Some friends and family members know that estimating the electoral college is a hobby of mine.  In 1996, I correctly picked 49 states.  Or stated another way, I correctly picked all 50 and 200K voters in CO got it wrong ;). This was before the era of Nate Silver, but in fact anyone spending an hour looking at state polls could have gotten 45 states correct.  So my 49 in 1996 hardly makes me an expert.

First point, the electoral college only matters if the popular vote spread is less than 3 percent. A candidate winning the popular vote by 3 percent or more will surely win the electoral college, probably with a couple states to spare.

But in a close election it does matter. Although I don’t think the election will be decided by less than 3 percentage points, if it were, the key states are FL, OH, and PA.  The New York Times had an extensive article on this just as I was doing my own thinking last weekend.  If Clinton can win any one of these three states, she will almost certainly win the electoral college. However if Trump can sweep all three, he is likely to win the electoral college.  This is because sweeping all three probably means winning Iowa and New Hampshire as well.  With those additional wins, the only path I can see for Clinton would be to win Arizona or North Carolina. Again, this is unlikely if Trump has won the other states discussed.   Bottom line, Trump needs that clean sweep of all three to win and PA has not gone Republican since 1988 (though it has been close in several elections).

With the latest polling numbers now giving Clinton an average lead of about 8 percentage points, talk of the electoral college key states seems almost silly.  If Clinton wins by 8, her electoral college map would look better than Obama’s in 2008.  For a sneak peak at the potentially bluest map in about 20 years, go to Nate Silver’s model and set the forecast on “nowcast.” This maximizes the impact of the latest polling.  This morning the nowcast shows Clinton winning all the usual swing states plus GA and AZ.  What a pretty map!

While glimpsing that map is fun, I would urge readers not to get too caught up in day-to-day polling changes.  Response rates for polls are very low.  What a poll is really testing is the propensity for respondents to pick up the phone and take the time to share their views.  With the Dem convention featuring so many good speeches favoring Clinton and Trump idiotically confirming everyone’s worst impressions by attacking a gold star family, is it any wonder that Dems were happy to answer pollsters with better number for Hillary? And that Republicans were less likely to answer those phone calls if they were in the random survey?  Larry Sabato of UVA puts it well:

You might ask: What about the wild swings in polling we observe with regularity, most recently after each convention? Some persuasive research has argued that it is explained by variability in the survey response rates of Democrats and Republicans. (See Andrew Gelman, et al., “The Mythical Swing Voter”). That is, short-term swings in candidate preference are caused mainly if not exclusively by variability in partisan response rates. Even small changes in response rates among Democrats and Republicans can produce sizable shifts in candidate support, given the very low overall response rates in most polls.

Clinton is totally dominating the news cycles right now.  This may not always be the case, so I would not be surprised if her average lead drops back to 3 or 4 percent, or possibly even closer for a few days between now and November.  I always go back to 2012 election won by Obama by 4 percent.  Demographics alone suggest if folks voted the same way in 2016, the lead would expand to at least 5 percent.  Given Trump’s now obvious hair trigger temper and obvious unfitness for office, how many former Obama voters can really switch to Trump no matter how badly folks feel about Hillary and her emails?  I just don’t see Trump assembling a winning coalition unless more terrorism causes the electorate to conclude we must have change if we don’t trust the change agent.  Not impossible, but very unlikely.

On the flip side, Trump’s current polling and the fact that local Republican primary campaigns are coming toward the end make it more possible that Republican congressional and senate candidates may soon formally abandon Trump, rather than just “distance” themselves.  If this happens, Trump’s numbers will likely stay at their current terrible levels or get even worse, making a double digit Clinton win possible and maybe even likely.

Bottom line, I expect a Clinton win in the range of Obama’s numbers in 2012 or 2008, with 2008 being the better estimate.  A double digit Clinton win is much more likely than a super close race that Trump manages two win.