View Mobile Site
 

Ask the Expert

Signal Photos

 

Tim Myers: Confusion over cause and effect

Myers' Musings

Posted: June 23, 2012 1:55 a.m.
Updated: June 23, 2012 1:55 a.m.
 

I make my living accumulating, calculating, analyzing and attempting to forecast numbers, primarily financial.

In 27 years of this undertaking, I learned I cannot predict numerical outcomes very well, but I can 100-percent predict people's reactions to numerical results.

Take the annual rite of school standardized test-score reporting: If a school scores high and/or shows consistent improvement in its scores, administrators and communications folks will immediately speak breathlessly about new teaching methods, interventions and parental involvement.

Conversely, those falling short of numerical thresholds will mumble statements concerning the tests "not showing the whole picture" and bemoaning the reduction of education to numerical metrics.

Ironically, scientific studies reported in the academic literature provide a murkier view. The only absolutely accurate predictor of performance on standardized tests still remains the educational achievement level of the child's mother. (The higher, the better.) Bizarrely, Steven D. Levitt of the University of Chicago revealed in his "Freakonomics" series an inexplicable relationship between the number of books a family owned and children's results on standardized test scores (The higher, the better.).

The educational administrators' reaction represents the classic error of cause and effect. Since the administrators bear responsibility for the outcome on test scores, they assume their actions either caused the good outcome or need excusing in the case of a poor outcome. Academics who study these matters know, however, that demographics and other meta matters govern 90 percent of these outcomes and will overwhelm even the best intentions and efforts of those putatively responsible.

These facts, however, did not stop Santa Clarita city officials from breathless celebration of FBI crime "statistics" released within the last two weeks that showed a year over year decrease in crime in Santa Clarita.

Now, no doubt exists that one of the major responsibilities of local government lies in the area of public safety. Municipal professional and elected officials would naturally tout numbers that would seem to indicate a good job in this area, particularly in Santa Clarita where in the last City Council election candidate Jon Hatami made the descent into criminal chaos, particularly of Canyon Country, at the neglect or active complicity of incumbent city government, the centerpiece of his (unsuccessful) campaign.

Even incumbent Mayor Laurie Ender jumped on the bandwagon musing in a candidate forum that the city should transfer a pending annexation area (Jakes Way) known (perhaps incorrectly) for high crime to Palmdale rather than deal with the problem in the blessed confines of Santa Clarita.

But many issues make this connection of municipal and police diligence between falling (or rising) absolute crime numbers tenuous. Not least of all, the FBI crime report itself in its introductory passages cautions against drawing conclusions or making policy based on their reports, since they include only the reporting of absolute numbers subjected to no statistical rigor.

Second, due to low absolute numbers of crimes in most suburban enclaves in the United States, a small change in a low base number when converted to a percentage can seem to indicate an exaggerated trend. For instance, take the number of homicides in 2010 and 2011 in Santa Clarita that increased from 1 to 5 (a 500 percent increase).

But, before we all go under our beds assuming the citizenry stands five times more likely to die in the mean streets of Santa Clarita, we need to examine the numbers more critically.

Two of the incremental deaths related to a double murder of an elderly couple and then the suicide of their son in law. One related to the violent denouement of a long-term feud between neighbors. These "ad hoc" type incidents, while shocking, could not in any rational way constitute a trend.

And then we come back to the error of cause and effect. City Manager Ken Pulskamp gushed at the recent news, attributing putative success to more "aggressive" policing. The fact remains that demographics govern crime statistics more than the actions of particular law enforcement, particularly on a long-term basis.

Employed, older, prosperous folks just seem to behave better in the aggregate. In fact, that same Levitt in his original "Freakonomics" tome posited that long-term crime-rate declines that began (unpredicted) in the early '90s until today were more darkly related to the broad legalization of abortion in the early 1970s that resulted in demonstrably fewer persons of certain demographics more likely in the aggregate to commit crimes.

Not something to particularly celebrate.

Tim Myers is a Valencia resident. "Myers' Musings" runs Saturday in The Signal.

 

Comments

Commenting not available.
Commenting is not available.

 
 

Powered By
Morris Technology
Please wait ...