Thursday, November 6, 2014

Election Post-Mortem and looking back at predictions

Well, here's my election post-mortem.  First, I summarize the results.  Then I check the accuracy of my Election Day and New Year's predictions, with some election takeaways at the bottom.

First, the results:

  • The GOP has gained at least 7 seats in the Senate, including WV, MT, SD, AR, IA, CO, and NC.  As of this writing, Alaska hasn't been called yet but Republicans are ahead there as well.  Louisiana will go to a runoff, which Republicans should be pretty favored to win.  Virginia also has not yet been called (which is impressive all by itself) but Mark Warner is ahead there and will probably remain so.  Assigning AK and LA to the Republicans and VA to the Democrats gives the Republicans a 9-seat gain, and a 54-46 advantage in the next Senate.
  • Perhaps more impressively, they've done better than expected in the governor's races.  As of this writing, they've lost PA and taken MA, Maryland, AR, and IL.  AK, CO, and CN have not yet been called, but the Republican trails all three.  Depending on how those turn out, we could see a net Republican gain of 2 (if they lose all three) to 5 (if they win all three) governorships, which would leave them somewhere around a 31-18-1 advantage (worst case) and 34-16 advantage (best case).
  • Even with a dozen+ House races as yet uncalled, the Republicans have gained their biggest House majority since the late 20's with, as of this writing, a 13-seat gain with the prospects for several more from the close races not yet called.


Now, the predictions:


The only race I got wrong was North Carolina, which I predicted Hagan would win without a majority.  Instead, Tillis won without a majority.  I did raise that as a real possibility, but I didn't predict that, and got it wrong.  Other notable wrong calls: Perdue did win Georgia without a runoff, not with a runoff as I (reluctantly) predicted, while my calls in Virginia and Kansas, while I got the winner right, were off enough to be worth mentioning - I didn't expect Gillespie to seriously threaten Warner, and I expected Orman to come close to Roberts.


I didn't do as well in the gubernatorial races (though, given the results, I'm not complaining).  I ended calling most of the close races wrong.  The Democratic sweep of FL, ME, and KS turned into a Republican sweep (which I also mentioned as a possibility, and really should have gone with).  My call that Quinn would win reelection in IL turned out to be wrong, and it looks like my call for Parnell in AK is wrong as well.  It looks like I was right about CO and CN, but I still hold out hope I was wrong there as well.  Perhaps people should start treating my calls of close races like this.  Then there's MA.  I said I'd consider it an upset if Brown lost, and I do.  But yeah, that one was spectacularly wrong.  And one final one: in my former home state of VT, I predicted an easy hold for Shumlin.  His Republican opponent, Scott Milne, led most of the night, ended up losing by less than 2 points, and ended up forcing Shumlin to be selected as Governor by the state legislature.  I will say again, however, if this is what my being wrong results in, I'm happy to be wrong.


Same thing as the gubernatorial races.  My guess underestimated Republican gains, and I'm perfectly OK with that.

Digging up my New Year's predictions:

Yes I'm going to look back at them as well.  You can find them here.

