| I can only guess that data like ages, income, # of children, etc can somehow be corelated to voting preferences. Maybe if a neighborhood is known to have lots of well-to-do two parent households with kids, it may lean one direction.
If I have your name and address from voter registration records, I can match you against the subscriber lists for Sports Illustrated, Guns and Ammo, The New Republic, National Review, Parenting, ad infinitum. By the time I finish with that I know a bunch of your hot buttons. Then I do the membership organizations: Sierra Club, the NRA, AARP, potentially hundreds of others. Now I know almost all of your hot buttons. I look at the home sales in your area, get the comparables, and now I know your income plus-or-minus a few percent. If I now encounter another voter that I can find nowhere in the magazine lists or the membership orgs, I will take his nine-digit zip and go look up the other people in that 9-digit zip. Odds are you all have comparable incomes, and if there is an obvious trend (40% of the people in the 9-digit zip belong to the the Sierra Club, and not one belongs to the NRA) I'll go ahead and make the bet that this guy is an environmentalist too. I might be wrong, but in the absence of any other data, it's a better bet than a wild guess. Most people in the campaign management business blow this stuff off as black magic. That's not because it doesn't work, it's because they don't understand it. So we are all safe from widespread use of this technology for another 15 years or so. The parties use it, the Vigeuries use it, but ask the campaign manager for Joe Candidate in Arizona about it, and his answer is "Huh?"
|
You're close, because that's about where the state of the art was five years ago. The point of this article is that they are using so much more now.