Monday, August 3, 2020

So You Want to Forecast an Election...

Right now, the Economist forecasting model gives Biden a 91 percent probability of winning the electoral college and a 99 percent chance of winning the popular vote. High!

This has led to some questioning of the model from various quarters, including the redoubtable Nate Silver. Silver leveled his criticisms on Twitter but Andrew Gelman, the Columbia statistician who developed the model with G. Elliott Morris and Merlin Heidemanns, put his replies on his blog. As Gelman says:

"[I]n general I don’t find twitter to be a good place for careful discussion...I like that Nate is publicly saying what he doesn’t like about our method. I’d prefer even more if he’d do this on a blog with a good comments section, as I feel that it’s hard to have a sustained discussion on twitter."

Boy, does he have that one right.

Anyway, I recommend reading Gelman's post. It's clear and well-written, avoiding unnecessary technical jargon. It'll help you understand the voodoo that forecasters do--both the strengths and weaknesses, ,which may be a bit different than you think. Note that Gelman and co. are revising the model a bit, though he doesn't expect the results to change all that much.

Gluttons for punishment can peruse the comments section to his post for more interesting observations and Gelman's typically lucid replies.

https://statmodeling.stat.columbia.edu/…/thinking-about-el…/

Image may contain: 2 people, text that says 'Right now, our model thinks Joe Biden is very likely to beat Donald Trump in the electoral college. Chance fwinning theeectora college Joeiden Democrat Chance fwinning theostvotes 10 around9 or 91% Predictedrange electoral collegevotes(270towin) Donald Trump Republican better or 99% 19in 249-413 around1 in 10 or9% less than or 1% 20 125-289'

No comments:

Post a Comment