Reported for violent imagery.
Also, congrats on the one person (so far) who clicked on the link.
ETA: Modified link w/in quote as to maintain 2-click rule.
ETA(2): Apparently someone else did the same in the original post.
I just clicked. 4,148,585 members! The Dope is certainly doing something wrong.
IIRC Silver’s election model allowed you to see, on any given day, not only the big-picture odds on who would win, but also what the odds would be if the election were held that day.
I imagine that in mid-October, the same-day model did show Clinton with well over a 90% chance of winning, based on her commanding lead in the polls. The reason the real model was never that optimistic is precisely that it takes into account the possibility of something like the Comey memo happening.
Right, because that’s what I said.
I guess that’s just going to be how people argue back against what I’m saying in this thread, since what I’m saying is too much trouble to process and respond to. At least Sage_Rat gets it.
I’m a sophisticated layperson with respect to statistics. Notice that I have not criticized the model itself in any way. I didn’t say he did he weighting wrong, was looking at the wrong polls, etc. I am criticizing how he presented his outputs.
Something good. OK, here’s the crux of the matter. In the future, in a tight race like that of 2016, do you think it’s worth it to look at 538 (or its future equivalent) on a daily basis? I don’t. I think it’s enough to know that the race is tight, is extremely sensitive to changes in swing states, and no one can reliably predict the outcome. I think that’s the correct epistemological processing of the situation. If you think otherwise, then tell me why.
Yeah, why not? It’s entertainment in the end, and it is fun to see the shifting of the polls and how that adjusts the model, etc.
That is fair. I don’t think Silver has promoted the website as “entertainment,” however, or as information not to be taken all that seriously.
It’s punditry with data. And while there is a good deal of politics there is a LOT of sports stuff there as well. And a large portion of visits to their site likely involves folks who want to argue about politics/sports while using data (either supporting their side or discounting it when it is against their side).
Political horse race polling is all about infotainment. That applies when CNN or FOX or doing it or when 538 does it. The actual campaigns aren’t basing their stuff on 538 after all (though they’ll look at it to see what’s being said).
It’s meta-information. Commentary on the commentary. Date on how to read the data. Newspapers and websites were broadcasting, in the original more literal sense of the term, the results of individual polls as if they were meaningful. Silver, starting in 2008 but continuing into 2016, placed them into the larger context.
Polls were not disinformation and not even misinformation but something akin to malinformation. Just like the Gallup and Rasmussun examples I cited earlier, they conveyed only small and distorted aspects of the whole. The daily combination and tracking that 538 provided knocked the jagged edges off individual polls. Think of it as a system that guides a ship sailing into unknown waters by paring away the extremes and forcing the vessel into a safer middle that has the best chance of reaching shore unharmed. Not a perfect chance, but the best chance.
I cannot stress enough how useful such a system is and has been to anyone paying proper attention. Losing one subtracts from the body of best understanding. Maybe it was already torn to shreds; maybe a replacement worthy of the original will arise. Either way, the death of 538 is not easily dismissible as if it has always been valueless. Believing that is simply wrong.
There was state-by-state aggregation before Silver. I forget the name of the site I was looking at in 2004, but it had a map with electoral college votes for each state, etc.
So how Silver took that concept and advanced it, I don’t know. I am not of the opinion that the guy has been poor compared to similar sites, etc., merely, well, what I’ve already said. I think the 2016 election was a tough test for aggregation, and my takeaway was… what I’ve already said.
I don’t think it was, but I also don’t know if it had any advantage over pre-2008 aggregator sites. I also don’t see how aggregation has advanced since 2016. Maybe it has; maybe it hasn’t.
Probably electoral-vote. com.
ETA: Example snapshot from 2004:
Better yet, here’s there scorecard from the 2016 election on Nov 5:
I found Nate’s analysis and numbers much more informative than that. That site makes it look like it’s a done deal, with Clinton 100+ electoral votes over Trump. Nate reported a 30% chance of Trump winning. Like I said before – 538 prepared me for a Trump win psychologically, whereas the others made it seem it was a foregone conclusion.
Here is a pretty brutal take-down of Silver (stronger than my own opinion):
Blog post on Taleb vs. Silver:
http://quant.am/statistics/2020/10/11/taleb-silver-feud/
This Twitter post is linked in the above:
Taleb says:
This is a guy who is smart with respect to statistics, apparently, and he has expressed my frustration with Hillary being given this very high chance to win per 538 and then cratering a week or so later.
I certainly think it’s fair to say that he did a better job than other sites.
…I think Nate Silver, politics and statistics wonk is just fine.
I think Nate Silver, pretending to be a public health expert and denigrating the work of epidemiologists and those working to keep us all safe is bad…and dangerous. Not sorry to see him gone at all.
This is a weird statement. If I’m going to flip a coin every day for the next three days, I’ll say that I have a 50% chance of winning best-out-of-three on heads. But if today I flip tails, then afterward I’m going to tell you that my odds are down to 25%.
You may be right! If you get into the links above, you will see PhD’s in statistics arguing at length over this kind of thing.
It’s far outside my area of competence.
But if you read his posts he reported ad nauseam that just because he said there is a 75% chance doesn’t mean that its going to happen, it just means that its more likely to happen then not. He put out likely correct probabilities and explained in clear language what they meant. Its not his fault that people who didn’t read his explanations misinterpreted his results. The only alternative I can see is simply to not bother reporting any polls and just say everything is uncertain.
As for Talib’s criticism, its answered in the final part of the second cite. Nate at the time was providing information about the likelihood of Clinton winning if the election were held today. Since 2016 he has been providing two sets of predictions, one for the if it were held today and another taking uncertainty of future trends into about. The latter statstics were very boring because they always said it was a toss up regardless of any event news of poll. But again Talib’s criticism applied to every other Pole pundit out there. Nate did a much better job of handling uncertainty than anyone else. in the business and as a result became very high profile making him a target for those who want to nit pick. Heck as a professional statistician with a doctorate, I’ve got a few nit picks of my own, but I do have to say that overall I am very impressed with his methods.
I’m not sure what this refers to, do you have a Cite?
Maybe stuff like this?
Or this?
Or this?
He received a lot of backlash for his comments.
Lessons from the Covid War appeared this week and it is also not favorable.
I haven’t read the book so I don’t know how much it overlaps with Silver’s criticisms. That we did almost everything wrong does not seem to be an arguable issue.
Thanks, I hadn’t seen this before. Yeah that’s bad. I thought perhaps that he was simply criticizing analysis methodology of certain papers. But here he is clearly speaking beyond his area of expertise. This is indefensible.
If that is true it is indefensible.
Sad that once people are famed for one area of expertise, they start to presume they are equally experts in everything. See Scott Adams. Or for that matter Cecil Adams.