I know this post is two weeks late and I am sorry about that. Still, better late than never.
This is a reply to Lorrie Goldstein’s column, “Stats, smoke and mirrors”, in which he rants about crime rates and Statistics Canada―the Canadian statistics agency. In my opinion, his diatribe is foolish, unfair and uninformed.
In an effort to be straightforward, this is what I will discuss:
- If decision makers don’t know how to use statistics, then they are the problem; don’t blame statistics for that.
- Mr. Goldstein’s “analysis” of crime rates is useless and superficial.
In his column, Mr. Goldstein complains that reporting a decrease in crime rates leads politicians to cut on policing and prison spending. I don’t know whether this true or not. But if it is, politicians are to blame because they should not be making these decisions based solely on crime rates. Statisticians create indicators so lay people like Mr. Goldstein can see trends and ask questions; they provide “red flags” when needed, and help monitor a situation. By no means, these figures are showing a whole picture.
In order to make rational decisions, any decision maker should look deeper into the data and find out more specific information. For instance, which kinds of crimes are responsible for most of the reduction? Are all types of felonies decreasing in frequency? Is there any particular offence that is occurring more often? Which demographic group is participating less in illegal activities? By answering this kind of questions, politicians can better understand the problem and choose where to cut spending, how to redistribute resources, or which kind of crimes to target.
This leads to my second point. Mr. Goldstein concludes that “crime has increased 321% since 1962.” So what? This statement alone informs nothing and it is completely useless. No rational decision can be made based on this “analysis”. Mr. Goldstein’s conclusion could have some weight if he had compared these figures with other indicators: has the gap between the rich and the poor increased as well? What about cultural diversity, unemployment, demographic density, drug addiction, or housing prices? Are these variables related to the raise in crime that you claim is true? Does any of these other indicators help us rethink the decline?
Had Mr. Goldstein done that, he would probably have realized that the Canada of 1962 is not directly comparable to the Canada of 2010. Had he known a little more about analytics, he would have realized that drawing conclusions from just one indicator is foolish, irresponsible, and sometimes, dangerous. Sadly, many business executives and politicians think like Mr. Goldstein.
To conclude, Mr. Goldstein’s belief that statistics are lies is largely fuelled by misconceptions about statistics and its role in decision making. Unfortunately, this kind of uninformed opinions had a lot to do with the absurd proposal to make Canada’s census voluntary.