a state-by-state basis, which we refer to as the “fundamentals.” One of these is the candidates’ ideology scores. In a deeply red state like Kansas, a fairly moderate Republican like Roberts rates as much closer to the median voter than a center-left candidate like Orman. Orman has also never been elected to office before, a factor which makes a candidate more likely to underperform in his polls, perhaps because of a lack of campaign experience.
In fact, our fundamentals-based estimate would put Orman 25 points behind Roberts — not 10 points ahead. It’s rare to see a discrepancy even half as large as that.
The fundamentals-based estimate is not so accurate, however. It won’t capture some of the more subtle qualities of the campaign, and on average it misses the final margin in the race by about 9 percentage points. By comparison, the average poll is off by something like 6 or 7 percentage points at this stage of a Senate campaign. The fundamentals estimate is mostly meant to serve as a stand-in when there is little or no polling data available. Otherwise, it doesn’t receive much weight in the model and mostly acts as a tiebreaker when the polls are very close.
In this case, however, we have only one poll of the Orman-Roberts matchup (and it was from the polling firm PPP, which has a series of methodological problems and which surveyed the matchup when it was still just a hypothetical possibility). So the model comes out somewhere between the survey and the fundamentals rating. It projects a narrow 2-point victory for Roberts, and gives him a 56 percent chance of winning against 44 percent for Orman.
For all intents and purposes, that makes the race a tossup. But it’s also a totally wild guess. The model is designed to recognize that the outcome is extraordinarily uncertain when the polls and the fundamentals diverge so much. So the margin of error on the forecast is enormous — the 90 percent confidence interval on the forecast runs from a 20-point Orman win to a 23-point victory for Roberts.
But if Roberts winds up beating Orman by a few percentage points, it wouldn’t be so surprising. That’s about the margin by which McConnell leads Democrat Alison Lundgren Grimes in Kentucky, after having trailed in several previous polls. McConnell, like Roberts, has extremely low approval ratings, but the partisan gravity of Kentucky may prevail in the end.
Another question is which party Orman might caucus with should he win. The default answer would be the Democrats. Orman was formerly a Democrat, he’s mostly taken the political positions of a moderate Democrat, and the Democratic candidate just dropped out of the race.
But the more Orman appears to be affiliated with the Democratic Party, the less attractive he might be to Kansas’s red-leaning electorate. Orman is suddenly about to become much better known to Kansas voters and to the national media. That could help him in some ways, but it will also increase scrutiny on his political stances. Furthermore, some voters, especially Republicans who voted against Roberts in the primary, might have considered voting for Orman as protest before — and they may reconsider those plans now that he has a chance to win.
If we do program the model to treat an Orman win as a Democratic pickup, then the Democrats’ chances of retaining the Senate would improve to 38 percent from 35 percent. We’re going to do some further thinking overnight about how to handle the case.