Correlation, causation, facts and racism 👮🏾‍♀️

Don’t ignore the facts, the media and public service messages remind us. Science matters says the bumper stickers. Statistical analysis is a good thing when you have a truly representative data set, when there are no individuals but instead are things that are entirely alike, such as machines churned out of a factory.

Most people agree that the criminal justice system should not treat blacks different than whites. That people shouldn’t be judged based on skin color or denied housing or a job because they are black. They note skin color isn’t something you can change. But what about science and statistics that say elsewise? It’s a very proactive but morally fraught question. That say that an African American is more likely to have drugs or be engaged in criminal activity? It doesn’t take a lot of analysis of arrest or jail records to find that there is a nexus between race and involvement in the justice system.

However, that is a bad analysis for a couple of reasons. For one, if police officers are applying this analysis in their everyday business – racial profiling – then they are biasing the sample. Stop more black motorists and search their cars, of course you are going to find more drugs and crime from black motorists. Put more cops in low income neighborhoods, then you are going to have more people arrested for crime there, as cops are looking for crime there.

Then you have the issue of individuals being unique. Statistics are just that – averages or median values, what exists in the middle but not in reality. Just because a person’s skin color is dark, doesn’t mean they are automatically a criminal. Likewise, just because a person lives in a low income neighborhood doesn’t mean they are committing crime.

Everybody accepts that race isn’t a good tool to judge people by as it’s not changeable. But other characteristics, while potentially changeable – occupation, income, housing location, dress, affiliation with groups or organizations – aren’t very good either as looking at statistics doesn’t necessarily describe an individual.

Observing statistics is a powerful tool. Combined with maps, you can really get an idea of problematic areas via means and medians. You can find inliers and outliers, help you evaluate risks. The problem with using statistics isn’t that they don’t provide good information – they do – but they don’t adjust for bias caused by the observer or the uniqueness of individuals.

Ultimately the problem with profiling and pigeon holing individuals is its not fair or just. Individuals are automous, they are not solely defined by one characteristic or attribute. Racial or any other type of profiling might be effective policing – in a statistical sense – but it’s not fair to individuals its applied to. Liberty comes at a cost, and sometimes that involves ignoring statistical evidence, instead choosing to respect human freedom and dignity over stopping and prosecuting more crime.