It's no secret Republicans will do anything to put Hillary down -- even if it's to aide a rival. They're banking on Obama getting the nomination, so they can replay the Willie Horton game again. They careless if it hurts the Black vote, since they're not voting Republican anyhow -- and it'll be a "call in the wild" for White males everywhere to be reminded of what can happen to "race traitors", too.

In Nevada they ran that Spanish ad that said Hillary didn't respect the Hispanic community and I firmly believe the ad backfired and gave Hillary the lead. He completely lost me when I saw John Edwards denounce the ads for the race baiting smear they were and Obama said nothing on the subject.

Waste of time, I can't say.  It wasn't for me.  Seems simple enough:  it tests if you are more or less likely to spontaneously assign a positive or negative valence to any variety of stimuli.  It is extremely useful if a person is interested in knowing to what degree they may harbor unconscious biases based on sex, race, gender, body-type, etc.

It is quantitative in the sense that it arranges your choices ordinally; so, as with any collection of numbers, the ordinal data can be submitted to standard statistical tests across different test-taking groups.  

Here is the value:  Say I want to know if a random sample of, say, male freshmen Obama-supporters in a predominantly white Midwestern University have more or less negative feelings about career-minded women.  I could give a Likert-type survey, but then I would stuck with the answers that they wanted me to believe of them--or even the ones they thought they believed.  Or, I can give them an IAT.  In the latter case the data are much harder to fake, and are both quantifiable (63% of Obama-supporters in our sample had more negative feelings about women and careers, as compared to 38% of Clinton-supporters.)