It warms my heart to see occasions on which political science can be useful to our national discourse. Immediately thereafter it saddens me to see how rarely the media and political class take advantage of it.

Yesterday's New York Times ran a board editorial on the Justice Department's decision, 17 years after the National Voter Registration Act was passed, to require states to enforce the law. It did not take long for states to get on board with the "Motor Voter" portion of the law that combined driver services like license renewal with voter registration. However, some (Republican-led) states were much less enthusiastic about the other requirements to register people who obtain services like food stamps, welfare, Medicaid, disability, or S-Chip. Classy. And subtle.

The assumption underlying both the Republican stonewalling and the NYT's enthusiasm for the "welfare voter" portion of the law is that registration leads to voting – specifically that registering poor people leads to Democratic voting. But is there any evidence that this happens? Registration, convenience voting, and turnout happen to be my corner of the world, so…lucky you.

It has long been recognized that voter registration is costly in the U.S., contributing to turnout comparatively lower than other Western democracies. We are one of the only countries in the world in which the burden of maintaining a valid registration falls on the individual and not the state. As a result, the characteristics that we assume are predictors of turnout – income, education, partisanship, etc. – are better predictors of registration. Once registered, it was long assumed that one was highly likely (on the order of 80% or more) to turn out. The moral of this strain of research (see Erikson 1981 or Highton 1997 for an overview) was that if we could somehow get everyone registered we would have exceptional turnout. Get a person over the hurdle of registration, we assumed, and there was nothing to stop him from voting.

The problem is that the things that make people register, namely interest in and attention to politics combined with the skills necessary to register properly, are the same things that make someone likely to turn out. So not all registered voters are alike. Some were motivated enough to register on their own and thus are probably motivated to vote too. But if someone else registers you, does that matter? If you lack the skills/desire to register you probably lack the same with respect to voting.

In short, efforts to increase registration have rapidly diminishing returns. More people of lower socioeconomic status will vote because of the changes described in the editorial, but the turnout rate among that group will continue to lag the wealthier and better educated. Each additional trick used to wring a few more registrations out of the population of eligible voters is reaching successively less motivated or less educated people. So it depends on how one characterizes Americans who are using the social safety net – do they not vote because they don't know how to get registered and find the polls or do they abstain because they just don't care?

Nearly every effort to make registration easier and voting more convenient has primarily served to increase the convenience of people who would have voted anyway rather than to bring new voters into the fold and increase turnout (see Oregon's disappointing experiment with all-mail balloting). Based on the available evidence it is highly doubtful that, as the Times states,

When advocacy groups sued Ohio and Missouri to force their public assistance offices into complying, huge groups of new voters surged onto the rolls — more than 100,000 in Ohio, and more than 200,000 in Missouri. Nationwide enforcement by the Justice Department could add millions more. The more people who have access to the ballot, the better the country will be.

"Access" as a hypothetical will increase. Will those who enjoy this new access use it independently of the unprecedented GOTV drives responsible for the noted turnout surges in Ohio and Missouri? I'll believe it when I see it. Like other groups that don't turn out well historically – college kids, for example – there is more to getting the poor interested in voting than simply registering them.