Okay, Maybe Trump Will Be the Republican Nominee

About a month ago, I wrote that if Donald Trump did not expand his support beyond the 30% he was polling, he would not win the Republican nomination. Like others, I was convinced that a candidate as bombastic and undisciplined as Trump would be unable to do so. Now, I take my place among the legions who were wrong about the Donald.

In the past month, Trump has strung together a series of double-digit victories. He won by almost 20 points in New Hampshire, 10 points in South Carolina, and 22 points in Nevada. His average vote share has been about 38 percent in those contests, and he has claimed 63 percent of available delegates. His base has clearly grown, and the polls in the reflect it. Now, my statistical model predicts that Trump is the overwhelming favorite to win the Republican nomination.

Now that the Republican field has been whittled down to five candidates, I need fewer assumptions to support my model. This time, I only make three explicit assumptions (the first two of which I borrow from my previous article).

1. The 20 remaining states and territories that are often considered “winner-take-all” will each award all of their bound delegates to a single candidate. This is a flawed assumption, because most of these states actually award only a fixed number of delegates to the statewide winner, allocating the remainder to the winners of each congressional district in the state. However, this sort of assumption is needed to create the simple model I am looking for. Moreover, though a statewide winner might not win every congressional district in a single state, they might offset those losses by picking off a few congressional districts in other “winner-take-all” states. Thus, the “winner-take-all” assumption is probably more accurate than it might initially seem.
2. The 34 remaining states and territories that award their delegates proportionally will do so invariably. This is also an imperfect assumption; some of these states do not award any delegates to candidates who do not receive some minimum percentage of the statewide vote (such as 10 percent). However, for simplicity’s sake, we will assume that the candidates still in the race always meet this threshold, and are thus always allocated their proportionate share of delegates.
3. Ben Carson will end his campaign after Super Tuesday, and his relatively limited support will be approximately evenly distributed among Trump, Rubio and Cruz.

I model Trump’s vote share under what I take to be the best and worst case scenarios for Trump’s campaign. In the best case scenario, Ohio Governor John Kasich stays in the race for the duration of the primaries, taking as much as 10 percent of the total vote away from other non-Trump candidates. In the worst case scenario, Kasich drops out after Super Tuesday, with his supporters presumably drifting to the Rubio camp. I ran 10,000 simulated elections under both sets of circumstances.

For the Super Tuesday states for which polls were taken within the past two weeks, I modeled Trump’s vote share as a normal distribution centered around the state polling average. For Super Tuesday states with no recent polling and states voting later, I assumed that Trump’s vote share could be thought of as a normal distribution with a mean centered around the national polling average.

The below histogram represents the results of my first simulation. The red line represents the delegate threshold Trump has to reach to win the nomination on the first ballot at the Republican National Convention.

This first histogram shows that if John Kasich remains in the Republican field for the long run, Trump will, in all likelihood, win the nomination. Of the 10,000 simulations I ran with Kasich in the race, Trump surpassed the 1,236 delegate marker 9,517 times, 95.17 percent of all contests. The minimum number of delegates Trumps won in these simulations was 1014, making it virtually impossible for any other candidate would win the nomination outright. Thus, if Kasich stays in, anti-Trump Republicans will probably have to pin their hopes on the slim chance of a brokered convention.

If Kasich exits the race after Super Tuesday, the story is slightly different. As the below histogram shows, an early Kasich withdrawal would modestly reduce the odds of a Trump nomination.

Without Kasich splitting the vote, Trump only wins the nomination outright in 79.01 percent of simulations. This represents a significant increase in the chances that a non-Trump candidate wins the nomination through a contested convention.

This analysis has two major implications. First, it suggests that if Trump holds on to his current level of support, there is little the Republican establishment can do to deny him the nomination. This means that if Republicans have any hope of selecting a respectable nominee, they must do everything they can to shrink Trump’s base. Luckily, even a three percentage point reduction in Trump’s current average vote share would be enough to reduce Trump’s chances of winning the nomination outright to 23.92%, as pictured in the below histogram.

As a result, Republicans operatives should continue their all-out negative ad campaign if they want to minimize their exposure to Trump.

The second take-away from my analysis is that Kasich’s withdrawal would markedly reduce the chances of Trump getting the nod for the GOP. If Kasich voters eschew Trump (as one would expect), their liberation could significantly boost the prospects that a less polarizing candidate becomes the Republican nominee. Thus, in order to avoid a Trump nomination, Republican operatives should continue their campaign to shame Kasich into dropping his candidacy.