Data-mining and Negative Advertising

There has been a lot written about microtargeting and data-mining in elections to identify voter segments that will have a high response rate to targeted appeals. This means that 1) if apporached (through advertising or other GOTV methods), they will vote, and 2) will vote for your candidate. This has apparently been one of the primary tools of the Republican party in the last few elections, and while it has been used by Democratic candidates with success, doesn't yet seem to be part of the strategy of the Democratic party as a whole.

The basic idea is that by identifying just the right target group, even if that group is really small, a campaign can get massive bang for the the buck in terms of dollars spent on GOTV. This is why polls leading into elections may be wrong, because a successful microtargeting effort will cause a given identified group (let's say, female Honda owners who live in apartments and have voted in 50% of the last 4 elections) to come out to the polls in much larger numbers than would be predicted by their percentage of the population as a whole. Aside from the excellent response rate, it means that campaigns can spend less money on shotgun-style advertising--for example, broadcast television ads, which have a very low response rate since they reach people who might not be for your side, or might not be able to vote, or might not vote, or don't even live in your candidate's district.

This is a proven technique, but it is predicated on the idea of response rate where response is actually voting. What if you worked it where the desired response was for a likely voter for the other guy to stay home on election day? This is the fundamental premise of negative campaign ads. They're designed to keep potential supporters of the opponent at home.

But negative campaign ads reach too many people--they reach hardcore supporters of your opponent, who are unlikely to be swayed. They reach opposing party loyalists who may believe the ads but will still vote because the guy is in the right party. They reach your supporters, who already hate the opponent anyway, so their reactions are kind of meaningless. They reach people who weren't going to vote anyway. Etc. It's expensive to reach all those people, and a waste of money of the vast majority of those reached will not respond in a meaningful way.

So, could you create voter segments to whom you could target negative advertising or publicity? Let's say that female Honda owners who live in apartments and have voted in 50% of the last 4 elections are weak supporters of your opponent and only need a little push to keep them from voting (thus denying your opponent critical votes). How do you microtarget them? Cable TV ads? Direct mail? And has microtargeting for negative ads been done? I'd be curious to know.