The pathetic state of "win forcasts" in US politics -check out this graph
Edit: Changed from polls to win forecasts to be more clear, although the former is used to generate the latter.
This isn't a pro-sanders post. This is a "wow, the polling system is broken, misleading, and outdated" post.
This is a screenshot from five-thirty-eight (argued to be the "most accurate"
On March 29th according to them, Clinton had a 85% chance of winning to Sander's 15%.
On April 1st, Sanders now has a 58% chance of winning.
The problem is that such polls are used politically and in news constantly. There is a sense that they are "mostly accurate except for the margin of error that they provide." The margin of error seems to imply "Ok, so its not exactly correct but its close." The problem with margin of error is that its a mathematical calculation that assumes the data is correct. The real margin of error in these polls is the methodoogy and/or datasets.
In this case, the margin of error of the polls was around 4 points. The margin of error in the data and methodology was 43 points. I'm oversimplifying here at bit (There wasn't enough data, but that didn't prevent them from making graphs and talking about it as if there was enough data to assume certainty).
The point is this: Have we come to a point were polling numbers are entirely arbitrary media tolls that accidentally (or intentionally?) do a lot of harm to democracy? I realize such claims have been made countless times for many years, but I have never seen anything like that chart until today.
= new reply since forum marked as read
...this is a probability estimate model, taking into account polling, turnout, endorsements and other factors.
As for the accuracy of polling?
Are Democratic primary polls accurate? In individual states they are good, with a few exceptions. At an aggregated level, they are remarkably accurate. The delegate-weighted polling margin has a total error of 3.1%. This is better accuracy than one would expect from the reputation of polling these days. The overall error is also not nearly enough to change my calculation from yesterday.
Above are Election Day results so far this season, plotted against poll medians taken shortly before the election. The correlation coefficient is +0.85, pretty good. Red symbols indicate caucus states.
The notable exceptions are Minnesota, Utah, and Idaho. These are all caucus states. However, they also have a high fraction of non-Hispanic whites (see Will Jordan’s graphs showing this here and here). Perhaps the anomalies are accounted for by some aspect of the combination: highly committed Sanders supporters and deliberating in a crammed room. As an aside, it seems to me that the caucus mechanism requires either cultural homogeneity or a willingness to cross cultural lines to gather in groups. If caucuses date from a previous time, is segregation the historical antecedent that gave rise to a caucusing/race correlation?
The most important quantity for prediction is the accuracy of all the polls aggregated. The delegate-weighted sum of polls is Clinton +17.9%. The delegate-weighted voting result is Clinton +14.8%. The difference, 3 percentage points in Sanders’s direction, is greater than the ~2% that one would expect from sampling error alone. However, it does not come anywhere close to the 22-point swing that I calculated is necessary for Sanders to get an overall majority of pledged delegates by June 7th.
http://election.princeton.edu/2016/03/28/the-effectiveness-of-democratic-primary-polls-for-prediction/
