Whatever explains the differences between Sweden and the other Nordic countries during the past few weeks, I think it's pretty clear that it doesn't have much to do with policy.
Some of you point out that mobility data don't necessarily capture everything that's affected by policy, which is fair enough, but the reality is that policy has been very similar across Nordic countries for a while and if anything has even been more stringent in Sweden recently.
If you want to claim that it's policy, you should be able to pinpoint some specific things other Nordic countries are doing that Sweden isn't and which could plausibly explain the difference, but I don't think you can.
(By the way, there is a mistake in the first chart above, the y-axis should be about cases per million, not deaths per million.)
Also, by comparing Sweden to its neighbors, I don't mean to endorse the view that it's methodological preferable. As I explained repeatedly, I think this view is profoundly misguided, and I think this shouldn't even be controversial.
I just want to point out that, while the people who make that comparison because they endorse this view continue to ascribe differences between Sweden and its neighbors to policy, it's not a very plausible for what happened during the past few weeks.
• • •
Missing some Tweet in this thread? You can try to
force a refresh
Here are some true claims: 1) The bet on which Sweden originally sold its strategy, that it would reach herd immunity quickly, has failed. 2) It doesn't follow that it was the wrong strategy. 3) The predictions of people who oppose Sweden's strategy were also completely wrong.
It's interesting how everyone keeps talking about 1, but systematically forgets about 3. According to the predictions opponents of Sweden's strategy made last Spring, there should have been more than 65,000 deaths by now, but this has been memory holed.
The truth is that, if you had told them back in April that there would only be 6,000 deaths in Sweden right now, they would never have called that a "failure", because they had predicted something far worse. It was only labeled a "failure" and usually far worse later 🤷♂️
The long term impact of school closures, not the temporary income drop of their parents, accounts for 90% of children’s overall welfare loss. The effect is concentrated on low-income families and closures also result in a substantial fall of government revenues in the long run.
Now, you can quibble about the model, but it should be clear that, if you only look at their immediate effects (which is what most pro-lockdown people do), you’re going to vastly underestimate the long term impact of lockdowns.
Everyone is making this dumb argument that, because a vaccine is coming, lockdowns are obviously the way to go. But this changes *nothing* unless you thought that the cost of keeping the epidemic contained for another 3-6 months but not for 1-2 years passed a cost-benefit test.
Je suis le premier à dire que le gouvernement est totalement incompétent et gère très mal cette crise, mais je suis tout à fait d'accord avec ce thread, il y a un gros biais rétrospectif à l'oeuvre dans les commentaires sur la reprise de l'épidémie.
Je me souviens très bien avoir remarqué ce qui à l'époque ressemblait à un tassement dans les données sur les hospitalisations fin septembre et m'être dit que ça allait peut-être se calmer.
D'ailleurs, je parle des données sur les hospitalisations parce qu'elles sont a priori plus fiables que celles sur les cas, mais on voyait la même chose dans les données sur les cas.
This spreadsheet is based on the best estimate of the number of uncounted ballots, not PA's dashboard about absentee and mail-in ballots, which is not updated correctly. If Biden's margin among mail-in ballots remain the same, he should win the state by ~80,000 votes.
This is much closer than what people in the media, who are using PA's dashboard about absentee and mail-in ballots, have been assuming, but it still looks as though Biden should win the state.
To be clear, this projection assumes that, in each county, Biden has the same margin in the remaining mail-in ballots than he's had in the mail-in ballots that have been counted in that county so far, so it does take into account where the outstanding ballots are from.
The dashboard everyone is using to get the number of outstanding ballots in PA says 763,311 mail ballots remain to be counted, but when you look at the results, it seems that in fact many of those have already been counted. So Trump's lead might be more secure than it seems.
I'm trying to check if, as I strongly suspect, the same thing is true in Butler County, but apparently they're having a little technical difficulty... I'm pretty sure there are significantly less than 763,311 outstanding ballots in PA though.
This should be normalized to account for demographic change in the country though. My guess is that, while Trump did overperform among minorities, the current narrative exaggerates the phenomenon and how much it mattered. The real action was likely among whites, same as in 2016.
What I mean is that "non-white" is a heterogeneous category and, in particular, it’s increasingly dominated by hispanics, among whom Republicans have always done better than among blacks.
I guess "normalize" isn't quite the right word, "adjusted to account for the change in the composition of non-whites" would have been better, but it didn't fit. In any case, while I do not doubt that Trump improved among hispanics, I doubt he did as well as Bush in 2004.