I’ve posted Entry #327 to my weekly Valuation-Informed Indexing column at the Value Walk site. It’s called Using Historical Return Data to Guide Your Investing Strategies Leads to More Uncertainty, Not Less.
Juicy Excerpt: Silver’s take was rooted in two realities: (1) three-point polling errors are common; (2) there was an unusually high number of undecided voters going into the election and that increases the chances for seeing a polling mis-prediction. Silver did not ignore the polls; his take was that Clinton had a two in three chance of winning because the vast majority of polls favored her. But he took notice of the limitations of polls, that there are some things that polls cannot measure; for example, some voters make up their minds on the day of the election and it is thus impossible for polls taken before the day of the election to predict effectively how they will vote.
Many Trump supporters dismissed polling altogether; not knowing what the polls said permitted them the luxury of giving in to their pro-Trump bias and concluding that he was certain to win. Many Clinton supporters pointed to the pro-Clinton polls as scientific evidence but failed to consider the limitations of the science, thereby permitting themselves a certainty as to the outcome that they pretended was rooted in what the polls showed but in reality was not. It shows respect for the science of polling to note its limitations. Truly scientific explorations demand an intellectual discipline that leads to more uncertainty, not less.