Update, 3:54a.m. Nov. 9:
I think we can safely all admit what a stupid idea VoteCastr from Slate was. Amazingly, left-leaning Slate saw the race going all toward Hillary. The photo at left is a good sample.
It shows Clinton eeking out a win in Florida and Iowa. Well, we now know that’s not true.
Generally, I find myself OK with basic polling because it doesn’t say so much who is going to win the election but where people stand at this moment, what might happen if the vote were this minute.
But VoteCastr is a prediction based on exit polling and past trends. It’s just a stupid idea. Remember the mantra: Journalists do a much better job telling people what already happened rather than what will happen. The more we focus on that, the more likely we are to give people what they need and want. (We also don’t fall into narrative bias, the idea that we tell the story that we think is right and that we want to have happen and that seems to fit what we’re looking at. Margaret Sullivan at the Washington Post has a great piece on exactly this.)
From 6p.m. Nov. 8:
Well, this election day of all days, Slate is testing a new Vote Tracking system that scares the crud out of me.
It combines voter turnout from previous elections with the results of those elections and compares it to turnout now to take a stab at who is winning.
My basic belief: journalists do a much better job telling the world what just happened, not what will happen (I forget which journalist I heard first say it, but it has stuck for many, many years).
What really scares me here is the Tuesday morning quarterbacking that will come. If this is accurate, then guesses become truths in future elections: “The Democrat has always won when the turn out in Macon, Georgia’s 12th precinct on election day has hit 21 percent of likely voters by 11 a.m.”
The effect that conclusions like that could have on voter turnout in the rest of the country bothers me. –Thurston