The midterm electorate in 2018 will undoubtedly be older and whiter than the electorate in 2016. That'll be bad for the Democrats. But it will also be more educated--and that's bad news for Republicans. That's because Trump's improbable victory in 2016 was built on the largest educational divide among whites we have ever seen (he did 34 points better among white noncollege voters than among white college voters). But those white noncollege voters are much less reliable voters in midterms than their college-educated counterparts, so the 2018 electorate will be skewed, perhaps heavily, away from Trump's favorite demographic.
And, as David Wasserman notes in his excellent article on 538 that delves into this issue:
The most recent found that whites with a college degree disapproved of Trump’s job performance 61 percent to 37 percent, with 51 percent disapproving — a remarkable level of intensity for a group that he carried just 10 months ago. By comparison, non-college whites approved of Trump 56 percent to 38 percent, with only 27 percent disapproving strongly.
This is an important point: higher turnout among the college-educated may help Democrats not just in highly-educated areas but in ones with a heavy noncollege presence as well.
None of this means Democrats will get the 24 seats they need in the House or pull off a miracle in the Senate. But it does mean that Republicans are not the only ones favored by midterm turnout patterns this time around. That should help level the playing field for the Democrats so they can take full advantage of the GOP's many and growing vulnerabilities.