Dem
52
Ties
1
GOP
47
We had an interesting question from reader J.W. in Los Angeles, CA, and we decided that it was meaty enough that we'd turn it into a regular-week item, instead of a Q&A answer. Here's the question:
It is generally understood that it can be easier to flip an open seat that is up for election. So it may be easier for Democrats to flip North Carolina because Sen. Thom Tillis (R) didn't run and the seat is open. So how does this play out in Texas, where the incumbent was defeated in the primary election? Is that now considered an open seat and thus potentially easier to flip? Are there examples of this scenario in the past and what were the outcomes?
We'll start with the easy part: It is indeed an "open" seat now, since the incumbent is not up for reelection. It's not voluntary in this case, but an involuntary retirement is still a retirement.
"This scenario" would seem to be "an incumbent loses his or her primary." This is not a common outcome, but it's not utterly unheard of, either. Here is (we believe) a complete list of all the senators who were defeated in a primary in the last half-century:
Clearly, it's become much rarer for incumbent senators to be primaried. That is likely because it's much easier today to be aware that the writing is on the wall, and to jump ship before being thrown overboard.
From the list, we will cut the senators who were appointed. We are also going to remove the two senators who were primaried, ran as independents, and won a three-way race (Murkowski and Lieberman). Finally, we will remove Arlen Specter, who changed parties shortly before the election in which he was primaried. After all those cuts, the list is:
That leaves us with nine cases, from the past half-century, that pretty closely parallel what is happening this year with Sens. John Cornyn (R-TX) and Bill Cassidy (R-LA). That is to say, a senator was elected in their own right, and then was unseated by a challenger who went on to contest a (de facto) two-way race.
Now, let us compare the results for the primaried senator with how the candidate who replaced them did. Of course, senators run alternately in presidential years and off years, so we'll go give data for two elections where it's appropriate (E-1 = "one election before being primaried" and E-2 = "two elections before being primaried"):
| Year | Senator | E-2 Margin | E-1 Margin | Replacement | Replacement Margin |
| 1978 | Clifford Case (NJ) | +23% (60% to 37%) | +28% (62.5% to 34.5%) | Jeff Bell | -10.2% (43.1% to 53.3%) |
| 1980 | Mike Gravel (AK) | 7.7% (45.1% to 37.4%) | 16.6% (58.3% to 41.7%) | Clark Gruening | -7.8% (45.9% to 53.7%) |
| 1980 | Jacob Javits (NY) | 17% (49.7% to 32.7%) | 7% (45.3% to 38.3%) | Al D'Amato | 1.4% (44.9% to 43.5%) |
| 1980 | Donald Stewart (AL) | N/A | 12% (55% to 43%) | Jim Folsom Jr. | -3.2% (47.1% to 50.3%) |
| 1980 | Dick Stone (FL) | N/A | 2.5% (43.4% to 40.9%) | Bill Gunter | -3.4% (48.3% to 51.7%) |
| 1992 | Alan Dixon (IL) | 13.8% (56.3% to 42.5%) | 31.4% (65.1% to 33.7%) | Carol Moseley Braun | 10.2% (53.3% to 43.1%) |
| 2002 | Bob Smith (NH) | 33.8% (65.1% to 31.3%) | 3.1% (49.2% to 46.1%) | John Sununu | 4.4% (51.1% to 46.7%) |
| 2010 | Bob Bennett (UT) | 31% (64% to 33%) | 40.3% (68.7% to 28.4%) | Mike Lee | 28.8% (61.6% to 32.8%) |
| 2012 | Dick Lugar (IN) | 34.7% (66.6% to 31.9%) | 74.8% (87.4% to 12.6%) | Richard Mourdock | -5.6% (44.4% to 50%) |
For the seven senators who were twice-elected before being primaried, the average margin of victory was 23 points in the election 12 years prior to their primary defeat. In all seven cases, the primary victor underperformed the result from 12 years earlier once they advanced to the general.
For all nine senators, the average margin of victory was 24 points in the election immediately before they were primaried. However, that is skewed by the fact that the Democrats did not field a candidate in Indiana in 2012. If we toss out Dick Lugar's easy win over an unknown Libertarian, the average margin is 17.6 points. In every case save one (John Sununu in 2002), the primary victor underperformed the result from the previous election once they advanced to the general.
For all nine seats, the average margin in the victory where the sitting senator was primaried was 1.6 points. That's 16 points fewer than the non-Lugar average from the previous elections, 22.4 points fewer than the with-Lugar average from the previous elections and 21.4 points fewer than the average from the elections 12 years prior to the primary defeat. In five of nine cases, the challenger who was successful in the primary went on to lose the general election.
It's only nine cases scattered over half a century, so you don't want to draw conclusions too firm. And undoubtedly there are many possible reasons that an incumbent would lose to a primary challenger, who then might lose in the general: a party in disarray, an insurgent fringe that's big enough to swing a primary but not enough to swing a general, some sort of scandal, partisan realignment, etc. Still, the data here gives no reason to question the conventional wisdom that a non-incumbent is noticeably weaker than an incumbent. And it may be that an insurgent primary winner may be weaker than a garden-variety candidate, by virtue of (generally) having gone scorched earth so as to take down the king of the hill. Whatever the extent of these phenomena, it may not be enough to flip Louisiana this year. But it's another reason to think that James Talarico (D) has an excellent chance of beating Ken Paxton (R) in Texas. (Z)