I’ve promised to give you all an overview of the general election, but before we get into all that, there is something we need to discuss first: polls. There’s a poll out that says Hillary is within one point of Trump in Georgia. It’s going to be a Hillary blowout! But wait: the same guys say Trump is within the margin of error in Pennsylvania and Ohio. It’s going to be a Hillary disaster!
Quinnipiac Maypolling: The Green Party is winning!
Polls are a statistically valid small sampling of a population in order to model within an acceptable margin of error and an acceptable degree of certainty the characteristics of that (much larger) population. An individual poll can never be fully trusted. That’s why sites like RealClearPolitics, Pollster and FiveThirtyEight are so popular: they apply math-y techniques across multiple polls to smooth out the error rates among them. But what if all the polls you’re smoothing all made the same types of mistakes?
Right now, in May, I am only looking at general election polls in order to indicate broad trends. Who is trending consistently ahead? What population is trending for which candidate? What states are following recent blue/red trends? Where in all these numbers is new ground being broken? We can examine these trends, but to assign hard percentages to them and call that “reality” is a fool’s errand.
The Likely Voter
The reasons for this are many. First, it’s difficult to model what a “likely” voter is in 2016. The Republican Primary brought out a strange mix of voters. The minority turnout is difficult to predict, I’m not sure how strong the African American turnout will be versus the last two elections. Polls show that both candidates are underwater in their approval ratings, but Trump is worse off to a historically bad degree. Significantly, the type of disapproval for each is not measured, nor the strength of those feelings of disapproval. Anti-Trump or anti-Hillary? I believe we’re comparing apples to oranges there, but it’s always hard to measure who will come out to vote FOR someone they’re not particularly enamored of, and/or who will come out to vote AGAINST someone they loathe with the white-hot heat of a thousand suns.
Another reason I am less than enamored of polls right now is that polls show a bias depending on who is running them. A bias can be introduced just by the choice of methodology (e.g. by under-sampling cell phones) or by the questions themselves (for instance, by not varying the order of choices – there is small but measurable unconscious bias towards selecting the first name you hear). These biases can be accidental or deliberate, and they can be occasional or constant. What’s weird about 2016 so far is that polls I used to consider reliable have shown consistent skews in the primaries, and I don’t know if they will smooth in the general.
Something else weird about these general election polls is that they are both measuring well-known people. Usually at this time, presumptive candidates have not been fully introduced to the general election voting population, and have not been framed by their supporters or by their opponents in a narrative that would make sense to the voters. Such is not the case with Hillary, a well-known political entity. You’d also think that such is the case with the Donald, as well, except that once again we are comparing apples to oranges. Donald is a well-known ENTERTAINMENT commodity. His business record, while assumed to be well-known, has actually not been fully aired; nor has his personal life. As a politician, he has no record, not even within the context of this campaign, because he has yet to take a coherent, consistent stand on ANYTHING.
So while support in the general seems to be settling in for Trump, I believe that his numbers will continue to drop as the Clinton War Room, the DNC, and – FINALLY! – the media begin to make all his many many many shortcomings much better known.
So, again, the polls are measuring two products that some people think they like, some think they don’t like, but know one really knows for sure what Trump is, and whether any individual will actually get out of the chair, find the keys and go stand in line to vote… well, even the person you’re asking isn’t 100% sure they’re going to do that.
Another reason the polls taken right now in this cycle are particularly untrustworthy is that the candidates are in two very different stages of their campaigns. Trump has vanquished everyone earlier than the stop-Trump movement had intended, prior to its ignominious collapse. Without a focal point to coalesce around, Republican voters are starting to fall in line and respond to polls in a business as usual manner. Hillary has had to continue her primary longer than anticipated due to Bernie Sanders still doing his thing. As such, within the context of the Democratic primary, a lot of anti-Hillary sentiment has yet to settle out.
In just a regular Red vs Blue year, there are a lot that can go wrong with general election polls in mid-May. There are even more things that can go wrong with them this year. So read them in context, with a very discerning eye and a healthy dose of skepticism. And remember that the people trumpeting one poll over another are doing so for very selfish reasons; examine those, as well, but most especially remember this:
The media is in the business of making money and have invested a huge amount in election coverage. It is in their best interest to make something look like much more of a horse-race than it is.