patrick.net

#politics

Good article here:

https://fivethirtyeight.com/features/the-fundamentals-favor-democrats-in-2018/

Relating to yesterday's discussion on polling-here's a good excerpt: (bold added)

"[It’s] been interesting to see how television pundits adapt to the post-2016 environment. Pretty much everyone on Monday morning’s “Morning Joe” panel predicted that Gillespie would win in Virginia despite Northam’s modest lead in the polls, for instance…

[The] segment was a bit worrisome in that it suggests that political pundits and reporters learned the wrong lessons from 2016. Contrary to the conventional wisdom, the polls weren’t that far off last year — they were about as accurate as they’d been in past elections. But they were filtered thru a lens of groupthink that was convinced Trump couldn’t possibly win — and so pundits routinely misinterpreted polls and ignored data showing a competitive race.

It’s healthy to take away the lesson from 2016 that polls are not always right… But that polls aren’t always right doesn’t mean that one’s gut instinct is a better way to forecast elections. On the contrary, the conventional wisdom has usually been much wronger than the polls, so much so that it’s given rise to what I’ve called the First Rule of Polling Errors, which is that polls almost always miss in the opposite direction of what pundits expect. That the “Morning Joe” panel thinks Gillespie will win might be a bullish indicator for Northam, in other words."