Cross-posted at overdetermined.net
As promised earlier, I'm focusing on the way voter files are used during the runup to an election. Last week we covered early voting; this week, it's time to focus on new registrants. Follow me to the magical land of beneath the fold...
As I've discussed earlier, voter registration is spiky when graphed over time. For example, in Virginia, there were about 300,000 new registrants this year; of those, over 60,000 came during the last three weeks of registration, from mid-September to October 6. This probably understates the spikiness, since there was no doubt another surge just before the Potomac primary on February 12.
With this kind of surge, getting these people into the database becomes a paramount priority. Less and less information gets appended to the file, since it makes more sense to just get as many names as possible in to run field efforts.
For example, a friend of mine is working as a volunteer coordinator for Obama. In his turf, many eligible voters have been contacted many times, and target lists are growing thinner. The faster he gets more addresses of registered voters in, the better. Obviously, a campaign with the kind of energy and manpower that Obama has is a special case; but regardless of the situation, as election day approaches, speed becomes more and more attractive relative to completeness.
During most of the cycle, you will update a file relatively infrequently, and when you do you will process the entire file. But towards the end, it makes sense to do new registrant updates. You will simply receive the list of new registrants and tack it on to the end of the last full update, without dropping any voters from the full update. These files will be much smaller--five-digit numbers, rather than seven or eight--and will have less data added. Instead, the goal is volume, volume, volume.