The Chronicle: Colloquy: Responses

There are several problems with quantifying broken windows that need to be addressed before we can hope to validate the theory. First, we need to remember exactly what the theory entails. In many discussions about it, the only thing spoken of is the correlation between disorder and serious crime. These elements are on the extreme ends of a four part causal chain. Disorder is theorized to cause fear. This fear results in a decline in social controls (through decreased social contact). Social control is generally operationalized as community cohesion. This lapse of community cohesion is theorized to cause increased serious crime. So, as I have conceptualized the theory, we have four major variables:

disorder
fear
community cohesion
serious crime

Further, these elements are theorized to form a causal loop, creating a "spiral of decay" -- the subtitle of Skogan's (1990) book. Of course, this theory cannot be modeled using the path analysis techniques Skogan utilized (the model would be nonrecursive). Skogan recognizes these limitations, but I believe that Kelling and Coles were overly optimistic in their appraisal of Skogan's work as providing empirical validation of broken windows. He simply showed a relationship between disorder and an act he classified as serious crime (robbery). A serious problem I have with this model is the treatment of crime and disorder as two distinct constructs. These factors are generally measured by public perceptions to survey questions on a three point Likert scale. Respondents are asked to rate specific acts as "no problem" "some problem" or a "big problem." PCA of the data (which is freely available via the web) suggests that this dichotomy is an academic construct not supported by the data. What I suggest is that acts can be arranged along a continuum of seriousness. It stands to reason then that if we artificially divide such a continuum, the two halves will correlate. I also suggest that response patterns will vary according to how the individuals perceive the threat to affect them.

The beauty of broken windows theory is that it cannot be modeled statistically as specified. Even the modern miracle of SEM cannot close the gap between speculation and empirical validation. (The model would be saturated and thus several competing models would yield the same solution.)

I believe the most promising aspect of research into community dynamics has been the evidence that disorder and crime increase transience. We have long ago established that poverty is highly correlated with crime. If we buy into the idea that middle class flight from disorderly neighborhoods will increase poverty in a neighborhood, then we can establish a viable economic theory. While we social scientists debate the merits of these overarching theories and sort out correlation matrices that have so many asterisks they look like star maps, I suggest that America's police stick to solutions that work on the neighborhood level. What are these solutions? That all depends. In the absence of an overarching theory to explain things, the police will have to come up with innovative solutions for the specific problems they face in their daily work. This idea, developed by Herman Goldstein, is called problem-oriented policing (POP). POP essentially applies the scientific method to solving problems. As a social scientist, I think it's a good idea.

: -- Adam J. McKee, University of Southern Mississippi (posted 2/5, 10:35 a.m., U.S. Eastern time)

previous response

next response

JOIN THE DEBATE

STEP 1: Your contact information (required)

Your name:

Your title & institution:

Your phone:

Your e-mail address:

STEP 2: Your comments

STEP 3: Submit!

Be sure to include your name and affiliation. Legitimate requests for anonymity will be honored.

The Chronicle maintains Colloquy as an open forum where readers can comment on issues. Submissions may be condensed or edited for clarity, but editing is kept to a minimum. We correct errors in spelling, punctuation, etc. In addition, we remove statements that have the potential to be libelous or to slander someone. In cases in which people make claims that could be libelous, we will remove the names of institutions and departments. But in those cases, we will not alter the ideas contained in the submissions.

Please read our privacy policy or our Colloquy help page if you have questions.

Check this box only if you would like this submission to be considered for publication as a letter to the editor in the printed Chronicle.