  • Democrats will neither retake the House nor significantly dent the Republican majority.  The results will be such that under normal presidential year conditions, no one will expect Democrats to retake the House in 2016 either.
Yep.  Republicans gained seats.  It'll take a little while to see whether anyone seriously argues Democrats can take back the House in 2016, but I will say that would require taking 30+ seats, which doesn't seem to me a standard presidential year gain in seats for the incumbent party.  I'll go ahead and call this accurate.
  • One, and only one, Republican Senator facing a Tea Party challenge will lose
Wrong, though with Roberts, Alexander, and especially Cochran, is was a close-run thing.
  • Primary choices will cost Republicans one of these three races: Alaska, North Carolina, Georgia
Wrong, fortunately.
  • Democrats will make a small net gain of 1-2 governorships, but Republicans will retain the majority. Their next realistic chance to take the majority won't come until 2018
Partly wrong.  Republicans gained governorships, but they kept the majority (odd turn of phrase there), and there doesn't seem to be any realistic way for them to lose it until 2018 at the earliest.
  • Tom Corbett will lose reelection
Yep.  Bit of a gimme.
  • One Republican who has not yet announced a run will enter one of the non-competitive Senate races and make it competitive
That would be (Sen.-elect) Cory Gardner in Colorado.
  • Related to previous: the Hawaii Senate race will tantalize Republicans by appearing competitive, but this will again be a false hope for them
Wrong.  Since Djou didn't run, there was no hope there to be false.  Though you could say there was false hope in the governor and 1st District races here.
  • Susan Collins will not retire
She didn't.
  • The Louisiana Senate race will go to a runoff
  • That runoff will not determine control of the Senate
GOP majority already assured.
  • A net gain of three Senate seats will be the Republican floor
Bit hard to say for sure what the theoretical minimum gain in the Senate was for Republicans, but I'd love to see someone argue it was less than three.
  • Tom Cotton will become the youngest US Senator
This was a roundabout way of saying Cotton would win, but the second part did end up being true as well despite some early threats from T W Shannon to make it false.
  • Rick Scott will again be elected Governor of Florida with more Floridians disapproving of him than approving
Should have stuck with this in my Election Day predictions.  Scott did win and, as far as I can tell, his net approval is still negative.
  • Jason Carter and Terri Lynn Land will both lose, but will come closer to winning than any candidate of their respective parties in ten years
This one is accurate, though it says more about the performance of Republicans in Michigan Senate races and Democrats in Georgia gubernatorial races than it does about Land's and Carter's particular strengths as a candidate.
  • If the Republican nominee doesn't hand her the race, Michelle Nunn will do the same
Both options were wrong.  Nunn neither won, nor did the best of any Senate candidate in Georgia the Democrats have put up in the last ten years - that honor goes to their previous nominee for this seat: Jim Martin.  I would say that the environment had a ton to do with it, but hey, rules are rules.


And, finally, the takeaways:

