Predictions
of elections reside in a world of experts, the poll dancers. Let’s check ‘em out.
Most predicted a Clinton win by about 3% of
the vote. Nate Silver, the guru of
statistics and outcome predictions, predicted chances for Clinton win at 70%.
Other outlets predicted a Clinton win at 85%. Even the odds-makers (bookies) gave her at
least 2-1 odds for a win*. Some more so. Indeed, so confident was her team that
Hillary Clinton left early from home for her hotel room near their NYC HQ anticipating
the prospect of making an early winning announcement. All were wrong. Only a very few predicted a Trump win, and
they were all discounted. OK.
Analysts
will be analyzing away for years, among them the so very many who were wrong.
They will conjecture, debate, write articles, write papers, write books, and
appear on television, ad nauseam.
The point here is not to give our own description of why. Our point is about the “who.”
The
“who” were acknowledged experts. All understood statistics; all were well-schooled
in polling; all were well-studied in prediction techniques; and all were well-connected
in political circles of both flavors. Almost all were dead wrong. I’m certain if asked the National Academy of
Sciences would have approved of their methods, their theory, and their
conclusions.
This
leads to consider the expert prediction of the future. The experts make a model that worked in the
past and use it to project into future.
In this case, the predictions were only
days into the future**. They were dead wrong. Even weather prediction is better these
days.
The
moral of this story? Probability. It is a measure not of certainty, but
uncertainty. It is wise not to blindly accept the view of the experts, no
matter how lofty their credentials. They
are only human, condemned by their methods, and victims of their own thoughts. So
many were wrong, and wrong big. This
gives us pause.
When
the unusual happens often, as in politics, a healthy distrust in predictions is
well advised going forward. We learned
this in sports when we hope for the underdog, because they can win despite
percentages.
Next
election, the political pollsters, these poll dancers, will remember the
“forgotten man,” but they may forget others. They cannot know. If they predict the correct outcome, they
will rest on laurels. If not, they will
analyze ad nauseam.
Poll
dancers really do provide a level of entertainment for every election, sexy for
some, depressing for others. This cycle, most have been completely undressed.
Yet, they remain always ready for the next show.
------------------
*Note
we have given three different measures, percentage electoral margin, percentage
for a win, and odds. The second two are
compatible, but the first is another type altogether. Yet another source of
confusion. Here is a technical point you may be hard-pressed to answer. If a dozen polls indicate candidate A has a
75% chance of winning, does this mean the chances of winning are exactly 75% or
higher? No, if all use the same methods.
It remains a probability. But what if
they use different methods? Then, there
can be no answer without more information.
The curious point is what “more” information do we ask for? This is unknown, and almost unknowable.
**Recommended
reading: Philip Tetlock, Superforecasting: The Art and Science of
Prediction.
Comments
Post a Comment
Please Comment.