How to lie with statistics (or not)
As the old saying goes: there are lies, damn lies, and then there are statistics. The Los Angeles Times published a story this week that the LAPD had rejected a study of racial profiling conducted by a Yale University legal scholar and endorsed by the ACLU. The reason cited for the rejection was that the study did not examine individual circumstances, and instead looked at more macro level statistics, such as finding that African and Hispanic Americans were 29% and 32% respectively more likely to be arrested after being stopped by the LAPD.
It is interesting that whenever there are complaints from individuals about racial discrimination or racial profiling, the standard response is that these are isolated incidents and we shouldn’t impune the integrity of an institution without broad statistical evidence. Years ago, when California voters passed Proposition 209 to prohibit affirmative action, its supporters successfully forced the legislature to remove laws on the books that required statistical reporting of racial statistics in state employment and contracting. I assume that the same thing has happened for the dozen or so other states that have passed similar ballot measures.
Trying to pursue an individual discrimination claim without statistical evidence is like trying to stop Americans from driving their cars. And passing laws making it more difficult to gather statistics to determine whether there is a discrimination problem is the equivalent of locking the door and throwing away the key.
I do agree that discrimination studies should supplement macro level statistical data with anecdotal data, however the jurisdictions need to ensure that the necessary data is being gathered and reported in order to get in front of a problem instead of having to react to it. And in the meantime I guess we need to just take LAPD’s word for it that they don’t have a racial profiling problem.
More information on racial profiling.
