Ann McGeehan, Director of Elections in Texas, spoke at the conference this afternoon about their experiences with the development and implementation of their statewide voter registration system in Texas. During her presentation, she made three very interesting points:

1. Data matching: McGeehan gave some data on the matching rate they are finding with their new system. (Caveat: I jotted these down as quickly as I could; if there are any changes to these numbers I’ll repost.) She stated that somewhere between 50 and 60% of their registrations or re-registrations are coming from Texans in drivers’ license offices (“DPS” in Texas). For all of those registration requests, DPS verifies the drivers’ license, not the election officials. For the rest, 85% give a drivers’ license number; 6% give the last four digits of of their social security number; and the remainder have neither form of identification. When they verify the license number or social security number, they get a 2% match failure rate with the drivers’ licenses, but over 50% fail when they match the last four digits of the social security number.
2. Process when verification fails: McGeehan stated that Texas developed a procedure for when this happens, which was unforseen under HAVA. She noted that since the failure to match might be due to problems in the databases, they are treating those who try to register but can’t be matched as first-time registrants (meaning they have to show identification when they show up to vote). It would be interesting to see what the other states have developed as procedures for this eventuality.
3. Hurdles to implementation: McGeehan noted that they were having troubles with implementation of their new statewide file. She said that it was a hugely complex task, and that one of the greatest problems has come from “dirty data” from the counties.

Now to the presentations by Kim Brace and Thad Hall on data matching and interoperability.