  • Like I said in evaluating my New Year's predictions, it looks like the Republican House is safe for awhile absent yet another goddamn wave (and I do think this was a wave, more on that below).  Their majority is big enough that, while it'll probably be reduced in 2016, it'll still hold.
  • Even if there is a Democratic wave in the next four years, the Republican majority in the governorships should be secure.  Even in the worst case scenario, the Democrats would have to gain 7 governorships in the next 4 years to retake that majority before 2018, which would entail basically sweeping the governor's races in 2015, 2016, and 2017.  That would mean, besides defending plenty of their own vulnerable seats, they'd have to take places like Mississippi, Louisiana, North Dakota, and Utah from Republicans.
  • On the other hand, once 2018 hits, Katy bar the door.  There'll probably be Republican open seats in states like Ohio, Iowa, Nevada, New Mexico, Maine, and Florida, plus Democratic open seats in California and New York.  2018 should feature plenty of open and competitive gubernatorial races.
  • The newfound Republican Senate majority, however, looks much less secure.  A 4-seat majority, while better than expected, isn't all that sturdy, particularly considering the tough map Republicans are facing in 2016.  It looks like there'll be yet another big tossup battle for the Senate and, if the Republicans do lose the majority, a further one in 2018.
  • As Sean Trende has noted, the Senate is a bit of a natural Republican gerrymander thanks to the greater number of small, heavily Republican states, but the Democrats have kept it even by overperforming in many of those.  One of the most important things out of this election is that the Republicans have put a big dent in that advantage.  By winning in West Virginia, South Dakota, Montana, Arkansas, and (presumably) Alaska and Louisiana, they haven't just taken seats, they've moved the baseline 6 points more Republican.  Once 2020 finally rolls around, we probably won't be refighting a lot of the races from this cycle.  Colorado, Iowa, and North Carolina will still be competitive, Georgia will probably be as well, and maybe Susan Collins will be ready to retire in Maine.  But most of their gains this year will probably be off the table, in a way that many of those in 2010 weren't.
  • Since Florida is my new quasi-home state, I'd like to mention a bit about it.  Republicans had a generally pretty good night, but there were some notable disappointments: Gwen Graham's narrow victory over Steve Southerland, and the disastrous attempt to dislodge Patrick Murphy. Just going by the partisanship of their districts, neither should have won, but that they did will probably come back to bite Republicans in an outsized manner.  The best time to stop a rising star is before they get too high, but since this didn't happen, expect these two to be recruited for important statewide runs in the future (particularly given the Democrats' otherwise rather thin bench).
  • That segues nicely into another important background story about both this election and the last one, which is the thinning of the Democratic farm team as Republicans have been taking over many of the traditional paths of advancement to higher office in many swing states.  The best example of this is in Ohio, where Republicans now control the state legislature, every statewide elected office, and 3/4 ths of the congressional districts.  It's a fairly similar situation in Michigan, which is partly why, while the Democrats got Gary Peters for Senate, they had to settle for a one-term former Congressman who lost his seat in 2010 as their nominee for Governor.  Nationally, Republicans now control at least 2/3 rds of state legislative chambers.
  • Back in my other new quasi-home state, Texas, a lot of people are dancing on the graves of the Battleground Texas guys after Greg Abbott beat Wendy Davis by 20 points in the Governor's race.  That's hardly an impressive result for Democrats, but it was replicated up and down the ballot.  According to the returns on the Secretary of State's website, there wasn't a single Democratic statewide candidate who reached 40% of the vote.  At the top of the ballot, John Cornyn beat David Alameel for Senator by 27 points.
  • One noteworthy thing from the Senate results: most of the Democrats' 'red state counterattacks' didn't turn out that well.  Alison Lundergan Grimes lost in Kentucky by 15 points - the second worst defeat of any of McConnell's challengers over the years.  Most of that's probably not her fault, but it does illustrate the difficulty of unseating a red state Republican incumbent in this environment.  Same thing in the rest of them.  Perdue ended up winning by 8 in Georgia, Roberts by 11 in Kansas, and Rounds in South Dakota by 21.
  • And, finally, was it a wave?  I tend to think so.  Republicans lost just one competitive Senate race (or two if you count Virginia, which was never supposed to be competitive), out of what appeared to be a good ten or so right before Election Day.  Sean Trende found a 'break point' in the 2010 Senate results - any state with PVI D+2 or less, Democrats tended to loss, while they won almost all D+3 and greater states.  This year, that point was at D+1 and, if anything, a little more solid.  In 2010, Democrats won three seats past the break point: Nevada (D+1), Colorado (TIE), and West Virginia (R+8).  This year, they won one at the break point (New Hampshire (D+1)), and one past it (Virginia (TIE)).  I think the House results also show evidence of a wave.  Let's say Republicans add another three seats from the uncalled races (a very reasonable proposition).  That would give them a net gain of 16 seats.  That's 1/4 th of the gains from 2010, but considering it came on top of many of the 2010 wins.  Republican performance in the governor's races and downballot state races also suggests a wave.
Sorry for being a little slow to get this out.  It was delayed by Internet troubles (which I just thank God waited to show until after Election night) and the sheer length of the thing.

Tuesday, November 4, 2014

2014 Midterms: Official Predictions

Senate: 53 R - 47 D

R Pickups: AK, AR, CO, IA, LA, MT, SD, WV   D Pickups: None

Governors: 27 R - 23 D

R Pickups: AR, MA   D Pickups: FL, KS, ME, PA

Here goes:


Alabama - Jeff Sessions:
  Has to occupy the top spot by virtue of not having an actual opponent.

Delaware - Chris Coons,
Hawaii - Brian Schatz,
Idaho - Jim Risch,
Illinois - Dick Durbin,
Maine - Susan Collins,
Massachusetts - Ed Markey,
Mississippi - Thad Cochran,
Nebraska - Ben Sasse,
New Jersey - Cory Booker,
Oklahoma - Jim InhofeJames Lankford,
Rhode Island - Jack Reed,
South Carolina - Lindsey GrahamTim Scott,
Tennessee -  Lamar Alexander,
Texas - John Cornyn,
Wyoming - Mike Enzi:
  The forgotten races of 2014.  I'm sure their respective states will be so sad.  All these fall just below Alabama because they are actually technically contested races, even if it would take an Act of God for the underdog candidate to actually win.

Montana - Steve Daines,
West Virginia - Shelley Moore Capito:
  The Republicans' easiest takeovers.  I have these in the third tier solely because I dislike lumping pickup opportunities together with safe holds (this is also why I've never moved them off Likely R in the ratings).  No, I can't rationally explain why.

