DSeid
September 16, 2016, 10:13pm
121
Well a post on 538 today is entertaining!
Is Silver reading my posts? Or at least Wang’s?
… Another way to hedge against short-term swings in the polls would be to assume that polls tend to revert to where they’d been previously. Over the course of the campaign, Clinton has been ahead by an average of about 5 percentage points, although she has ranged between being virtually tied with Trump at her worst moments to being ahead of him by as much as 10 points at her best moments. In fact, the election has tended to ebb and flow between these boundaries like a sine wave:
If we had a model like this — basically, a version of polls-plus that partially reverted the current polls to the long-term mean of Clinton +5 — it would show her as about a 70 percent favorite instead of a 60 percent favorite. But would that be a good set of assumptions? Does the fact that Clinton has usually led Trump by a larger margin than she leads him now mean anything?
This is a complicated question, and one that we might want to revisit over the next couple of weeks. But the short answer is… I don’t know. We know that many news events — most notably, the political conventions — produce short-term “bounces” in the polls that partly or wholly reverse themselves after a few weeks. …
… So it’s plausible that Clinton’s “bad weekend” could be one of those events that has a relatively short-lived impact on the campaign. …
… I’d advise waiting a week or so to see whether Clinton’s current dip in the polls sticks as the news moves on from her “bad weekend” to other subjects. …
DSeid
September 16, 2016, 10:21pm
122
Seriously for him to not give Wang’s model a shout out as the one using that model consistently is pretty low class.
CarnalK
September 16, 2016, 10:49pm
123
I would agree but Wang has mentioned and linked a couple times to infer Silver is playing at clickbaiting. They are clearly getting into a little pissing match this election.
DSeid
September 16, 2016, 11:42pm
124
Wang is not the only one who has made that observation. It’s pretty widespread and has been discussed here much too. It’s a fair cop: Silver’s site needs clicks; it’s his living. Wang does not. Silver has to have damn thin skin to get pissy over that.
The critique he is explicitly responding to though was this one :
… To generate a long-term prediction (their polls-plus forecast), FiveThirtyEight establishes a prior based on other factors such as the economy. (No, dear reader, I do not want you to tell me what those factors are!) This method can work – Drew Linzer at Votamatic and Benjamin Lauderdale have used it to make a long-range forecast, and have analyzed their model rigorously.
In my view, FiveThirtyEight’s polls-only approach does not take full advantage of this year’s polling data. It is possible to generate a long-term prediction using polls only. Here at PEC, that’s what we do to create a prior expectation for where polls will drift to. Think of it as “polls-plus-more-polls.” Here is how we do it.
As I explained in May, we have lots of data on how far polls can move during an election year. Here are time series from 16 Presidential campaigns. The red traces shows the ±1 standard deviation interval for the Democratic-minus-Republican margin, relative to the final outcome. Campaigns fall between the red traces about two-thirds of the time (68% of the time, to be exact). … <snip a bunch of math> … I used m-bar to establish a prior. This achieves the same effect as the “-plus” in “polls-plus” – but it uses older polls instead of econometric factors. I use m-bar for all of 2016 to estimate where the race is centered. I then estimate the range +/-S around m-bar, over which the final result may fall.
In the final step, PEC uses that range as a Bayesian prior. We take the random-drift calculation (see our banner) and combine it with the prior to get an estimate of where the election is likely to end up. That generates the “Bayesian win probability” … Now, this leaves the problem of how to estimate S. S is closely related to the red trace for the standard deviation in the first graph. Before 2004, S was large – up to 6 percentage points. Since 2004, S has been smaller, less than 3 percentage points. I believe this to be a symptom of polarization in politics, in which people choose sides.
So…which S do we choose? If 2016 is like 2004-2012, then S is 2-3 percentage points. … In 2012, I estimated S as the standard deviation of the Meta-Margin. This year, I am currently using national polls (current average, Clinton +4.7%) and a value for S of 6 percent (a “de-polarized scenario”). Sometime in Septemnber, I will transition over to using the average and standard deviation of this year’s Meta-Margin. …
Really a very polite discussion about using economic fundamentals as the prior as 538 uses vs a sort of version of polls-plus that used the long-term mean of Clinton +5, or more precisely +4.7, as the prior, which what PEC does.
Silver comes off as insecure and immature honestly.