I thought the Humphrey Institute poll was interesting, so I went back tonight to read its marginals. Unlike my earlier analysis of the SCSU Survey, the demographics here looked fine. They still have the party ID result that says 12% more DFL than Republican. I didn't see too much else to question except for this question. All of the voters in the sample are considered likely voters, meaning they indicated they "will vote" or will "probably vote". The next question is

Are you currently registered to vote in Minnesota? Yes......................................................................................................98.6% No..........................................................................................................0.6%
Plan to register (vol)...........................................................................0.7%
Don't Know...........................................................................................0.1% Refused.................................................................................................0.0%

Now this poll is run by Larry Jacobs, who co-authored in October a report on voter turnout in Minnesota. He noted in that report,

Same-day registration has contributed to Minnesota�s strong voter participation, accounting for 15 percent to nearly 21 percent of the state�s turnout. (Minnesota is one of just six states -- including Wisconsin -- that permits same-day registration. Most of these states lead the country in voter turnout.)

It was 15% in 1998 and 2002. Now, some of the people who are considered registered to vote in the Humphrey sample will be in the same-day registration because they've moved (if they reported themselves as registered but not registered in the right district), so they haven't missed 15% of the sample. But they've certainly missed some. Do we think that is random, or are they undersampling a specific demographic group? How about young professionals moving to the suburbs? If they recently moved, they might not be in the sample. Who are they? And do they vote more D or R?

How big a problem is this? I really don't know. I just haven't seen anyone talk about the effect of same-day registration on polling here before. Ideas? Please put them in comments.