I posted new turnout rates based on official canvasses available on the Secretary of State web sites, available at http://elections.gmu.edu/VAP_VEP.htm.
. These turnout rates will likely be the final rates, since the two states that have not certified their results are simply awaiting legislative approval early next year. I have calculated a VEP turnout rate of 39.9% in the 2002 election (using an estimate of total ballots cast). Minnesota had the highest VEP turnout rate of 63.4% and Virginia came in last at 28.6%. I thought I would also make everyone aware that I have had conversations this week with the people who put together population estimates at the Census Bureau. They have indicated to me that they may not produce even 2004 VAP estimates because of budgetary constraints. In fact, we may not get new age-sex-race population projections for some time, which could pose a problem for people who need these numbers to weight surveys. They urged interested people to start complaining to the powers that be so that they will be able to make this a priority.
These numbers will not be released because the Census Bureau is still struggling to figure out how the 2000 estimate (based on the 1990 census) predicted 174 million persons, while the actual census counted 181 million. In a related matter, the Census Bureau is unable to correct the 1990-2000 estimates for the 2000 census. The reason is only partially related to the previous prediction problem - also factored in is an OMB directive which prevents the Census from reconciling the 2000 race categories with the 1990 race categories, which is necessary to make age-sex-race population projections. So, we may never get updated 1990-2000 population estimates even if the Census can figure out the prediction problem. This could affect any study that uses these numbers - turnout rates, cancer rates, accident rates, etc. Out of that list, I think turnout is the least important, and makes me concerned that this may pose real health risks.
There is further funny business going on at the Census Bureau. As some of you may be aware, the Bush Administration has agreed to release the adjusted census numbers as the result of a lawsuit brought by Oregon Democratic legislators. The Census Bureau has released the original adjustment to Oregon, but the Census Bureau is now claiming that the original undercount projections were wrong, and that in fact, there was essentially no undercount in 2000 (a revised 0.16% undercount from a 1.2% undercount). The undercount problem is related to the prediction problem, since the Census Bureau has figured out that some of the prediction problem can go away if they can wish away the undercount (never mind the political ramifications since minorities are more likely to be undercounted). I gathered in my conversations that these new adjusted data are available for purchase, though they are not advertised on the Census Bureau web pages, and I have not yet investigated purchasing them. Have a Happy Holiday! -Mike ================================== Dr. Michael P. McDonald Assistant Professor Dept of Public and International Affairs George Mason University 4400 University Drive - 3F4 Fairfax, VA 22030-4444 Office: 703-993-4191 Fax: 703-993-1399 Efax: 561-431-3190 mmcdon at gmu.edu http://elections.gmu.edu/