Trumped: How Likely is a Donald Trump Victory?

As the Republican and Democratic primaries are coming to a close, it is clear that Donald Trump and Hillary Clinton are going to be the nominees for their respective parties. While Clinton at the top of the Democratic ticket does not come as a surprise, the thought of Trump receiving the Republican Party’s nomination seemed impossible to most people just a few months ago. Indeed, just about everyone waspredicting that Trump would follow the flavor-of-the-month rise and fall, including Nate Silver’s estimate that Trump had a less than two percent chance of winning the nomination.

Now that Trump is moving on to the next phase of the election cycle, there has been concern among both Democrats and Republicans about what a Trump presidency would look like. Most pundits have become a little gun shy when it comes to Trump predictions, while some have even gone as far as to say that he exceeded expectations in the primaries and there is a good chance he could do it again in the general. Which got us thinking- does he actually have a shot?

As one of the few people who predicted Trump would not be a flash in the pan, we were interested to see if we could come up with a plausible path of victory for Trump this November. As we have discussed in the past, the presidential races have become very predictable and our PAAR model shows that Democrats are favored to win, no matter the nominee on either side. However, since Donald Trump is far from the “generic Republican” that our model examines, we decided to take another look at our PAAR model through the lens of a Trump candidacy in order to attempt to account for the unique elements of a Trump campaign.

The outcome of this exercise determined that there are only two plausible paths for Trump: 1) he increases support among white voters by 4 points, or 2) he increases support among white voters by 2 points and also dismantles the support Clinton has among minority voters. Unless Donald Trump is able to completely rebrand himself for the general election, we can safely assume that the second option is a non-starter; given his rhetoric during the primaries, this does not seem like a farfetched assumption. Therefore, his only real shot is to expand the white vote. This means a focus on the Rust Belt states, who tend to have a higher proportion of white voters compared to states like Florida, Virginia, and Colorado.

While a 4-point increase among white voters does not sound like much, it is a HUGE mountain to climb when you put that number into context. He would need a landslide victory among white voters to even make this election close. And when we say landslide, we mean to the levels of Ronald Reagan in 1984 where he won every state but one, and the District of Columbia.

While Trump is the presumptive nominee, and therefore technically could be the 45th President of the United States, we would put the odds of this happening as slim at best. Regardless, there will be plenty of polls over the next five months that show a close race both nationally and in key states. You can ignore them all and focus in on two questions: 1) Is Trump performing better than expected among minority voters? And also, 2) what is his lead among white women? If the answers to these questions are 1) no and 2) less than 20 points, you don’t need to worry too much about a Trump presidency and you can go about your day as usual. If the answers are 1) yes or about 2) 20 points, then there is reason to worry… or celebrate if you are hoping for a Trump win.

Trump's Needed Coalition and Map

As we all know, the most important number for every presidential candidate is 270. That is what you need to win in the Electoral College to become President of the United States. As we mentioned above, there are two technically feasible ways for Trump to reach this Electoral College goal. Those options are that he increases his share of the white vote by 4 percentage points or increases the white share of the electorate by 2 percentage points and hope Hillary Clinton loses most of the Obama coalition. Given the low odds of the second scenario playing out, this election comes down to Trump increasing his support among white voters by 4 points over Mitt Romney’s track record in 2012. If he were able to pull this off, our model predicts that Trump would win 291 electoral votes compared to Clinton’s 247.

Three years ago, when we attempted to review the Republican’s path in 2016, we settled on three different options. The one that seemed most plausible was that Republicans could focus on Colorado and two of the upper Midwest trio of Minnesota, Wisconsin, and Iowa, to break the 270 electoral vote barrier needed to win the presidency.

The theory behind a Trump win these days is similar; however, instead of Colorado, his best path is through the Rust Belt, putting WI, MI, OH, and PA into play (he would need to win all four of these to hit 270). This all sounds good in theory, but this path is fraught with a few pitfalls. Most strikingly, the share of white voters has been shrinking each election, and in states like Wisconsin, many white voters are quite liberal and likely will abhor the thought of President Trump.

Trump's Inevitable Demise

Throughout his campaign, Trump’s rhetoric has sparked an explosion within the Republican party – a party that, up until recently, relied upon strong support from white-collar white men and married white women. Yet, Trump’s rise has left many of them questioning party loyalty. We fully expect most Republican voters to “come home,” however, as Ronald Brownstein notes, Trump’s success has stemmed largely from a specific type of voter. Exit polls throughout the primaries have reflected his strength among blue-collar men who are often over the age of forty-five, making less than $50,000, and are self-identified Republicans.

