Curtis Gans of American University's Center for the Study of the American Electorate has just published a new report that analyzes nationwide registration data. Based upon the 34 states that have reported registration statistics, he finds that 68% of the voting age public is eligible to vote. This is unchanged since 2002.

Gans also offers some surprising information on partisan registration. He has analyzed the 13 states that have supplied partisan registration voting data, comparing them to prior years. Relative to 2002, the Republicans have actually closed the registration gap. In 2002, the Democrats had a 7.0% registration advantage over Republicans in these 13 states. This year, their advantage is down to 5.8%. I won't report the actual figures because they are inflated (due to deaths and geographical movement), but the trend lines here are, as Gans argues, valid (so long, of course, as Democrats are no more likely to have died or moved than Republicans).

Unfortunately, Gans only offers data for midterm elections -- so we cannot determine whether, in 2004, the Republicans greatly increased their share of registered voters in 2004 and the Democrats have reduced that increase this year. Also, registration seems to square only loosely with final vote outcomes. Between 1978 and 1982, Republican registration increased by 0.8%, but they won fewer seats in the latter year. Between 1990 and 1994, it also increased by 0.8%. (Though I would note with interest that the last time the GOP lost as many seats as Cook and Rothenberg are predicting they will lose, their share of registered voters fell by 2.5%.) And, of course, being registered to vote and actually voting are two entirely different things. So, registration itself does not really give us much purchase on the actual voting public. Also, his report does not -- at least by my read of it -- mention which states are in play. This limits the scope of our analysis -- especially if we think that regional or other state-by-state variables might affect turnout.

Nevertheless, this data offers some interesting qualifications on the "disspirited" storyline that the press has embraced. If Republicans are disspirited, they do not seem to be expressing it by altering their long-term registration habits in these 13 states. The trend line of the last 44 years has not been interrupted in any kind of significant way due to Republican morosity. In the last 44 years, the Republican percentage of registered voters has grown, on average, by 0.47% per midterm year. In the last 28 years, it has grown by 1.4% per midterm year. This year's growth of 0.8% is actually identical to the 1998 - 2002 growth. The Democratic decline, meanwhile, has averaged 0.96% per midterm year in the last 44 years. In the last 28 years, they have declined by an average of 0.55% per year. Between 1998 and 2002, they declined by 1.2% -- though this might have been a "correction" for their 2.4% increase in registrants between 1994 and 1996 (their largest in the 44 year period). This year, their rate of decline is 0.4%.