Home 2018 Elections Grading the 2018 Election Predictions, Polls in Virginia and Nationally, From “A”...

Grading the 2018 Election Predictions, Polls in Virginia and Nationally, From “A” to “F”

4

See below for my grades, in descending order from “A” to “F”, for pre-election predictions made by individuals and organizations that covered Virginia. Personally, I took a vow not to make any more predictions following 2016, but generally, I agreed with Ben Tribbett, Chaz Nuttycombe and Larry Sabato when it came to Virginia, and strongly DISAGREED with Trumpster/right-wing radio host John Fredericks. I was hoping that Tim Kaine would beat Corey Stewart by 20 points or more, while in the end Kaine beat Stewart by 15.7 points – but regardless, he won big. See how everyone did, below.

A Ben “Not Larry Sabato” Tribbett’s final Virginia House ratings were strong, with VA-10 a “Likely Dem Pickup,” VA-02 “Leans Dem Pickup,” VA-07 as “Tossup” (that one, in hindsight, probably should have been about the same as VA-02, “Leans Dem”) and VA-05 (“Slight-Lean R”; probably should have been “Leans R”). According to Ben, if Dems won VA-10, VA-02 and VA-07, we’d pick up around 40 nationwide. That’s basically correct, with Dems on track to pick up around 36-39. Ben also had Kaine crushing Corey, of course. Nice job!

A Chaz Nuttycombe had Tim Kaine winning by 21.3 points, which was about 5.6 points too optimistic, but still in the right direction. On VA-02, Nuttycombe had it “Luria by 1.5% to 4.5%,” which was right on the money. On VA-05, he wrote, “I’m honestly torn, I’m thinking Riggleman wins but my gut says Cockburn by <1.5%.” His head was right on this one, his gut wrong. On VA-07, he predicted a Spanberger win by <3.5% (yep!). And on VA-10, he had it “Wexton by 10-15%,” which absolutely nailed it. Nice job!

A Sabato’s Crystal Ball’s predictions were solid this time around. As they wrote on Wednesday, “the real-time seat projections from both the New York Times and FiveThirtyEight were suggesting that the Democrats would win a 229-206 majority in the House, for a net gain of 34 seats, exactly the seat change we picked in our final selections.” As for the U.S. Senate, the Crystal Ball’s final 2018 forecast had a net Democratic loss of 1 seat, which currently seems possible, IF Bill Nelson somehow comes out ahead in the Florida recount. The Crystal Ball forecast Dems winning Arizona (looks like that’s going to happen), Nevada (yep), West Virginia (yep), Montana (yep), Florida (???) and losing North Dakota (yep), Missouri (yep), Indiana (yep), Texas (yep) and Tennessee (yep). So…yeah, not bad at all – Crystal Ball no longer “shattered,” as it was after 2016.

Here in Virginia, the Crystal Ball had Tim Kaine as “Safe D,” of course. In the House, the Crystal Ball had VA-10 as “Leans Dem” (should have been “Likely Dem” IMHO), VA-02 and VA-07 as “Leans Dem” (nailed those!), VA-05 as “Leans GOP” (yep!) and no other race as competitive (bingo!). So…overall, great job by the Crystal Ball this cycle!

A- Overall, 538.com’s final predictions for the U.S. House of Representatives nailed it, with an average Democratic gain of 36 seats (currently, we’re at +32 seats, with at least four more likely to be called for the Democratic candidates). Here in Virginia, 538.com did pretty well overall, predicting:

  • Tim Kaine would beat Corey Stewart 56.9%-41.1% (+15.8 points). The final results were Kaine 56.9%-Stewart 41.2% (+15.7 points). NAILED IT!
  • VA-01 would go 58.5%-41.5% (+17 points) for Rep. Rob Wittman (R) over Democrat Vangie Williams. The final results? 55.3% Wittman-44.7% Williams (+11 points), so Wiittman DID win by double digits as 538 predicted, but Williams did six points better than the 538.com prediction.
  • VA-02 would go 50.6%-49.4% (+1.2 points) for Rep. Scott Taylor (R) over Democrat Elaine Luria. That prediction was off by several points, and also got the winner wrong, with Luria beating Taylor 51.1%-48.9% (+2.2 points for Luria).
  • VA-05 would go 51.5%-48.5% (+3 points) for Republican Denver Riggleman over Democrat Leslie Cockburn. The actual results had Riggleman defeating Cockburn by nearly 7 points (53.3%-46.7%).
  • VA-06 would go 60.7%-39.3% (+21.4 points) for Republican Ben Cline over Democrat Jennifer Lewis. They were very close on this one, with Cline beating Lewis 59.9%-40.1% (+19.8 points). Basically nailed it.
  • VA-07 would go 48.8%-48.5% (+0.3 points) for Democrat Abigail Spanberger over Rep. Dave Brat (R). The actual results were Spanberger 50.3%-Brat 48.5% (+1.8 points). So…they got the winner, but slightly underestimated her victory margin. Not too bad.
  • VA-09 would go 60.5%-39.5% (+21 points) for Rep. Morgan Griffith (R) over Democrat Anthony Flaccavento. The actual results were Griffith 65.3%-Flaccavento 34.7% (+30.6 point for Griffith), so actually 538.com was significantly (nearly 10 points) too optimistic for Flaccavento.
  • VA-10 would go 53.6%-46.4% (+7.2 points) for Democrat Jennifer Wexton over Rep. Barbara Comstock (R). In fact, Wexton ended up beating Comstock by 56.2%-43.8% (+12.4 points), so 538.com got the winner right but underestimated the margin by about 5.2 points.

