I’ve written before about bad stats, but a video currently getting a lot of attention on the web is an excellent example of bad research. Which doll is the nice doll purports to show racial self-hatred in black children.

rnrn

The video is a version of Kenneth and Mamie Clark’s doll experiments. Anyone who has studied research methodology should immediately pick up some of the problems with this video:

rnrn

rn
• The same population size and selection. The MSNBC article covers 4 children, all black, with no evidence given of their background. The sample could easily be biased, and there is no control group.

rn

• Children are known to have a tendency to please authority figures, which is why interviews with children are problematic. Often a child will give the answer they believe will please the interviewer or audience. If a child perceives the interviewer as feeling racially inferior (if black) or superior (if white), or notices cues that the interviewer is expecting a particular answer (often too subtle for adults to notice) he/she will respond accordingly.

rn

• At least one of the children may have been displaying gender bias, perceiving the white doll as female and the black doll is male (the genders – if they were gendered – were not clear in the video). Most cultures display a gender bias that perceives women as more attractive than men.

rn

• According to anthropological research many societies showed a preference for light skinned women (in particular) even before contact with the West (see for example Fair Women, Dark Men). While this may be “colour prejudice” it is not a self-perception of racial inferiority.

rn

• The order of questions biases the research. The children are asked to indicate the “pretty” and “nice” dolls first; they may choose the white doll for any of the reasons above. When asked to indicate the “ugly” or “bad” dolls they are logically forced to indicate the black doll.

rn

rnrn

As it happens the Clarks’ original research was criticised as being unfalsifiable, but bad research like this just keeps popping up whenever someone wants to “prove” a point.

rnrn

Bad Research ultimately hurts everyone. In the short term it is effective at persuading most people (often to their detriment), but in the long term it hurts the credibility of those who use it. It+¢G‚¬G„¢s a lose-lose situation. Friends don+¢G‚¬G„¢t let friends do Bad Research.

rnrn

For my next trick I won’t be looking at climate change research …