In this study, the researchers attempted to fight casual biases in based on casual stereotypes and prejudices. This is not about blatant racism; this is about subtle, subconscious bias.
This study came about because some groups of people have been proven to receive inferior treatment because of their race, when all other factors were equal. Not surprisingly, doctors tended to treat blacks with less serious attention. Similarly, it's been found that in a white room, with the same resume, someone an employer believes is black has a significantly less chance of being interviewed and hired.
In this study, a number of researchers attempted to find methods that could help to fight these prejudices and get people to treat people identically despite these casual, subconscious biases.
A number of methods did not work. This is not surprising. One of the most effective methods was to present subjects with "counter stereotypical" imagery. For example, showing the subjects images of black people in very positive roles was an effective behavior changer. That said, even its effects were somewhat temporary.
This suggests that if we want to eliminate casual prejudiced and biased behaviors, including discrimination in hiring and medical practices, people need to consistently be presented with imagery that bucks hurtful stereotypes, and shows marginalized groups of people positively.
The next time you say that constantly sexualizing, diminishing, or brutalizing women in the media doesn't hurt anybody, think about this.
A side effect of the study will note that the relationship between participants that actively admitted biased views and those that reflected biased views in their practices was very little. Almost nobody admitted to racial bias, but it was surprisingly common. So the next time you say, "I'm not prejudices toward X", consider that maybe, just maybe, you could be, but don't notice it. Maybe you're no more enlightened than the more than 11,000 participants in this study.http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2155175
(While an academic paper, there's a link to download it, free of charge.)