Presidential election, 2016

[pbrower2a]

(10-11-2016, 09:38 AM)Bob Butler 54 Wrote: I'm not as much into chasing polls as many contributors to this thread.  Rather than building my own maps, I've been visiting 538's who will win the presidency page.  At this point I'm pretty happy with what I'm seeing.

The page features a list of 14 swing states that were and are considered up in the air.  Right now only two -- Arizona and Georgia -- are leaning for Trump, and Arizona is getting mighty close.

Is 538 s as good as anyone if one wants to keep an eye on the race without doing too much work?

Nate Silver has a model that worked well in 2008 and 2012. Yes, his predictions jump around much, but they jump to reflect the political reality. Trump has gone from enticing to appalling for many people.  He has appalled me throughout this year, but I can't judge that as a model for the rest of America.

2012 was easy once four states seemed likely to decide everything. Barack Obama got solid leads in states with about 260 electoral votes, with Colorado, Florida, Nevada, Ohio, and Virginia showing about even chances for both. Colorado and Nevada were going to vote together, but neither this pair nor Florida, Ohio, nor Virginia was similar to the other state or pair of states. Any one of those could clinch the election for Obama, and they were far enough and dissimilar enough that they couldn't fall for the same appeal tailor-made for one but not the others. A special appeal that could win all of them would pull several states from reasonably-sure states for Obama to reasonably-sure states for Romney. Romney had to win all four and Obama needed win either Colorado and Nevada or one of the other three states.

One can ignore a state like Indiana (he wasn't going to win it without also winning Ohio), Arizona (he wasn't going to win Arizona without also winning Colorado and Nevada), or North Carolina (which he wasn;t going to win without also winning Virginia).

I called those states "independent events". Multiply the chances for Romney in each state, subtract that from "1" and you get the Obama chance. If the pair and the three others have a 50-50 chance of going either way, then the chance of Obama being elected would be .50*.50*.50*.50 =  1-.0625, or .9375. So what happens when the chance becomes 30% in one and 70% in the other? 

I could ignore such a state as Indiana (Obama wasn't going to win it without also winning Ohio, which would clinch) or North Carolina (which Obama was not going to win without also winning Virginia).  the Romney chance goes to 5.25%. Having one state become a 90% chance for Obama whbile another become a 90% chance for Romney is even worse, cutting the Romney chance down to  2.25%. One becomes a sure thing for Obama and the other becomes a sure thing for Romney? 0% chance for a Romney win.

The model was simplistic enough, and it seemed to fit reasonably well. Reality forced adaptations that caused me to state that Obama had practically won it by taking out the random chance of a Romney victory. Romney was in deep trouble in his Presidential bid.