To measure how close an election was, I believe the best way is to look at how many votes a loser needed to have won in order to change the outcome. The closest elections in my lifetime were 2000, 1976, 1960, and 1968. (Data from this article.)So how close was the election this time? The NY Times reports:
Gore could have won in 2000 with about 500 more votes in Florida.
Ford would have won in 1976 with about 18k more votes in Ohio and Hawaii.
Nixon would have won in 1960 with about 60k more votes in Illinois and Texas.
Humphrey would have won in 1968 with about 106k more votes in New Jersey, Missouri, and New Hampshire, assuming Democratic control of the House.
As of Wednesday, Mr. Trump’s lead in Michigan had shrunk to 10,704 votes, or 0.2 percent, according to the National Popular Vote Tracker maintained by the Cook Political Report.So by this measure, Trump won by about 100k votes. This was about the same as Nixon's margin in 1968, and about a third of Obama's margin in 2012.
Mr. Trump’s lead in Wisconsin has narrowed to 22,525 votes, or 0.8 percent. In Pennsylvania, his lead slightly grew on Wednesday, to 70,010, or 1.2 percent.
Update: This NY Times story on Election Facts says that the margin was 12882 (MI) + 24081 (WI) + 65690 (PA) = 102653, with Trump'a margin in Florida being +112,911, and a few votes still being counted.
But the story is misleading by saying:
Hillary Clinton definitely won the popular vote, and that lead is only going to grow. ...No, Clinton did not win the popular vote, but only got a plurality of the popular votes.
Yes, the polls were off, but not in extraordinary ways.
The AP poll reported that Clinton led by 17 percentage points. The Princeton Election Consortium said that she had a greater than 99 percent chance of winning. She barely campaigned in Wisconsin, Michigan, and Pennsylvania because pollsters had assured her that she had huge leads there. Many experts complained that Nate Silver was giving too much encouragement to the Trumpsters, but even he gave Clinton a 95% chance at one point.
The NY Times says that these errors are balanced by the fact that Clinton did better than expected in some blue states. Maybe so, but the election was played out in the battleground states, and most of the polls were very badly wrong there. (A couple of polls, like IBD, did well.)
Update: This says that the margin is 80k, as of Dec. 2.