Poll Politics and the Misleading of the American Electorate

The ancient, archaic AOL website is currently leading today’s news feed with an article reporting that Bloomberg has just published a poll indicating that Donald Trump now has a two point edge in the Florida presidential race. This is another example of chumming the waters with click bait designed to attract apprehensive voters from both camps to read and comment on the article.

AOL did NOT report that fivethirtyeight.com, Nat Silver’s website, is still showing Clinton as a 70/30 favorite but millions of people are following Silver and, seeing that 70/30 edge, thinking that it means something.  (AOL did however report that realclearpolitics.com (RCP) is currently reporting that Clinton still has a razor-thin 1.6 percent lead in Florida.)

Adding to the confusion: the propensity among many news organizations to conflate raw vote totals with electoral college votes.  This is important because the raw vote total – the so-called popular vote – doesn’t matter in a presidential election. Therefore, reporting national raw vote totals simply confuses an already too confused issue: what’s going on in the mind of the American voters, and who is (probably) going to win.

Thus, while RCP is reporting that Clinton is holding onto a thin lead in Florida, RCP is also telling visitors that Clinton still enjoys a totally insignificant 4.7 percent advantage in the NATIONAL raw vote totals.

In view of AOL’s sensationalist editorial standards, and click bait headline policies, it is difficult to take anything one finds there at face value, except for the fact that millions of people do just that every day.  A number of these factoids raise again my concerns about excessive reliance on polls in general, and on Nate Silver’s numbers in particular.

Here is a snapshot of the most recent polls from Florida, as listed on Silver’s website.  Note that this listing does not include the Bloomberg poll. Two of these thirteen polls have Trump ahead of Clinton. That would be great news, except for the fact that seven of the 13 polls listed here are rated by Silver himself at “C” or less, and two are not rated at all.

Only TWO of the 11 polls showing Clinton ahead of Trump are rated at “A” or “A-“. Without the Bloomberg Poll, Trump leads in only one of these polls but, with Bloomberg, Trump has two high quality polls that show him leading in Florida.

So, according to Silver’s summary chart (below), Clinton now has a 70/30 edge over Trump, but AOL is currently reporting – to millions of AOL users – that Trump is now two points ahead of Clinton on Florida, but that is only on the basis of one poll. AOL is also reporting that Clinton is still head of Trump according to RealClearPolitics.

AOL did NOT however report that fivethirtyeight.com, Nat Silver’s website, is still showing Clinton as a 70/30 favorite but millions of people are following Silver and seeing that 70/30 edge, thinking that it means something.

Take a closer look at what is going on here.

Clinton is head by three according to SurveyUSA, the ONLY “A” rated poll that shows Clinton ahead. Auto Alliance/ESA/Pulse Opinion Research (rated C+ by Silver) has Clinton up by three. SurveyMonkey (C-) also has Clinton up by three.

Now here’s where it gets very interesting. A second poll from the C+ rated Auto Alliance, taken during the same period of time (Oct 17 to Oct 24 for the first one and Oct 16 to Oct 23 for the second one) with the SAME sample size has Clinton up by FIVE rather than THREE points. Same time period, same sample size. A third Auto Alliance survey, spanning Oct 13 to Oct 20 and therefore overlapping the two previous surveys, has Clinton ahead by SIX points.

Same sample periods, same sample sizes, but three different outcomes. (I am using the LEADER Column not the ADJUSTED Leader Column for this analysis.)

Look through the results on this poll. Go back to Silver’s website and examine it yourself. (Note that I changed the listing order to show the most recent polls first. Silver lists his polls according to the weight he ascribes to each one, which is another questionable strategy since the observer is therefore prejudicing the outcome.)

Let me leave you with some questionable speculations.

Polls are only as good as the questions they ask, and the sampling methods used to select respondents.

Pollsters who use volunteer respondents are suspect because it is very easy for a modern marketing operation to flood such polls with partisans of the same persuasion. Polling organizations claim that they have algorithms that filter out these respondents, algorithms that are probably also responsible for sending me advertisements for feminine hygiene products on a daily basis.

If someone commissions a poll with the ulterior motive of showing Trump behind in order to light a fire under people who are on the fence but leaning toward Trump, the result will be a surprise turnout favoring Trump when the polls show Clinton ahead, and vice versa.

Therefore, the best way to improve Trump’s chances at the ballot box is to show Trump behind in the polls. No modern political campaign wants to be ahead at the polls; they all want to be the underdogs, or at least that is what they should want if they really knew what they were doing.

Of course, Trump, in his arrogant stupidity, has been saying that he doesn’t believe the polls….until the polls favor him. Equally stupid Clinton surrogates crow over their lead in the polls, when they should be decrying them. (To her credit, Hillary Clinton herself has started doing just that.)

In all likelihood, Clinton will win the only poll that counts, the one on November 8, but…once the dust settles….political scientists and information junkies like me are going to have to take a long, hard look at how polling itself distorts the electoral process.

Remember your Heisenberg: the observer always affects the outcome of the event or, in other words, vote early and often for Hillary Rodham Clinton, because this might be your last chance to vote in a Democratic election.