Andrew Gelman blogs about whether voters and nonvoters are different.
Baldassare finds a lot of differences in California. Yet Gelman points out that "political scientists generally hold that voters and nonvoters aren't really so different", and refers to a blog posting with some nice links by Sides.
I did a quick skim through Sides material, and it's consistent with the excerpt Gelman quotes from Highton: "How can one estimate what nonvoters would do if they were to vote? . . . we attributed to nonvoters the perspectives of voters in their respective [demographic group]"
This is creative, but leaves us with the assumption that within a demographic group the opinions of voters and nonvoters are in aggregate the same. That's not necessarily true, since voting is a form of involvement.
It's very common in consumer packaged goods research for people who are willing to provide information about their purchases (i.e. cooperate with detailed surveys) to be far higher purchasers of new products than the average household -- even matching on demographics.
This is because the people who are cooperating are people who are more interested in packaged goods products than the average person, and this involvement causes them to pay more attention to new products. This means that the relative sales of new versus established products is higher (directly analogous to a political voting choice).
It's no great leap of faith to theorize that -- controlling for demographics -- people who vote are more involved in the process are nonvoters and may have different / more informed opinions.
But there's likely to be a lot more literature out there that what I've skimmed quickly.