It's out--the definitive take on the 2016 election! And it's pretty different than what the exit polls told you. My colleagues and I at CAP have synthesized all the available public data and harmonized these data with actual election returns down to county level. The result is a set of internally consistent estimates of turnout and support by group in both the 2012 and 2016 elections, what changed between 2012 and 2016 and how important each of these changes were to the outcome in 2016. In short, the real story.
Here's what we found:
- There were far more white noncollege voters in the 2016 election than shown by the exit polls. Our estimates indicate that 2016 voters were 45 percent white noncollege and just 29 percent white college-educated. The exit polls claimed there were more white college voters (37 percent) than white noncollege voters (34 percent). This is not just wrong, but massively wrong, especially in the context of Rustbelt swing states.
- White college-educated voters voted Democratic in both the 2012 and 2016 elections. In fact, Clinton carried white college voters by a solid 7 points in 2016. The exit polls said both Obama and Clinton lost the white college vote. Again, wrong. The new data indicate that white college graduates are edging toward being a Democratic demographic.
- There was a smaller margin shift among white noncollege voters toward the GOP in 2016 than people suppose. Our data indicate that Trump had a 31 point margin among white noncollege voters, representing a 6 point shift in his favor, not the 37 point advantage and 11 points shift the exit polls showed. This 6 point shift was actually smaller than Trump's margin gain among black voters (8 points).
- The decline in black support margin was just as important to Clinton's fate as the decline in black turnout. She would have carried Pennsylvania, even with a decline in black turnout, if she had held Obama's black support from 2012. She would have carried Wisconsin, even with a decline in black support, if she had held Obama's 2012 black turnout (2016 black turnout was down an astonishing 19 points in the state).
- Our simulations show that Clinton would have won the 2016 election if she had held black support and turnout from 2012. She would have carried, albeit narrowly, the states of Michigan, Wisconsin, Pennsylvania, Florida and North Carolina.
- Our simulations also indicate that she would have won the 2016 election if she had held Obama's modest support among white noncollege voters from 2012. She would have carried, with more robust margins, the states of Michigan, Wisconsin, Pennsylvania, Florida and Iowa (and just missed carrying Ohio). In fact, if she could merely have reduced the shift toward Trump among these voters by one quarter, she would have won the election.
- In Arizona, Georgia and Texas, where Clinton improved over Obama's performance in 2012, she did better among both white college and white noncollege voters. The key factor in all three states was relatively large shifts in her favor among white college graduates.
There is more--much more--in the report. I urge you to take a look at it.