Economics and politics - comment and analysis
20. May 2016 I Will Denayer I Countries and Regions, General, General Politics

Can Donald Trump win the presidency? Forecasts, predictions and prudence

After reading a lot of material, my conclusion is that Trump can win the presidency, but that his path to the White House is very narrow and that it is very likely that he will lose and probably by a generous margin. In this, I agree with almost everybody. The only problem is that almost everybody has been wrong before and never more so than this year.

The US presidential election is not direct. The electorate votes per state. The winner of each state wins a certain number of ‘electoral votes.’ The candidate who accumulates 270 of these votes wins. The political geography is such that some states consistently go to either the Democrats or to the Republicans (consistently means that this has been the case in at least the last six presidential elections). The great majority of these states won’t change. Other states are considered to be ‘toss-up’ states – they can go either way. Most of the ‘toss-up’ states are ‘purple,’ meaning that the urban electorate votes predominantly Democrat and the rural parts vote predominantly Republican. As you will see, there is no agreement about which states are toss-ups. It is fair to say that the most important ones are Ohio, Pennsylvania and Florida.

So how do things stand in these states? As Cillizza wrote in the Washington Times some days ago, if Clinton wins Florida (this was a near certainty until a couple of weeks ago, but has become a neck and neck race in the meantime) and she carries the 19 states (and D.C.) that have voted for the Democratic presidential nominee in each of the last six elections, Clinton wins the election. These states give her 242 electoral votes, Florida adds another 29 and it is all over (see here). For Trump, the road to the White House is much narrower. Only 13 states have gone for the GOP in each of the last six elections. They only total 102 electorate votes. That means that Trump has to find 168 more electoral votes. Cillizza proceeds to produce five scenarios that deal with toss-ups (see here and here). If any of these scenarios become reality, it becomes very difficult, almost impossible or completely impossible for Trump to win. Let me give you the gist of Cillizza’s analysis.

Cillizza’s scenarios are based on demographic changes that are taking place at state level: the growing percentage of Hispanics and the percentage of black people within the state. He also looks at income. He makes a lot of these demographic changes, although they look rather minimal to me. Even so, it is unproven that more Hispanic voters mean more votes for the Democratic party – there is no general pattern here. I do not understand why Cillizza did not look at population growth and the percentage of young voters. If Trump does abysmally bad in one group of voters, it is certainly the new generation of first time voters. Aside from this, I am skeptical about his measure of consistency (states that went to either Democrats or Republicans in the last six presidential elections). I will explain this below.

In Scenario 1, Trump loses North Carolina. Cillizza considers this ‘likely,’ because the state has a large black population (22 percent) and a growing Hispanic population (9 percent). Furthermore, a decent chunk of North Carolina’s white population is affluent and highly educated. Cillizza assumes that this factor plays to Clinton’s advantage – I return to this below. In Scenario 2, Trump loses North Carolina and Arizona. Arizona went to the GOP in 2008 when McCain was running (Arizona is his home state). In 2012, Romney won Arizona, presumably because of the large Mormon community. On the other hand, 30% of residents in Arizona are now Hispanic, a factor that should favour Clinton. In Scenario 3, Trump loses North Carolina, Arizona and Georgia. Today, Georgia is the 10th most Hispanic state in the country by population and it ranked 10th in Hispanic growth between 2000 and 2011. It was the eighth closest state in 2008 and the 12th closest in 2012. Scenario 4 deals with the Midwest. Obama won Indiana in 2008 and that same year Missouri was the closest state by margin of victory in the country. Finally, in Scenario 5, Trump loses Utah because he did very poorly in the state during the primary as he does not seem to have the support of the Mormons (see here and here).

Cizilla also points out that out of 25.2 million GOP votes cast in primaries so far, Trump got less than 10 million: 60.5 percent of the GOP base did not vote for Trump. Trump did worst in the caucuses that are open to independent voters: he lost 7 out of 10 (see here).

Let me mention a couple more forecasts before I deal with Cillizza’s analysis.

Sabato from the University of Virginia forecast last year that the Democrats will get 247 votes, the Republicans will accumulate 206 electoral votes and six states, with a total of 85 electoral votes, are toss-ups. Today, Sabato considers no states to be toss-ups. He predicts that Clinton will get 347 electoral votes and Trump 191 (see here).

The Cook Political Report predicts 304 electoral votes for Clinton, 190 for Trump and 44 will be up for grabs in Iowa, New Hampshire, Ohio and North Carolina (here and here). According to the Rothenberg & Gonzales Political Report (see here) 263 electoral votes go to the Democrats, 206 go to the Republicans and the remaining toss-up states are Colorado, Florida, Ohio and Virginia. As you can see, there is not much agreement among the forecasters.

