« Regular Wine Consumption Cuts Down on Dementia in Women, Increases Fun Factor | Home | A Communications Disservice »

What Statistics Can Tell You and What They Can’t

By crooky | April 16, 2008

I mentioned in a posting last week that I recently wrote a test for a job with the RCMP as a Criminal Intelligence Analyst. Part of the test was an essay question that asked something along the lines of “statistics are all well and good but why is it a good idea to go out and gather qualitative data?” My answer was something along the lines of the following:

During my grad school studies during my Advanced Research Methodologies course, I decided to take a run at some data in the FBI Murder Database. (Fascinating database if you haven’t had the opportunity to muck around in it.) While looking at some of the categories of data that were available for analysis, I noticed that “infanticide” was listed as a subset of murder.

Delicious Murder

If you’re not familiar with the term - infanticide is when someone kills a child that is less than 12 months old. If you don’t have kids (I do), you need to understand that this is a truly heinous crime because no matter how gifted your infant is, there isn’t a hope in hell that they could defend themselves against malicious intent. Infanticide is the moral equivalent of raping and then killing someone in a vegetative state.

I was looking at the issue of infanticide from the perspective of a policy wonk and there are few more prickly pear of policy problems (P4s) out there. (And no, Gordon Campbell, I’m not trying to say that P3 +1 = Infanticide. Everyone knows that P3 + 1 = Armageddon).

When you look at a behavioural trend like infanticide from the numbers, you automatically try to find a correlation between the incidence of infanticide and other factors. In this case, I saw some correlation between the relationship of the accused to the victim and the socio-economic status of the family of the victim.

These numbers might lead a wonk to believe that the root cause of infanticide is socio-economic status and the accompanying problems that complicate family dynamics. I propose that those hypotheses are wrong.

What I found was that there have been numerous studies around infanticide where the accused murderers were interviewed and it turns out that there is a marked difference between the motivation to kill and the gender of the accused.

The papers pointed in the direction of a biochemical/psychological motivation for women to nurture crying/colicky babies that men lack. When women kill babies, it tends to be done with malice and forethought. When men kill babies, it’s more often a crime of passion and an impulse. The men in the studies could be said to have momentarily lost their grip on reality while trying to manage a squalling infant and accidentally killed the child by shaking or striking it.

Back to the original question - why is qualitative data an important acoutrement to quantitative data? Because if you just looked at the statistics, you might think the best policy to reduce infanticide is to work on socio-economic status.

In fact, I believe that to reduce the (larger) proportion of infanticide incidences committed by men, there is a need for a public education campaign that lets men know how they might react to a squalling child. A little forethought and perspective could go a long way to save a few childrens’ lives.

By way of analogy, that is how I feel about the relationship between quantitative and qualitative research. Quantitative research is the metrics. Qualitative research is the context.

*********************
Aaron “Crooky” Cruikshank is the Principal and Founder of Friuch Consulting. He has written professionally about science and technology for ten years.

Share/Save/Bookmark

Topics: Policy, Research Methodologies |

Comments