A- Decision Desk HQ had 233 Democratic House seats as its “mean,” and currently we’re talking about around…232-234. So, they NAILED THAT! On the Senate side, their “mean prediction” was Republicans with 52 Senate seats, which is still possible, although more likely Republicans will end up with 53 Senate seats. Still, not a bad prediction at all! For individual House seats in Virginia, Decision Desk HQ had VA-10 with a 70.7% of going to Democrat Jennifer Wexton (bingo!); VA-02 with a 50.6% chance of going to Democrat Elaine Luria (yep!); VA-07 with a 52% of going to Dave Brat  (not quite!); and VA-05 with a 59% chance of going to Republican Denver Riggleman (yep!). Very good overall!

B+ The Washington Post-Schar School final polls of Virginia-10 had Wexton up 11-13 points, which was exactly right – nice job (despite Comstock absurdly calling it a fake poll)! The poll had Kaine +19 points in VA-10, which was a bit high, as Kaine ended up beating Stewart in VA-10 by about 12.4 points. By the way, the Post’s final national poll had Dems up 51%-44% among “likely voters,” and right now – with many votes still to be counted in deep-blue California, Dems are up 51.5%-46.7%. So…not a bad final poll!

B The Cook Political Report had 58 “competitive races” House seats as either “Likely Democratic” (12 seats), “Lean Democratic” (16 seats), or “Toss Up” (30 seats). That’s decent overall, although as it turned out, a lot of the Cook Report’s “Toss Up” races went Democratic. That includes here in Virginia, where Cook had VA-02 and VA-07 as “Toss Up,” with VA-05 as “Lean Republican” and VA-10 as “Lean Democratic.” As it turns out, VA-10 was a lot more than “Lean,” while VA-02 and VA-07 both went “blue.”

B Inside Elections with Nathan L. Gonzalez predicted a Democratic gain of 25-35 seats as the “most likely outcome,” with “larger gains possible. So they kind of covered themselves there, but in the end were basically correct. In Virginia, Inside Elections had VA-10 as far too timid “Tilt Democratic” (should have been “Lean” or even “Likely”), with VA-07 as “Toss-Up” (that’s fine), VA-02 as “Tilt Republican” (not quite), and VA-05 as “Likely Republican” (pretty much, yes). Not bad overall.

B-/C+ Christopher Newport University’s Wason Center final polls and ratings had Abigail Spanberger leading Dave Brat by 1 point in its “likely voter model” and by 3 points in its “most committed voters” model. That basically nailed it, with the final result (Spanberger +1.8 points) right in the middle of those two models. The Wason Center correctly predicted that VA-10 would flip blue, but it was wrong that VA-05 was a “Toss Up” and also wrong that VA-02 was “Lean R” (note: its final poll had Taylor up 7 points, 50%-43%, which was wildly off). It was also ridiculous to call VA-01 “Likely R,” when it was clearly “Solid R.”

C-/D+ The New York Times’ final polls had Rep. Scott Taylor ahead of Elaine Luria 45%-42% in VA-02 (in fact, Luria won by 2.2 points); Jennifer Wexton up 48%-41% (7 points) over Barbara Comstock (Wexton won by significantly more than that, 12.4 points); Dave Brat up 47%-43% in VA-07 (in fact, Abigail Spanberger beat him by 1.8 points); and Leslie Cockburn up 46%-45% over Denver Riggleman in VA-05 (in fact, Riggleman won by nearly 7 points). So basically, the NY Times missed almost every House race here in Virginia.

F Trumpster and right-wing radio host John Fredericks was wayyyy off, both nationally (he thought Republicans would hold the House by one seat – NOT! – and that Republicans would pick up five seats in the Senate – absolutely not!!) and here in Virginia (he incorrectly had Comstock winning, Brat winning and Taylor winning; he did manage to get the Riggleman vs. Cockburn race right). Of course, this is the same guy who completely missed who would win the 5th CD GOP nomination (he had the favorites as Sen. Bill Stanley, Martha Boneta, Cynthia Dunbar, Mike Del Rosso and…FINALLY Denver Riggleman, the guy who actually won). And this is the guy who predicted, wildly incorrectly, on November 6, 2017 that Democrats would only pick up a net of ONE SEAT (!!!) in the Virginia House of Delegates (note: we picked up 15 seats!). Oh, and Fredericks predicted Ed Gillespie beating Ralph Northam (in fact, Northam romped), Jill Vogel beating Justin Fairfax (ditto) and John Adams beating Mark Herring (ditto). And note that Fredericks also was the one who insisted, over and over and over again, including minutes before EW Jackson won the 2013 GOP Lt. Governor nomination, that he absolutely would NOT win the nomination (note: Ben Tribbett, on Fredericks’ show, kept telling Fredericks that he was wrong, but of course he didn’t listen to Ben, who is an infinitely better political analyst and forecaster than Fredericks is). In sum: #MASSIVEFAIL