There are a few sociology majors over at the Universityof Maryland, college Parkwho should be demanding their money back. It seems that sociology professor, Phillip Cohen, is claiming that because the rise in single-mother families in Washington DC has continued to rise while the rate of violent crime has begun to drop, single-motherhood is not a cause of violent crime. Now, I’m not saying that single-motherhood is a cause of violent crime; what I am saying, is that Prof. Cohen is misusing statistics to state that it is not.
Let’s say, for instance, that I notice that from the 1960’s through the mid-90’s, the number of dead goats in the state of Indiana consistently rises from year to year, then begins a steady decline from the mid-90’s to the present. I also notice that the rate of goat milk consumption also increases from the 1960’s through the mid-90’s. I might be inclined to think that the rate of goat milk consumption in Indiana has caused the increase in the number of dead goats in Indiana. Then, when the number of dead goats begins to decrease while the rate of goat milk consumtion continues to rise, I might begin to believe the rate of goat milk consumtion did not cause dead goats in Indiana. In other words, the rise in demand led directly to an increase in the goat population which meant that more goats would die in any given year.
Prof. Cohen is making a similar claim in relation to single-mothers and violent crime. The now negative correlation between violent crime and single motherhood is proof that single motherhood never caused violent crime. I’ll give him some credit. He provides a disclaimer that allows for plausible deniability of his ignorance of statistics:
“Looking at two aggregate trends is never enough to tell a whole story of social change, of course. However, if two trends going together doesn’t prove a causal relationship, the opposite is not quite as true. If two trends do not go together, the theory that one causes the other has a steeper hill to climb. In the case of family breakdown driving crime rates, I don’t think the story will make it anymore.”
What Prof. Cohen’s disclaimer shows is that he does grasp the statistical concept at work here. But his grasp of statistics isn’t present in the rest of his article.
There have been a number of sociological studies that have shown a correlation between the rise in single motherhood and the rise in violent crime, but I know of no one who claimed that single motherhood is the only factor involved. Further, the fact that violent crime is now dropping while single motherhood continues to rise, does not necessarily indicate the disappearance of the positive correlation between these two factors.
Returning to the example of the dead goats in Indiana, the rise in the number of dead goats may have been due to more than one factor. The increased demand for goat milk may have led to more goats being brought into Indiana. Poor regulation of the goat farming industry may have combined with poor veteranary standards in the 1960’s to create an unhelathy environment leading to the death of a lot of goats. Therefore all three factors may have led to the increase in dead goats. In the 1990’s imporved regulation and better veterinary care may have reduced the number of dead goats even while demand for goat milk continued to increase. If the goats live longer and more productive lives, fewer goats will be necessary to meet the demands, fewer goats will also mean fewer dead goats. Demand for more goats will alwayts lead to increasing numbers of dead goats, but other factors may intercede causing a drop in the overall number of dead goats despite the increasing demand.
Apparently the media has not discussed Prof. Cohen’s observation that single motherhood does not cause violent crime and this is distressing to Prof. Cohen. This may be for good reason. Social science research has demonstrated a correlations between the two (correlation is not causation). There are a number of other factors being discussed as possible reasons for the decline in violent crime such as “rising incomes, improved law enforcement technology, and community relations.” There may be other factors as well.
Using statistical analysis, it is possible to determine approximately how much impact any single factor has on the rate of violent crime. Let’s suppose that single motherhood accounts for about 25% of all violent crime. Income and law enforcement also account for 25% each. Then let’s say that single motherhood increases, but income also increases and law enforcement becomes more effective. The net result would likely be a decrease in violent crime. But if I were to ignore those other factors and only graph single motherhood and violent crime, then I could be just like Prof. Cohen who misuses statistics to create lies. You could be just like Ampersand at Alas, a Blog who most likely never took a stats class and simply believes his professor’s misrepresentations because he is saying what Ampersand wants to hear. Or you could demand that a proper statistical analysis be performed to determine the impact of each of these factors on violent crime in order to determine whether or not single motherhood has a significant impact on violent crime.
This kind of stupidity really gets my goat.