Thursday, September 29, 2005

How to interpret the EDS / EAC Report

I've received a few emails regarding the EAC Election Day Survey report prepared by Election Data Services.

In particular, my comments regarding race and absentee ballot usage seem to be contradicted by Chapter 5 of the report.

I'm happy to be proved wrong, but it's very important to understand how EDS is reporting their results. An individual "case" in the report is not the individual, but the "jurisdiction".

This means that you cannot assume anything about individual behavior, and more importantly you cannot derive national averages from these data without weighting by jurisdiction size.

Thus, this paragraph (from Chapter 5 of the report) ...

Predominantly Hispanic jurisdictions reported the highest request rate for absentee ballots, 13.6 percent, slightly more than twice the lowest reported rate in predominantly non-Hispanic Black jurisdictions, at 5.7 percent. Predominantly non-Hispanic White jurisdictions reported a rate, 10.9 percent, slightly lower than predominantly Hispanic jurisdictions. Predominantly non-Hispanic Native American jurisdictions reported a rate, 6.1 percent, slightly higher than predominantly non-Hispanic Black jurisdictions.

... tells us very little about the relative rates of absentee ballot usage among whites, blacks, and hispanics.

It says what is says. For instance, among "predominantly Hispanic" jurisdictions, the average rate of absentee ballot request was 13.6 percent.

Those jurisdictions may vary in population size from densely urbanized counties in Arizona, New Mexico, and California to rural counties in Southwest Texas, northern New Mexico, southern Colorado, and northern Nevada.

Minimally, if we want to make a rough estimate of national averages, we need to weight the jurisdictions by the population in each county.


Post a Comment

<< Home