And here is the crux of the challenge for Trump: how do you win with an ever shrinking portion of the electorate? As represented in figure one below, the number of white voters has sharply declined over the last 20 years. According to exit polls compiled by the Roper Center for Public Opinion Research[1], the proportion of white voters in the electorate has dropped 17 percentage points over the last ten Presidential elections. In 1976, nearly nine-in-ten American voters were white, while in 2012, white voters represented only 72% of the national electorate. Over the past ten Presidential elections (1976-2012), with the sole exception of 1992, the proportion of white voters has decreased from the previous election.

Figure 1: White Voters as a Percentage of the Electorate
(Presidential Elections, 1976-2012)

Source: Roper Center

With this pattern in place, Republican nominees need to do progressively better every election among white voters if they are unable to increase their support among minority voters. Therefore, matching Romney just won’t work and would mean a worse defeat for Trump. If the trend holds, Trump would need to surpass Romney by 4 points nationally among white voters to make the math work in his favor. However, it’s not just Romney’s level of support he would have to surpass – it is every Republican candidate going back to 1972, with the exception of Ronald Reagan in 1984 and Richard Nixon in 1972. And as a reminder, those were two of the biggest blowout elections in modern history, with both candidates winning 49 out of 50 states (DC went with the Democrat in both years).

Figure 2: GOP Levels of Support Among White Voters
(Presidential Elections, 1972-2012)

Source: Roper Center[2]New York Times[3]

Trump, Gender, and the White Vote

A 4-point rise would be tough, but is it even possible? To answer this question, we decided to look at the breakout of white men and women voters over time.

Among white men, Trump excels with blue-collar workers who tend to have little or no college education. However, his road to the nomination has left many white-collar Republican men to question their party allegiance. Indeed, Trump’s message of economic populism and appeal to the despair of the middle class has isolated many upper class conservatives, who felt more represented by the likes of Romney and John McCain. Nonetheless, let’s just say that white men who have voted in the past “come home” and Trump somehow even pulls in new white men to the voting booth. Let’s assume that he gets 67% of their vote in the general election. This number is generous and has only been reached one other time in recent history – Reagan’s 1984 reelection.

If he can pull off the 67% number among white men, what would he need to win among white women? As shown in figure 3, white women have consistently comprised a strong base for GOP winners. With the exception of the elections in 1992, 1996, and 2000, a majority of white women have voted for the Republican candidate.

Figure 3: White Voters by Gender as a Percentage of the GOP Electorate
(Presidential Elections, 1972-2012)

Source: Roper Center[4]New York Times[5]

As the chart shows, there have been only three instances where the Republican candidate has done better among white women than white men. If Trump matches the best performance ever among white men (67% in 1984), the magic number among white women is 60%. Historically it has been done, but not often (1984 and 1972).

Another way to put this number in perspective is looking at the gap Trump would need to win among women. If he needs to win 60% of women that would mean a 20-point win among this cohort. Once again, historically this has only been done twice.

In his piece referenced earlier, Brownstein highlights Democrats’ longstanding failure to attract the majority of white women voters. In fact, from 1972 to the present Bill Clinton was the only Democrat to win more white women than his Republican opponent during his re-election campaign in 1996.

Brownstein notes that we can look at women further, in terms of two groups: married/less educated and single/college educated. Among white married women or women with little or no college education, the Republican party has won by double-digits in every general election since 2004[6]. Romney even received a 20-point lead among these women in 2012. However, among college educated and single women, Republicans have fared much worse. Once again, Trump is either going to need to maximize the white non-college vote or increase his support among college educated women. This is very unlikely, considering his remarks about women and the fact that he will be running against a female opponent.

The historical numbers point to this being a tough task for Trump, but looking at the reality of this year’s election, this feat is even more difficult. Indeed, Trump is not well positioned to gain the support needed among white women.  According to a recent poll from the Washington Post, 66% of white women view Trump unfavorably, and 55% view him very unfavorably.[7] The same poll had similarly low favorability ratings among white men with 51% viewing Trump unfavorably and 32% very unfavorably, but it is hard to imagine that Trump would become the first Republican candidate to lose the white vote since 1964.

