That's correct. The overall polling was very accurate in predicting the popular vote. State by state voting was less accurate, and the margins were very thin. But on the morning of the election, 538, who I consider the gold standard on polling analysis, had Clinton as a 75% chance to win. In baseball, that's about the odds that a given batter will make an out. But a player for the Minnesota Twins just reached base in 13 consecutive at bats. A small sample size was the reason that everyone talked about how wrong the polls were. Understanding statistics and probabilities is the key to understanding polling data. I think if the election had been held a week later, the polling shift after the Comey announcement would have been more pronounced, and the odds smaller. Polling has become less reliable, almost certainly due to the advent of mobile phones as the primary phone in almost every household. I don't think the methodologies have caught up yet.