Finally, there is the latest prediction from Quinnipiac University (see here). It puts Clinton and Trump neck and neck in Florida and Pennsylvania and Trump now leads in Ohio. Critics of the surveys asserted that the samples understate the size of the non-white vote (see here). This seems to be the only realistic path to the presidency for Trump: he has to win all of the states that Romney won in 2012 as well as Florida, Ohio and Pennsylvania.

The following two maps show versions of the political divide.

Figure 1


Figure 1: The electoral divide according to the Los Angeles Times. Source: Times reporting.

Figure 2


Figure 2: A projection of the electoral vote by Freedomslighthouse. Source:

Cizilla’s most likely case is North Carolina. Obama won North Carolina in 2008 with a 0.32% margin, but Romney won in 2012 with a 2.14% margin (and after the Democrats held their National Convention in Charlotte). Will Clinton win North Carolina? There is a lot of cherry picking here. Missouri was the closest state by margin of victory in the country in 2008 (McCain won by 0.1%), but it was the 12th in 2012 (when Romney won by 9.38%). I do not see how the point can be made that Missouri is evolving in a Democratic direction. Arizona was the 10th closest state in 2008 (McCain won with almost 8.5%) and the 19th closest in 2012 (Romney won by almost 9%). It is true that these results were better for the Democrats than was generally expected (with the exception of Bill Clinton in 1996, the Republicans have won Arizona every time since Truman in 1952), but where are the signs that the state can go Democratic in the election?

Trump did indeed especially poor in the Utah primary, but this is because Cruz has a lot of support among the evangelicals. Trump and Cruz have been insulting one another for months. But this is no reason to expect that Cruz will not throw his support behind Trump once the latter becomes the official nominee. Cruz will pick up the pieces, make no head waves and work towards 2020. I do not believe that Utah will go to Clinton (Romney won Utah in 2012 with 72.6% of the votes).

Cillizza says little about the independents, the group which may very well be the elephant in the room. Trump has shown that he can bring in voters who do not identify as Republican (see here). Some of the Republican open caucuses have been the sites of the greatest voter turnouts thus far and in some states more have turned out to vote for Trump than for Clinton. Yes, Trump did especially poorly in the open caucuses. So did Clinton.

Many commentators assume that Trump’s ‘populism’ is being carried by low income earners with modest levels of educational attainment. But this is not so. Nate Silver shows in a recent article that the majority of Trump’s voters are, at the least, solid upper middle class (see here). According to data that Silver derived from exit polls and Census Bureau statistics, the median household income of a Trump voter so far in the primaries is approximately $72,000. That is well above the national median household income of approximately $56,000. It is higher than the median income of Clinton and Sanders supporters. Silver estimates that 27 percent of American households had incomes under $30,000 last year. By comparison, 20 percent of Clinton voters did, as did 18 percent of Sanders supporters. Only 12 percent of Trump voters have incomes below $30,000 (see here). Silver’s analysis clearly shows that Trump’s campaign does not thrive on the archetypal white, middle-aged industrial worker or the middle aged unemployed. Nor are Trump’s voters particularly lowly educated. About 44 percent of Trump supporters have college degrees, which is higher than the 29 percent of American adults who have at least a bachelor’s degree (see here). It is therefore a mistake to assume that more affluent areas will automatically vote against Trump.

Third, as every model-maker and forecaster knows, the use of historical data for predictions assumes the existence of steady-state conditions (called ‘constants’) as well as the absence of new variables. Once these conditions change, ‘history’ becomes suspect or altogether useless in predicting the future. It is simply not realistic to assume that this is not the case now. Over the last few years, and certainly since the start of the primaries, the American political landscape radically changed: popular movements, both on the left and on the right, engage in a rebellion against the establishment, the Democratic party is more divided than ever and the American economy has performed well for some but left many more behind: the middle class is shrinking, inequality and poverty increase and millions of Americans are still unemployed. In 2012, no one spoke about NAFTA. Now free trade, bank bail outs and political corruption are dealt with on a daily basis by both Sanders and Trump. This context is new and hence predictability is for the most part gone.

The northern Appalachia is an example of how things have changed. This was once the land of the Democrats – both Clinton and Obama won several states there by a landslide. But what will happen in November? Of the 420 counties in the Appalachia, Trump won all but 16, including a sweep of western Pennsylvania, eastern Ohio and the western uplands of Virginia (see here).

What about Pennsylvania? It may turn out to be crucial. As said, a new Quinnipaic survey suggests that three of the purple states – Ohio, Pennsylvania and Florida – may now be too close to call. The poll puts Trump four points ahead in Ohio and only one point behind in Pennsylvania and Florida. If Trump chooses Kasich as his running mate, the two of them will take Ohio (Kasich’s home state). It is naïve to think that this will have no influence on nearby Pennsylvania. An article on FiveThirtyEight deals with the popularity of the 2016 presidential candidates when it comes to the number of likes on Facebook in Pennsylvania (see here and here). Granted, this is not a solid parameter, but it also seems imprudent to just ignore it.

