Pennsylvania benchmark + writeups released. 55% Clinton 45% Sanders
http://www.benchmarkpolitics.com/2016/04/pennsylvania-final-county-benchmarks.html?m=1
"
Throughout this primary season and beyond, I will be posting models like this that allow you to follow along with on primary night and not act confused. Watch with purpose. That is the point of all of this, to be able to say with data and analytics to back it up that one candidate or another is doing good or bad, using hard metrics. The use of the model is in knowing whether areas yet to come favor your candidate or disfavor them. You can make predictions on the final margin of victory early, with only 10-15% reporting.
I want to make sure that everyone is clear, this baseline model is becoming predictive, but its true value is not in its state-wide predictions. Polling is not yet adequate enough to get reliable county-by-county predictions that will be accurate to any degree as much as the state polls are. That being said, the demographics can be extrapolated, and the counties and precincts can be modeled. What this does is try to say how a candidate is EXPECTED to do, given the demographics, crosstabs, and previous results this year as well as past contests. If a candidate consistently is beating their baselines, you can bet they will beat their model baseline as well.
All race dot maps are courtesy of Weldon Cooper Center for Public Service, Rector and Visitors of the University of Virginia (Dustin A. Cable, creator).
http://www.benchmarkpolitics.com/2016/04/pennsylvania-final-county-benchmarks.html?m=1