The numbers above explain why Trump is unlikely to win on a national level. But, as we said above, presidential elections come down to the states and garnering 270 electoral votes.

State By State

Current consensus states that Trump would have to continue to appeal to lower income white voters in the Rust Belt in order to win. Teasing this out a bit further, let’s say Clinton wins all the states Obama won except WI, MI, OH, and PA. This would put Trump at exactly 270 electoral votes. Assuming that Trump wins 67% of white male voters overall, a number we have generously given him in this hypothetical situation, he would need to extend his lead 17 points on average among white women in these states. We can immediately identify reasons why this would be hard to achieve. For one, these states may have high numbers of Trump’s core supporters (white, wary of foreign trade, hit hard by the recession with slow employment gains etc), but white women in two of these states have been voting for Republicans in smaller numbers over the past three elections.

Figure 4: White Women as a Percentage of the GOP Electorate
(Presidential Elections, 2004-2012)

Source: CNN Exit Polls (2004-2012)

In Pennsylvania, the GOP vote among white women has increased, while in Michigan it has remained relatively consistent from 2004 to 2012. In the other two states in which Trump would need to win, Ohio and Wisconsin, there has been a decrease in support for Republicans within this demographic cohort. Interestingly, Obama did comparatively well among white women in these states for the 2008 election.

Wisconsin has the highest percentage of white voters, but it is also likely the most liberal of the Rust Belt states. For one, it is the only one of these states we are counting as necessary for Trump where a majority of white women voted for Obama in 2012. In 2012, white women voted 53% to 46% for Obama against Romney, while white men voted 56% to 42% in favor of Romney. Among white women, Trump would need a 7-point improvement from 2012, when a majority of their votes went to Obama.

Ohio, a crucial swing state in 2012 that ultimately went for Obama, the vote among white women favored Romney but not by much. Overall, 46% of white women voted for Obama versus 53% who voted for Romney. White men in Ohio were much stronger for Romney (62%) than for Obama (36%). In this state, Trump would need a 4-point increase among white women from 2012 if he can bump his support among white men up to 67%. This equates to a 14-point gap between Trump and Clinton.

In Michigan, the state with the lowest percentage of white voters of the four, 47% of white women voted for Obama in 2012 while 41% of white men did the same. Here, Trump would need to bump up support by 10 points to have a chance of winning, or a 26-point gap.

The gap was a bit wider in Pennsylvania for the last election, with 45% of white women for Obama and 54% for Romney. For white men, only 39% voted for Obama and 60% voted for Romney. If Trump can hit the 67% mark among white men, he would need a 6-point increase or a 20-point gap.

We typically do not focus on gaps when looking at polling numbers. However, since we expect to see a higher undecided number in polls than normal this year, the gap should be the focus of dissecting polls to gauge whether Trump is close to hitting his mark or not.

Conclusion

The odds of Trump winning are slim at best. He will need to run up the score among white voters while maintaining the level of support Romney received from minority voters. Trump’s path to victory rests on winning four states that Obama won in the last two elections, three of which have not supported a Republican nominee since the 1980’s. Trump will need a historic win among white voters to make this contest even close.

In the meantime, we are likely to see national and state polls that show a closer race than many will be comfortable with. These can, and should, be ignored. The only numbers that should be paid attention to are the spread among white women in these four states and any support among minority voters that show a possibility of moving to Trump. While anything is possible, it would take an incredibly poorly run campaign by Clinton or a major event to derail the inevitable.

For a PDF version of this white paper, please click here.

[1] Roper Center for Public Opinion Research. http://ropercenter.cornell.edu/polls/us-elections/how-groups-voted/

[2] Roper Center for Public Opinion Research. http://ropercenter.cornell.edu/polls/us-elections/how-groups-voted/

[3] New York Times Election Results. http://elections.nytimes.com/2008/results/president/national-exit-polls.html

[4] Roper Center for Public Opinion Research. http://ropercenter.cornell.edu/polls/us-elections/how-groups-voted/

[5] New York Times Election Results. http://elections.nytimes.com/2008/results/president/national-exit-polls.html

[6] http://www.theatlantic.com/politics/archive/2016/05/white-women-measure-of-trumps-success/481305/

[7] https://www.washingtonpost.com/page/2010-2019/WashingtonPost/2016/04/14/National-Politics/Polling/question_16174.xml?uuid=DJ281AIwEeaLsfEkpD-E3A