With 16 days – and about 160 news cycles – remaining until the presidential election, if you look at the news and data coming out of establishment media outlets you get the sense that the election of Hillary Clinton is a foregone conclusion. Yet, if you look at polling data on alternative outlets or even look at social media traffic you get the sense that Donald Trump will win in a landslide. The discrepancies in the information trends dictate that someone must be wrong. Unfortunately, without the tendency or the wherewithal to investigate, potential voters are being bombarded with headline numbers that are rapidly absorbed as indisputable fact due to repetition.
We should never forget the effect that confirmation bias has on this process. Whatever you want to believe, there is a way to manipulate data in order to justify your belief. I participated in a version of this ‘goal seeking’ back when I was an investment banking analyst. Rather than build up the assumptions and properly model out the most likely outcomes based on fundamentals and probabilities, a range of numbers would be deemed acceptable and then the calculations would be performed in a way to justify those numbers. The objective was to win clients. A valuation figure presented to the company needed to be as high as possible without being perceived as unreasonable. Once we won the business and as reality set in more and more, the valuation numbers invariably came down because they didn’t meet the rosy forecasts we had advertised in order to gain the business in the first place.
Why aren’t we taught logic and statistics? It’s a literal tragedy that our populace is not prepared to think critically and quantitatively. Without going off on a roller coaster of a tangent, I’ll simply say that it is not entirely accidental in my view. My degrees in mathematics, statistics, and economics don’t make me anyone special, but I have spent years interacting with these topics and understanding the shortcomings of data analysis – particularly how information can be manipulated using statistics and how theoretical models often don’t apply to reality. The good news is that you don’t need to spend years interacting with the material if you have five minutes to spare, common sense, and an open mind.
Polling is tricky business. Polls are conducted because it would be too time consuming and expensive to contact every single person. Even with our current communications technologies the costs are still prohibitive. Statistics as a discipline attempts to estimate measurements and probabilities given incomplete information. Various techniques are available, but the most basic point to keep in mind is that a random, large sample of data is more accurate than a nonrandom, small sample of data.
But that’s not all. The groups and amounts that are sampled have a massive impact on the reliability of a figure. Political polls that are regularly conducted have various ways to tweak their phrasing and adjust parameters in order to ‘goal seek’ the results they are looking for. One of the most obviously detectable is the manipulation of base rates. Statistics jargon aside, imagine taking a poll to figure out whether New Yorkers prefer the Yankees or the Mets. Now imagine that you gathered your data outside of Citi Field where the Mets play. Common sense tells us that the result of this poll would not represent the overall preference of New Yorkers due to the sampling bias of the base rate fallacy. Simply put, too many Mets fans will be counted.
So when ABC News comes out with a poll that has Hillary Clinton leading Donald Trump 50% to 38%, most will only digest the blaring headline. Clinton up by 12% over Trump: the election is over! However, something doesn’t quite fit. Trump is holding multiple rallies every day and drawing crowds in the tens of thousands each time. When Clinton has a rally – a rare happenstance – she’d be lucky to get one tenth of the turnout that Trump gets. When a poll comes out that doesn’t match the facts on the ground, it’s worth checking into the base rates that the poll publishes in the fine print. Here’s what ABC News had to say about their poll conducted by Langer Research Associates and Abt-SRBI:
This ABC News poll was conducted by landline and cellular telephone Oct. 20-22, 2016, in English and Spanish, among a random national sample of 874 likely voters. Results have a margin of sampling error of 3.5 points, including the design effect. Partisan divisions are 36-27-31 percent, Democrats, Republicans, independents.
A poll where 25% more Democrats are sampled than Republicans cannot provide a valid estimate because it commits the base rate fallacy. The professionals who conduct these polls know this, and they can come up with all sorts of hand waving justifications for the discrepancy of the sampling rates if they have to. This sort of data manipulation shouldn’t be surprising in an era of media collusion, evaporating ratings, and imploded credibility. Enough people have been awakened to the reality that establishment media outlets are the ultimate peddlers of fiction. If you disagree with me, get your news from CNN, and view them as mostly credible and objective, then you are to be congratulated for overcoming your confirmation bias and making it this far.
But why is this manipulation taking place?
My general inclination is that the objective here is to drain the confidence from Trump’s voter base and embolden Clinton supporters. Feeling fatigued and defeated – so the political calculation goes – Trump voters will be less likely to vote because they feel there is no chance Trump will win because that’s what various polls and talking heads are saying. That might be true in normal times, but herein lies the ultimate media manipulation miscalculation: these are not normal times. These are the same people who told everyone that Trump would never win the nomination and that Brexit would never happen. They weren’t making educated and statistically sound predictions based on reality. They were attempting to influence the results through perception management. It’s the same thing that the Federal Reserve and the IMF do when they forecast massive economic improvement only to be horrifically wrong year after year. How many ridiculous mistakes from the supposed experts does it take for everyone to realize that for them agenda matters more than accuracy?
I believe that the repeated victory laps will actually backfire and that we are headed quickly towards another Brexit moment. I think that Hillary voters are more interested in casting a vote against Trump than a vote for Hillary. They know that Hillary is corrupt, but they are scared of Trump. If the media continues to trumpet headlines declaring that a Clinton victory is a foregone conclusion, then a decent chunk of her voters will breathe a sigh of relief and, lulled into a false sense of security, not actually get out and vote because they aren’t truly passionate about their candidate. Trump voters, on the other hand, are convinced that the system is rigged and that the media is in the tank for Hillary. They are totally passionate about Trump – as evidenced by his rallies – and they will vote for him no matter what the polls say. They will never be discouraged. They view themselves as disenfranchised by the political class in Washington and see a vote for Trump as their only way to fight back.
Time will tell, but I can’t shake the feeling that the media manipulation miscalculation is in full swing.