The editors of the Wall Street Journal remark:
Either candidate had a path to win by our deadline. But it’s already clear that the biggest early losers are the pollsters. The mainstream media polls all had Mr. Biden winning in a walk with a popular vote margin in the upper single digits. They were off in particular on Florida. The outlier pollsters like the Trafalgar Group, often derided by their colleagues, seem to have better judged the electorate.
Mr. Biden still seems likely to win the popular vote, but the margin will be narrower than predicted and the Electoral College was still up for grabs. Lamenting the Electoral College is a hardy perennial, but both parties know that it’s the measure of victory or defeat. Both campaigns focused on those swing states.
The tight race reminds us that democracy is surprising, and humility is good journalism practice. Especially in this year of Covid-19 and an economic recession, Democrats thought they were set up for a blue wave. But while they looked set to retain a House majority, a Senate Democratic majority was moving out of reach as GOP incumbents held on to win in Iowa and North Carolina.
The primary stumbling block to governance is not how divided the country is but how divided the Congress is. At present the most right-leaning Democrat in the House has views farther to the left than the most left-leaning Republican. That in turn is a product less of differences in the electorate than in our dysfunctional, self-serving political parties.
The coming reapportionment will further muddle things. Although states like North Carolina, Georgia, and Florida are purpling, the bluest states are going to lose districts. Sadly, not enough people are taking the underlying message seriously: the governing strategy of states like California, New York, and Illinois is deeply flawed. It’s not just the climate; California has a nice climate, too.
Well, my prediction was completely off and the modeling/polling accuracy so far looks very bad. At this point, it still seems likely that Biden will win but it was not nearly as strong of a showing as I estimated.
At this moment (10:45am CST) it’s still hard to say. Biden could win but it will be by razor-thin margins. Who will control the Senate is still largely up in the air but some of the Republican incumbents who were thought might be defeated are holding steady.
The problem with that is that the Congressional leadership in both political parties have little experience with compromise—it’s been power politics all the way. They just don’t look at things that way.
And how do you forge compromises with people you consider morally reprehensible?
On compromise, I don’t know how to change the hearts of the Congressional leadership; since they are doing what has worked for them.
Almost feels like it would take a gruesome “car-crash” before the Congressional leadership realizes or the people select new Congressional leadership that realizes compromise is a necessity of a representative republic.
How do you compromise with the reprehensible? In the dark, when none is looking, an issue that is electoral suicide to punt and a compliant press agrees.
I feel bad for Nate Silver; he didn’t do the polls, but he rates them, polishes them up and resells their aggregate as his own product. I suspect that most of the close states will drift towards Biden as mail-in returns come in.
The Senate races look worse, 538 reported that Democrats had a three out of four chance of taking the Senate, and Gideon just called to concede to Collins. There do not appear to be any polls this year that showed Collins winning, let alone getting a majority in the first round.
Next you’ll be telling me that astrology doesn’t work or that professional wrestling is fixed.
My modest proposal for regulating polling is to allow polling to continue; just prohibit taking money for doing it.
On polling; the consensus was the methodological error in 2016 was they didn’t weigh samples by education, since a large proportion of Trump supporters were thought to be “white non-college educated”. So pollsters fixed it by weighing by education.
There was an alternate theory it wasn’t “education” that was causing sampling issues, but individuals that support Trump or the Republican party were less likely to respond to polls. The theory didn’t address whether this was due to desirability bias (shy Trump voter) or more prosaic reasons like they are more frequently out with family and community, and have less time to answer phone calls from strangers.
I think this election proved it was the 2nd theory that was true. That implies polling cannot be accurate until one can measure how less likely these potential voters are to respond to polls. Which is an impossible circle to square.
As I’ve mentioned Scott Adams has another explanation: Trump supporters trolling pollsters. That’s actually a much worse problem than refusing to answer since it doesn’t take a lot of trolls to introduce a lot of error.
I don’t have a link for it, but in the 538 election live-feed last night Dan Hopkins wrote this after Florida was called, which I copied:
Besides the factors already mentioned, I have to wonder if the pollsters are calculating the margin of error correctly.
As Galbraith said about economic forecasting, the purpose of political polling is to make astrology look respectable.
That comments gets at something I read. After the 2015 UK general election where the polls missed by ~5% (outside the margin of error), a poll consortium did a research study. They picked addresses at random and physically sent someone to the address to do a survey. They would try contacting someone 9 times before they gave up. The study found using physical contact, they could “re-create” a poll for the election was only 1% off.
Most of the people I know send their calls to voicemail, and some hate it so much that their mailbox is always full.
I don’t know how to quantify a survey self selected for those so lonely they are just waiting for someone to cal.