Figure 3


Figure 3:  The ‘Facebook primary’ in Pennsylvania. Counties in Pennsylvania were Donald Trump has the most likes on Facebook of all candidates during the primaries are in orange (the purple spot in the middle is Harrisburg, Pennsylvania’s capital).  Source: Politics PA (

Bob Casey, a Democratic senator from Pennsylvania, recently characterised his state as Philadelphia in the east, Pittsburgh in the west and Alabama in the middle (see here). Many of the rural ‘Reagan Democrats’ seem to be turning to Trump. Will Clinton win Pennsylvania? Many forecasters do not consider Pennsylvania a toss-up state. I hope that they are right.

The Appalachia, once the centre of American steel and coal runs from New York state up north to Pennsylvania, Ohio, Kentucky, West Virginia, Virginia and from Tennessee, North Carolina down south to Georgia, Alabama and Mississippi. Why are some of these regions shifting in the direction of the Republicans, or, more accurately, Trump? The standard explanation deals with the collapse of local industries, outsourcing, the loss of manufacturing jobs, economically deprived communities, a shrinking middle class. The feeling exists that something has gone terribly wrong, that Obama did not fix it and that the Democrats may well have created a lot of the trouble (see here). Bill Clinton’s NAFTA and his financial deregulation are cases in point. As Howard Fineman writes in the Huffington Post, large numbers of cities and towns have never recovered from economic stagnation and the off-shoring of industrial jobs. Resentment of Washington is sky high in these communities (see here). The problem with all of this is that Hillary Clinton casts herself a defender of the Obama economy, but for many this is more of the same that they do not want. The margin of error is not high here either. For example, Obama won Virginia in 2012 with a 3.8% margin. Will Clinton win Virginia?

Finally, an unnamed writer wrote a piece for Zero Anthropology with the provocative title ‘Why Donald J. Trump will be the next president of the United States’ (see here). It is not necessary to agree with the overall assessment of this writer to see that she or he makes several valid points.

First, the democrats will go to the election as a divided party. No one knows exactly which role this will play, but it is certain that some of Sanders’ supporters will stay home on election day and some even indicate a preference for Trump (one poll shows 20% of Democrats defecting to Trump) (see here and here). Of course, at one point Sanders will throw his support behind Clinton, but how many of his supporters will follow? The fact remains that Hillary Clinton embodies most of the things that Sanders’ ‘political revolution’ opposes. The behaviour of the DNC (Democratic National Committee) and the irregularities during the primaries created great frustration and resentment among Sanders’ supporters (see here). How many of Sanders’ supporters will vote for Clinton?

Figure 4


Figure 4: Clinton’s favourability ratings among Sanders’ supporters. Source:

Second, Trump managed to accumulate an incredibly high un-favourability score (see here). But what does this actually mean? No one ever explained or showed the causal link between un-favourability and ‘un-votability.’ As the Zero Anthropology writer rightly notes, those who make this assumption commit the mistake of assuming rationality where there is little. It is, for example, well possible that some will vote for Trump, not because they like him (or his views), but because others (as for example the establishment) view him negatively. People vote for candidates on the basis of attractions and repulsions that cannot be logically understood (see here). Aside from that, Trump is not alone. Clinton has the second highest un-favourability score in history. How will this pan out? Nobody knows.

Third, there is the question of women and minority voters (see here). Trump loses among women, just as Clinton loses among men in quasi equal measure. In theory, this favours Clinton, because women are more likely to vote than men. On the other hand, in this election, 44% of US men are following this election ‘very closely,’ while only 31 % of women do. This may or may not be relevant. We do not know. Finally, there are the minorities. According to Cizzilla, the non-white vote will do Trump in (see here). It may well be, but in 2012 71% of eligible voters in 2012 were ‘white’ and nearly 74% of those who voted were of this non-existing colour. Projections show a slight change in these proportions as the following figure shows.

Figure 5


Figure 5: Growing diversity of voters by race. Source: PEW Research Center.

Many Hispanics and blacks will not vote for Trump, but these are also the most disenfranchised groups and the biggest victims of voter repression. Furthermore, it cannot be assumed that because Trump turns against one minority group (for example Hispanics) that this will lead him to lose votes of other minority groups (for example blacks) (see here). Finally, there is the biggest minority group of all, at least in terms of political visibility. The dominant ideology of the mainstream press is that many will not vote for Trump because what he said about Muslims. This is ideological whitewash. We know that islamophobia is widespread (see here).

Some people think that there is nothing to worry about. Those were the people who laughed when Trump announced his candidacy, the same people who said that Trump would self-destruct in no time. Today, Trump is the Republican nominee. Is that nothing to worry about, even if it remains ‘unlikely’ that he will win the White House?