Thursday, November 4, 2010

Trees and Crime: Correlation and Causation, Part 652


Once again I am moved to post on my favorite bugaboo: mistaking correlation with causation.  This time the subject is a study of the affect of neighborhood trees on crime in Portland that The Oregonian reports on in today's paper.  Part of my problem is with the study itself but part of my problem is fundamental mistakes in the way it has been reported on by The O.

Let's start with an excerpt of the abstract of the study (which was authored by two Forest Service employees):

In general, the authors find that trees in the public right of way are associated with lower crime rates. The relationship between crime and trees on a house’s lot is mixed. Smaller, view­obstructing trees are associ­ ated with increased crime, whereas larger trees are associated with reduced crime. The authors speculate that trees may reduce crime by signaling to potential criminals that a house is better cared for and, therefore, subject to more effective authority than a comparable house with fewer trees.

This is all fair enough, the study looks at the correlation between trees and crime controlling for a host of other observables. They find some statistically significant correlations and are careful in the abstract not to make a causal statement. Yes, there are reasons to expect that trees signal greater police and neighbor vigiliance (think the broken windows theory) and thus asct as a deterrent to crime. But there are a host of other reasons to think that areas with higher crime might be associated with lower income, shorter tenancies and thus lower investment in trees so the true causal realtionshiop is far from clear.  In simpler terms, in high crime areas, the crime might be a disincentive to invest in the planting and upkeep of trees, in which case the causality runs in the other direction.

Unfortunately the authors do nothing to try and uncover the true causal link, and so all they can do is talk about correlations and speculate as to the causal story as the abstract illustrates. To this extent they are honest. The title of the study is a big problem however: "The Effect of Trees on Crime in Portland, Oregon." This is simply inaccurate, this is not a paper about the effect of trees on crime (a causal statement), it is about the association of trees and crime.

Perhaps this is the root of the problem with the reporting then. Here are three snippets from the article in The O.

1) The title of The Oregonian article (this is the web version): Tall trees help protect houses from crime, says study conducted in Southeast Portland

Uh, no.

2) "Houses with tall trees had less crime because trees made the area look more desirable."

Uh, no (this doesn't even get their speculative story right - if an area is more desirable, why isn't it more desirable to criminals too?)

3) "Street trees, a bigger tree canopy and more trees were all associated with less crime." [Italics mine]

Ah, yes, yes, yes! Do you see the crucial difference in the third statement from the first two?

Now this association is interesting in its own right, so that is not a problem, but you have to be very careful about crossing over from correlation to causation.

Class dismissed.

1 comment:

Jeff Alworth said...

Ah, but you fail to consider the Ent angle. Perhaps these are crime-fighting trees! I know the apple tree in my front yard looks shiftily at the neighbors when they stroll by.