How FiveThirtyEight Failed to Predict Hogan’s Victory

1
Share the News


fivethirtyeight

After FiveThirtyEight, the website of statistics whiz Nate Silver, gave Lt. Gov. Anthony Brown a 94-percent chance of winning Maryland’s gubernatorial election Tuesday morning only to have Republican Larry Hogan beat him by five points, they had some explaining to do. And they did.For one thing, FiveThirtyEight pointed out that 94-percent favorites are expected to lose 6 percent of the time. OK, fair. Nate Silver wrote that the Brown win and other gubernatorial upsets may indicate “that pollster ‘herding’ — the tendency of polls to mirror one another’s results rather than being independent — has become a more pronounced problem.” And FiveThirtyEight’s model relies exclusively on polls.

But there was a Maryland-specific problem as well. “[T]he most reliable public pollsters stopped surveying the race a month before Election Day,” meaning the Baltimore Sun and the Washington Post. Their latest polls were conducted in early October and put Brown ahead by an average of 8 percentage points. Later polls were all partisan, and FiveThirtyEight doesn’t look at those.

Interestingly enough, as pointed out by the Baltimore Business Journal, there was an internal Hogan poll from Oct. 30 that predicted Election Day perfectly, showing a 5-percent Hogan lead.

 



Share the News

1 COMMENT

Comments are closed.