For 9/15/2009:

On the topic of causality - one of the things I most frequently encounter when reviewing the literature on crime and deviance is not necessarily a blatant misuse or misunderstanding of causality when scholars are presenting their empirical work, rather it is once this work is picked up on by mainstream media outlets that terms like ‘causes of crime’ start to be misused.  Most of the truly good work in the field is peppered with conditional statements and caveats, which are then abandoned as they become presented to general audiences.  I have selected two cases below which highlight this misuse of causality.

The first selection is from the Heritage Foundation, a conservative think tank located in Washington D.C.. Drawing from a review of the major theories in criminology, Patrick Fagan has managed to isolate the breakdown of the American family as the root cause of the current trends in violent crime. (http://www.heritage.org/research/crime/bg1026.cfm). Fagan reduces decades of research into twelve succinct bullet points that lend credence to this connection between the breakdown of the American family and violence. Among various logical fallacies (such as deducing individual patterns from neighborhood level data) and selectively avoiding issues such as spuriousness (single family households are highly correlated with economic disadvantage and residential instability) and more nuanced theoretical arguments (rise in single family households is one of many social phenomenon that has co-occurred with the rise in violent crime beginning in the 1970s indicating perhaps a greater underlying trend), he irresponsibly misrepresents the connection between marriage and desistence from crime. When laying out the root causes of violence, Fagan acknowledges findings from Robert Sampson and John Laub’s age-graded theory of social control (without actually acknowledging the authors themselves) by stating that ‘Criminals capable of sustaining marriage gradually move away from a life of crime after they get married.’

 This relationship has been the point of a somewhat contested debate between scholars who propose a dynamic life-course perspective to offending (that is to say that people can age out of crime) and those that reduce offending behavior to a single latent propensity (arguing that the underlying motivation to offend is relatively stable across the life course – most notably Gottfredson and Hirschi 1990 – discuss more in person). Within this argument is a fundamental issue of social selection vs. causation. Proponents of the latent propensity model would argue that the marriage effect is irrelevant as individuals who are most likely continuously engage in criminal behavior are the least likely to maintain stable relationships or marry.  Thus the crime reducing effect of the marriage treatment is overstated as those who are going to desist are more likely to self-select into marital relationships.  Marriage, unlike other adult roles, is something that cannot be randomly assigned and evaluated (such as work programs or education), therefore estimating the true effect of marriage on desistence of crime is a sticky process – and something that PSA can speak directly to! Rather than outline my own design here, I will instead indicate a recent paper that utilizes PSA to get at the heart of this issue.   King et al (2007 – Criminology) using propensity matching techniques, concluded that even when considering propensity to marry, marriage serves to suppress offending for males.  Moreover, those who are the lowest propensity to marry benefit the most from marriage –lending more credence to the dynamic life-course perspective.

The second selection is an article on America’s most dangerous cities published by Yahoo Finance (http://finance.yahoo.com/real-estate/article/106978/America's-Most-Dangerous-Cities). In this article, Zach O’Malley Greenburg attributes the high crime in the ‘most dangerous cities’ to operations of Mexican Drug cartels. He further examines the unique situation in Detroit (America’s Most Dangerous City) and attributes the high levels of crime to a break-down of the industrial economy, suburban flight and a lack of resources dedicated towards crime prevention.  The issue with this article (and moreover the idea of ranking cities on a scale of dangerousness) is one of measurement error.  The unique social/cultural/historical forces at work in each city makes cross – city comparison damn near impossible. Moreover, metropolitan areas with large suburban sprawl (think Washington D.C) tend to have inflated crime/ victimization rates when considering only the city proper. For instance, a man from Fairfax, VA working in D.C. is mugged on his way to the metro station in downtown D.C. If we are interested in examining the violent crime rate in the city of D.C. (# of offenses / population), this offense is only counted in the numerator of the crime rate, while the denominator reflects only those people living in the city proper.  Thus central cities in large metro areas will have very biased crime rates.  The selection of high crime cities (as America’s Most Dangerous Cities) are therefore inherently biased.