Quantifying the Kasich Spoiler Effect
A small portion of my spreadsheet modeling the Illinois results.
According to exit polls, if John Kasich dropped out of the presidential race, 45% of his voters would stay home. 37% would vote for Ted Cruz. 18% would vote for Donald Trump.
These are the numbers FiveThirtyEight computed last week, which I then used after the March 8th primaries, and which were confirmed, broadly speaking, by exit polls after the March 15th primaries. Today, I am using them once again. After all, it is now mathematically impossible for Kasich to actually win the nomination during the primary season, and, thanks to Rule 40(b) of the Republican Party (which I'll be discussing a lot more in the upcoming parts of Conventional Chaos), it is almost impossible to foresee him even winning enough states to have his name entered into nomination, and even less possible to foresee the convention changing Rule 40(b) to help out Kasich, given that the convention will almost certainly be controlled by a mix of Cruz and Trump supporters, both of whom want Kasich out. By any rational measurement, Speaker Paul Ryan has a better chance at the GOP nomination than John Kasich.
So the only effect of continuing the Kasich campaign -- and Kasich surely knows this -- is to take votes away from Ted Cruz, allowing Donald Trump to win an outright majority of delegates and win the presidential nomination. (In fact, because staying in reduces the chance of a brokered convention, Kasich has a better shot at the presidency if he drops out!) One wonders: is Kasich simply so overconfident about his appeal to the Northeast that he actually believes he can win, or is he deliberately helping Donald Trump in the hopes of winning the vice-presidential slot on a Trump ticket?
For now, that's not a question we can answer. Here's one we can, though: exactly how much Kasich is helping Trump? We'll go through the March 15th states one by one and see how many delegates were lost to Trump because of Gov. Kasich. (We'll do the same for Marco Rubio, simply for transparency, though Rubio has since done the honorable thing and dropped out.) MISSOURI Actual Result w/o Rubio Cruz Alone Trump 37 20 0 Cruz 15 32 52 Kasich 0 0 - Rubio 0 - - Trump Margin +22 -12 -52 In Missouri, Kasich blocked Cruz not just from winning, but from climbing over the 50% threshold, which would have activated a winner-take-all clause that would have handed Cruz all 52 of Missouri's delegates. Instead, Trump walked home with the majority of those delegates... despite winning by only 0.2%.
You see, in the first half of the GOP primary calendar, all states were roughly proportional, so every candidate would win at least some delegates as long as they won a reasonable number of votes. In a winner-take-most contest, however, the rules make it so that a winner gets most of the delegates if he wins by 0-2%, and all (or nearly all) of the delegates if he wins by much more than that. So Cruz and Trump can finish in a dead heat in Missouri (40.5% to 40.7%), but Trump is awarded 71% of the delegates! If Cruz had finished two or three percentage points worse, Trump could have won everything in Missouri -- even without winning a majority.
Here in the second half of the GOP primary calendar, there are only 3 proportional states left. 10 are pure winner-take-all, 3 elect unbound delegates, and the rest (11) are some version of winner-take most. That's why a spoiler like Kasich is much more dangerous now than he was even a week ago: in a winner-take-most world, Kasich may only peel off 1 or 2% of Cruz's vote relative to Trump, but that's enough to swing many of the upcoming races from a Cruz triumph where he takes home two-thirds (or more) of the delegates to a Trump triumph where he takes home two-thirds (or more) of the delegates. What's that give us? President Trump.
If Kasich stays in the race until March 22nd and beyond, this is a scenario we're going to see over and over again. For now, on to Ohio. OHIO Actual Result w/o Rubio Cruz Alone Trump 0 0 66 Cruz 0 0 0 Kasich 66 66 - Rubio 0 - - Trump Margin -66 -66 +66 In Ohio, Kasich won his first (and likely only) victory of the 2016 primary, in the state where he currently serves as a popular multi-term governor. It was a big victory, too, because Ohio's delegates are winner-take-all. Cruz could not have caught Trump in Ohio on his own, according to exit polls, so Kasich's single win took 66 delegates away from Trump, which Trump otherwise would have won. That's a significant win that does, actually, offset the (significant) damage Kasich did elsewhere.
At last report, Kasich is currently searching for other states where he is a beloved multi-term governor. "I’ve proved that I can win big in areas where I’m the governor, and now it’s time to take that approach nationwide," Kasich said. If he fails, then we will watch him do lots of damage for the rest of the primary... with no offsets like Ohio tonight. ILLINOIS Actual Result w/o Rubio Cruz Alone Trump 54 48 43 Cruz 9 12 26 Kasich 6 9 - Rubio 0 - - Trump Margin +45 +36 +17 In Illinois, the Kasich Effect is clear: the statewide vote was a decisive Trump win, with Trump taking 39% of the vote, Cruz taking just 30%, Kasich 20%, and Rubio down at 9%. But eliminate Kasich (and Rubio), and suddenly it's a squeaker: 49% Trump to 48% Cruz. Because Illinois is a winner-take-most state, that's a big deal.
Illinois uses a peculiar "loophole primary" system. Don't ask; it's complicated. It took me well over an hour to remodel the results without Rubio and Kasich in all 18 districts. Suffice to say that, at the end of the day, Cruz's much closer finish would have allowed him to take away 11 of Trump's delegates... but Kasich's presence in the race denied him that.
