Ok – if you visited the site prior to the election, you saw the page where I estimated when they’d call the state versus when the state closed.
Here’s the data points: The call time comes from NPR. Any poll times that I got incorrect as part of the original posting were adjusted to the correct call time. I call it a successful call if it’s within 15 minutes (partial credit if within 30).
What went well:
The easy states (those that were blowouts – candidate differences >10%)
New Hampshire (got lucky)
Well, most everything else. You can’t estimates based on polls unless you want to use it as a baseline for which states will be called later. I went with the generic assumption that states will be called based on poll accuracy. Major flaw in theory is that you have to factor in how accurate the polling is. More on that in the next post.
Going to leave this to the experts, but it’s fun to try and guess every four years.