Byron Tau has a good piece about how a massive tide of elderly Republican voting drove the election result:
Voters over 65 favored Republicans last week by a 21-point margin after flirting with Democrats in the 2006 midterm elections and favoring John McCain by a relatively narrow 8-point margin in 2008....
In the 2006 midterm campaign (regarded as a more useful basis of comparison than high-turnout presidential elections), voters 65 and older essentially split their vote evenly between Republicans and Democrats — a stark comparison with 2010. But that’s not the only noticeable change in voting patterns: In this cycle, in addition to losing the senior vote by more than 20 points, older voters also grew as a share of the overall electorate. In 2006, seniors made up 19 percent of the voting public; this year, they represented 23 percent.
We've gotten used to this, but the rise of the young as a strongly Democratic constituency, and the old and a strongly Republican one, is a novel phenomenon. In 1994, voters under 30 and voters over 60 both split their votes 51-49 for the GOP. You can check the demographics of the House vote going back to 1982 here. Basically, there was little difference between the youth vote and the elederly vote for most of this period. A split opened up in 1996, with young voters going Democratic by 55-45 and old voters going Republican 51-49, but the split closed in half the next election and disappeared altogether by 2000.
In 2004, the young-old split began to reappear, with young voters going Democratic by 12 points, and the old going Republican by 8. It's continued to widen, to the point where the 2008 and 2010 electorates had a 16- or 17-point gap between the young and the old.
One effect of this change is to make Democrats increasingly vulnerable in off-year elections, where the elderly tend to turn out and the young do not. It's possible that we'll see several cycles in which Democrats make gains during presidential years and lose them during off years, at least relative to trends. (Which is to say, sometimes Republicans will do well in presidential election years, and sometimes Democrats will do well in off-years, but the age split will diminish those effects.) And, of course, in the very long run this split bodes well for Democrats, but the standard John Maynard Keynes caveat applies.