Even in exit polls with the NY Times seal of approval.
According to a Capitol Weekly early-voter exit poll, Hillary Clinton was leading Bernie Sanders by less than 10% in the Los Angeles area vote-by-mail balloting ahead of last Tuesday. According to results posted to the Los Angeles County website, Clinton was winning vote-by-mail ballots by 66.6-33.4%, for a discrepancy of more than 23%.
The discrepancy cannot be easily explained by demographic factors: the results of the Capitol Weekly exit poll were weighted by age and race. Moreover, the exit poll had 21,000 respondents, and was praised–prior to election night–by mainstream elections journalists, including Nate Cohn of the New York Times.
Twenty-one thousand respondents. That’s a very respectable size for the absentee ballots, even for California, if the Washington Post is to be believed. It truly is an astounding coincidence how an exit poll with a sample size that large could be off by 23 percentage points. And one that was specifically weighted to take into consideration age and race, factors that were known to benefit Hillary more than Bernie.
A poll that surveyed absentee voters by sending emails to people known to have returned their June 7th ballot as of June 4, 2016. Not guessed at, not presumed, but people they knew had voted,”weighted by geography, party registration, age, ethnicity and gender to match the voters who have already cast ballots…” Using such weighting eliminated the effect that younger, male Sanders voters were over-represented in the poll. In other words, they had already adjusted their raw data to account for that.
Indeed, this was the most Clinton-favorable poll of the many that were conducted. All the rest had Clinton leading by only two percent (2%) with margins of error ranging from 4-5 percent. So how could this most-favorable Clinton poll still screw it up that badly? Such a respected polling company, too. I mean I could understand it being off by say 10%, even though that figure would still likely far exceed the poll’s margin of error, but 23 percent? That’s simply astounding to me.
Exit polls this year really do suck. At least in the Democratic primaries, anyway.