Arkansas - Tom Cotton,
South Dakota - Mike Rounds:
  Not as sure as Montana and West Virginia, but a Democratic win in either would be a huge shock.

Kentucky - Mitch McConnell,
Michigan - Gary Peters,
Minnesota - Al Franken,
New Mexico - Tom Udall,
Oregon - Jeff Merkley,
Virginia - Mark Warner:
  It would be very surprising if the incumbent party lost any of these.  I'm about as confident in these as I am in the South Dakota and Arkansas predictions but, for the reasons listed two entries up, they're separate.

Georgia - David Perdue (after runoff):
  I'd be surprised if I was wrong about this, but it's not out of the question.  I should also say that I'm much less confident that there will be a runoff than I am that Perdue will win.  The rule I follow is that if and only if (vote share of candidate with most votes) - (vote share of candidate with second most votes) < (vote share of all combined other candidates) will no candidate have a majority.  As of tonight, Perdue's margin of victory is .03 points less than the Libertarians share in the RCP average.  That's close enough that, enough though I am officially predicting a runoff, I would be utterly unsurprised if Perdue won outright.

Alaska - Dan Sullivan,
Colorado - Cory Gardner,
Iowa - Joni Ernst,
Louisiana - Bill Cassidy (after runoff, but Mary Landrieu wins the first round):
  There's a decent chance at least one of these wrong, but I'd still be surprised if one were.  I am, however, supremely confident there will be a runoff in Louisiana.

New Hampshire - Jeanne Shaheen,
North Carolina - Kay Hagan:
  Thom Tillis and Scott Brown have closed admirably and, as of Election morning, are both less than a point behind in the RCP average. I'd say there's an even chance at least one of Brown or Tillis ultimately wins, but the likeliest individual outcome is still both (narrowly) Shaheen and (with a plurality) Hagan.  Republicans should lose some competitive races, but we're really getting into uncertain territory now.

and. finally,

Kansas - Pat Roberts:
  Jerry Moran's long nightmare may finally be over.  I'm going against the poll average here, but I think the undecideds will ultimately break enough for Roberts to very narrowly put him over the top.  He should still win some sort of 'Worst Incumbent Candidate of the Year', but I think he wins the seat as well.  This is my lowest confidence choice, along the lines of 'Orman, Roberts, pick one, any one', and an Orman win is nearly as plausible as a Roberts win.


Alabama - Robert Bentley
California - Jerry Brown
Idaho - Butch Otter
Iowa - Terry Branstad
Nebraska - Pete Ricketts
Nevada - Brian Sandoval
New Mexico - Susana Martinez
New York - Andrew Cuomo
Ohio - John Kasich
Oklahoma - Mary Fallin
Oregon - John Kitzhaber
South Carolina - Nikki Haley
South Dakota - Dennis Daugaard
Tennessee - Bill Haslam
Texas - Greg Abbott
Vermont - Peter Shumlin
Wyoming - Matt Mead:
  The easy holds out of the way.

Arkansas - Asa Hutchinson
Pennsylvania - Tom Wolf:
  And the easy pickups.

Arizona - Doug Ducey
Georgia - Nathan Deal (no runoff)
Hawaii - David Ige
Maryland - Anthony Brown
Michigan - Rick Snyder
Minnesota - Mark Dayton
New Hampshire - Maggie Hassan
Rhode Island - Gina Raimondo
Wisconsin - Scott Walker:
  Closer, but I'd consider it an upset if any of these lost.

Massachusetts - Charlie Baker:
  At this point, I'd consider it an upset if Coakley won.

Alaska - Sean Parnell
Colorado - John Hickenlooper
Connecticut - Dan Malloy
Illinois - Pat Quinn:
  Two near-2010 losers survive another close election, but I wouldn't be surprised to see either go down.  I feel the same about Hickenlooper and Parnell.  These should all be close.

Florida - Charlie Crist
Kansas - Paul Davis
Maine - Mike Michaud:
  The three most confounding races of the cycle, but I'm going for a full Democratic sweep.  Still, I wouldn't be surprised if Republicans end up winning all three, and they all should be very close.


Plus a House guess of a Republican pickup of 10 seats.  I follow the House much less closely than I do the Senate and governors, so this isn't constructed out of individual race calls - it's just a result that seems reasonable from everything I've read.