So, let's recap: Kasich's presence in Ohio took away 66 of Trump's delegates. But Kasich's presence in Illinois and Missouri gave a total of 48 delegates back to Trump. Kasich's "big win" was truly a Pyrrhic victory: a few more "wins" like that and we'll be toast! NORTH CAROLINA Actual Result w/o Rubio Cruz Alone Trump 30 31 35 Cruz 27 30 37 Kasich 9 11 - Rubio 6 - - Trump Margin +3 +1 -2 North Carolina is one of the last proportional states on the calendar, so Kasich and Rubio dropping out has less of an effect there. Still: it would have changed Cruz's narrow loss there into a narrow win. Not bad. FLORIDA Actual Result w/o Rubio Cruz Alone Trump 99 99 99 Cruz 0 0 0 Kasich 0 0 - Rubio 0 - - Trump Margin +99 +99 +99 Nobody could stop Trump from winning winner-take-all Florida. CNMI Actual Result w/o Rubio Cruz Alone Trump 9 9 9 Cruz 0 0 0 Kasich 0 0 - Rubio 0 - - Trump Margin +9 +9 +9 Ditto the winner-take-all Northern Marianas Islands, where Trump won his first outright majority win (he got 76% of the vote in the tiny island territory). TOTALS Actual Result w/o Rubio Cruz Alone Trump Margin of Victory +178 +133 +137 Trump Delegates Won on 3/15 229 207 252 Rubio did nothing but damage Tuesday night (as predicted), handing Trump 22 free delegates just by spoiling. Kasich's presence did ultimately benefit the #StopTrump campaign by winning Ohio, but his 66 delegate win there was offset not just by Rubio's defeat, but by 21 delegates Kasich himself gave to Trump elsewhere. Kasich's net benefit on Tuesday was therefore 45 delegates denied to Trump. If he'd done this, then immediately dropped out, I'd give him a big round of applause for his help in the anti-Trump effort.
But he didn't.
With Arizona coming up next Tuesday night, with its 58 winner-take-all delegates, Kasich -- who notably is not the popular incumbent governor of Arizona -- may well be able to wipe out all the gains he won this week... and then some!
In fact, let's do a quick model of Arizona, based on the most recent polling. That poll gives Trump 31%, Cruz 19%, and Kasich and Rubio 10% each, with 30% undecided.
We will follow our assumptions above. We will further assume, based again on FiveThirtyEight's analysis of the March 15th exit polls (excluding the favorite-son states of OH and FL, for obvious reasons), that Ted Cruz will get 43% of late-deciding votes, Trump 31%, Kasich 16%, and Rubio 8%. What that gets us: ARIZONA - Projected Actual Polling w/o Rubio Cruz Alone Trump 40% 42% 44.8% Cruz 32% 36% 44.7% Kasich 15% 19% - Rubio 12% - - Trump Margin +8% +8% +0.1% Kasich's presence turns the Arizona race from a Trump/Cruz dead heat into an easy Trump win. 58 delegates are on the line, and Kasich hands every single one of them to Trump, gift-wrapped with a ribbon on top.
Meanwhile, up in Utah, which votes the same day, we face another Kasich-induced quandary. The state is proportional, with 40 delegates... unless a single candidate wins 50% of the vote, in which case Utah becomes winner-take all.
We don't have any good recent polling of Utah, but we can still make a rough model. We'll assign undecideds and exclude dropped-out candidates not named Rubio (because we simply don't know how Carson and Bush voters reassigned themselves). We grab the last poll, do some math, and boom: at the end of February, Utah looked something like this: 34% Cruz, 31% Rubio, 27% Trump, 7% Kasich. Let's feed those numbers into our model... UTAH - Projected Actual Polling* w/o Rubio* Cruz Alone* Trump 27% (12) 35% (14) 37% (0) Cruz 34% (15) 47% (19) 63% (40) Kasich 7% (0) 18 (7) - Rubio 32% (14) - - Trump Margin -7% (-3) -12% (-5) -24% (-40) *Projected delegate counts are in parentheses.
Kasich's presence in Utah spoils Cruz's shot at breaking 50% and taking home all 40 delegates. This gives Trump 14 free delegates (while scoring only 7 for Kasich, incidentally).
So, based on current projections, Kasich took 45 total delegates away from Trump on March 15th... and, on March 22nd, it's a coin flip Kasich will give at least 14 and as many as 72 delegates right back to Trump. Average damage: 46 delegates. Just enough to finish cancelling out all the good Kasich did on March 15th in Ohio.
And that's just two states. There are 17 states with binding primaries voting after March 22nd, most of them either winner-take-all or winner-take-most.
Can you begin to see why Kasich's presence in the race is so dangerous?
Please, someone phone the GOP Establishment in their secret headquarters deep inside the Bloomberg corporate offices and tell them that, unless they really want Donald Trump to be the GOP nominee -- unless they want to completely shatter their own party -- they need to hustle Kasich out of the race. They have about four days left before the damage Kasich deals becomes severe. If that happens, our hopes of denying Trump hinge on a series of iffy propositions like denying Trump wins New York, California, and Indiana. We might pull it out anyway, if Kasich gets out by May 3rd, but we really, really don't want to go there. Surely the Establishment will listen to reason, pull some of the strings they have so often pulled to hurt conservatives, and get Kasich gone.
...right?
UPDATE: Go read Sean Trende, whose column today says the same thing differently.