[lmm]

It's quite easy to ignore even very clear evidence when it goes against one's politics. And that kind of result is massively politicised for obvious reasons (e.g. there are clear differences on standardized test scores - but some argue this reflects biases in the tests themselves).

[kirsebaer]

One bias is so-called "stereotype threat".

> negative stereotypes raise inhibiting doubts and high-pressure anxieties in a test-taker's mind... even passing reminders that someone belongs to one group or another, such as a group stereotyped as inferior in academics, can wreak havoc with test performance.

http://www.apa.org/research/action/stereotype.aspx

http://www.npr.org/2012/07/12/156664337/stereotype-threat-wh...

http://steinhardt.nyu.edu/news/2008/1/29/Stereotype_Threat_A...

[lmm]

Of course as with everything in this area, there is no consensus; studies have contradictory results and some dispute whether stereotype threat even exists.

[Tuisto]

I seriously doubt "stereotype threat" is going to survive much longer. http://www.sciencedirect.com/science/article/pii/S0022440514...