Does Trump have a decent chance of winning? Pollster averages say so.

As Hillary Clinton wraps up the Democratic nomination, attention turns to the contest between Clinton and Trump. The current Huffington Post Pollster average has Clinton and Trump alarmingly closely matched, 42% to 39%.


The closeness of the polling is leading some to believe Trump has a good chance at winning. But that belief is severely misplaced.

National Polling Versus States

One important thing to remember is that a national poll of the popular vote has little bearing on the outcome of a presidential race that’s decided by electoral college. This does not necessarily immediately advantage Clinton. There have been several Presidential contests where the winner of the popular vote has lost the election. If national polls are to be believed, Trump could still win even if he is a few points behind Clinton nationally. Luckily, this is not the case in this election.

Looking at individual states makes Clinton’s prospects look much better. State polls show Clinton winning in North Carolina (Romney +2 in 2012) and running close behind Trump in Georgia (The last Democrat to win Georgia was Bill Clinton in 1992). Overall, averaging state polls (and when no polls are available, 2012 results) gives Clinton 324 electoral votes. That’s over 60 more than the 270 required to win, and it doesn’t include any of the 75 electoral votes deemed a toss up. If you assume Clinton and Trump slit electoral votes in tossup states, Clinton would win by almost 200 electoral votes. 

Bernie or Bust?

Although Clinton has all but clinched the nomination, Sanders plans to continue campaigning through the Democratic Convention. On the other side, Trump wrapped up the Republican nomination weeks ago. Republicans have had plenty of time to consolidate around their nominee, but the continued contestation of the Democratic race means that a substantial portion of Sanders supporters refuse to support Clinton in the general election. For instance one polls shows that 25% of Sanders supporters won’t vote for Clinton come November.

Luckily, virtually all partisans participating in primaries end up voting for the nominee by election day. In 2008 Democratic Primary, there were many Clinton supporters claimed to refuse to support Obama in the general election. One poll of North Carolina even found that the majority of Clinton supporters weren’t likely to vote for Obama in the general election. Obviously, they ended up changing their minds as Senator Obama became President Obama with a resounding victory over McCain in November (Including a victory in North Carolina). All the evidence shows that the same thing should happen with Sanders supporters voting for Clinton in the general election, despite what polls may claim today.

Methodological Issues

Furthermore, the quality of the polls included in the Huffington Post average are suspect. The last three polls to be conducted (and those that are weighted more heavily in the average) are all internet based. The next two are “automated” phone interviews, where interviews are conducted by machines instead of people. The next 5 most recent are all internet polls.

The first poll that actually uses live interviews is the 11th oldest in the average, meaning it gets little weight. It’s results? Clinton is up by 13 points. In fact, when you only include live phone interviews, the average comes out to Clinton 49, Trump 40, a 6 point difference in spread from the average that includes all polling methodologies. 

Average of Polls that use only a Live Interview Methodology



What’s driving the differences between live interviews and online polls? When you dive into the cross tabulations, one thing that sticks out is the spread of Hispanic voters. Morning Consult, an online poll that has Clinton up by just two points. Among Hispanic voters, they have Trump winning 31%. Survey Monkey, partnering with NBC News, gets similar results from their online poll: Clinton is ahead by three points with Trump getting almost 30% of the Hispanic vote. Does that really seem believable for a guy that tweets out this?

There is more evidence that these polls are wrong than Trump’s tweets. For one, as a point of comparison, Mitt Romney won 27% of the Hispanic vote in 2012, perhaps as low as 23%. Trump exceeding this margin seems ludicrous.

Secondly, other polls of Hispanics give Trump remarkably high unfavorability ratings with Hispanics. Latino Decisions, a polling firm that specializes in latino opinion, finds Trump unfavorability at 87%. Back in January Gallup found the same statistic at 77%, and CBS found it at 82% in April. According to Latino Decisions that high unfavorability translates to only 11% support for Trump, not 30%. They called 2,200 latinos using bilingual live phone interviewers. Compare that to the 355 online respondents in Morning Consults Hispanic sample that gives Trump 31% of the Hispanic vote.

While most live interview polls refrain from giving Hispanic cross tabulations that would make them directly comparable to their online counterparts, the difference can be inferred from the overall results. And regardless of the direct differences, any poll that gives Trump 30% of the Hispanic vote (like the top to in the Huffington Post average) is very likely wrong.  

It’s clear that national polling averages are at best misleading and at worst clearly biased in favor of Trump. While the election is months away and he certainly has some chance of winning, nothing about national polling averages suggests the race is close.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s