From the Judge's Chambers

Data-driven or driving the data?

By William C. Whitbeck


Frederick Jackson Turner (1861-1932) was a historian who liked maps. He also liked measuring things. And when he put his two avocations together in 1890, he found something very interesting. For years, the Census Bureau had made it possible to draw a map of where people were living in the United States. Using this map, one could draw a line between areas where there were more than two people per square mile and areas where there were less than two people per square mile. That line represented the American frontier.

The frontier had been moving Westward almost since our inception as a country. But by 1890 the Western territories were settled and as Turner looked at his map he concluded that it was no longer possible to draw a clear line. According to the numbers, the frontier was a thing of the past.

Turner saw the closing of the frontier as a seminal event in American history. In 1893, he put forward his famous “Frontier Thesis,” asserting that, “This expansion westward with its new opportunities, its continuous touch with the simplicity of primitive society, furnishes the forces dominating the American character.” In a sense, Turner was one of the originators of the concept of American exceptionalism. But he was also a new breed of cat, a historian who was into measurement.

We are still in the habit of measuring things. Sometimes we do it poorly, sometimes we do it well. And sometimes we do both at the same time. Take a famous example: COMPSTAT, a data-driven crime-tracking system that former Mayor Rudy Giuliani and former Police Chief William Bratton used in New York City in the 1990s. The results were spectacular. New York’s murder rate and overall crime rate dropped significantly. Giuliani and Bratton were able to claim a least a partial victory in the war on crime; the numbers don’t lie they said, and the numbers proved that crime in New York was down dramatically.

But suppose the numbers contained some lies. Radley Balko, writing in Reason magazine, reports that there is a scandal brewing over COMPSTAT. According to Balko, Long Island’s Molloy College conducted a survey of retired high-ranking police officials in New York. The responses showed that, “police commanders faced heavy pressure from higher-ups to reduce felonies to misdemeanors—or in some cases to not report crime at all—in order to make the numbers look prettier.” In other words, the cops may have been cooking the books.

Now, this is not an unusual occurrence. From body counts in Vietnam to exaggerated results from Head Start to numbers games with school headcounts in Michigan, those actually on the ground have shown a repetitive, and entirely human, tendency to report that which favors them.

But does this mean that we should not focus on results when evaluating spending on governmental programs? Or that we should forego data-driven analysis in favor of more touchy-feely anecdotal reporting? Of course not. It simply means that we should be very careful not to steer the data to some foregone conclusion.

Everyone—but in particular Mayor Giuliani and Police Chief Bratton—wanted crime to go down in New York and perhaps they drove the data they received from COMPSTAT to that end. But, as Balko

points out, you can’t fake homicide . . . and there is no question that homicides decreased in New York in the 1990s when Giuliani and Bratton were in charge. And there is also no question that they were focusing on just that result.

So, like Frederick Jackson Turner, we who are involved in government should be into measurement. If the census data showed that the frontier was closed in 1890, then the data we receive today can show us how effective we are in what we do. The taxpayers that are footing the bill deserve no less. Sometimes it is as simple as seeing that a line on a map had disappeared. And that the first act in the American drama was over.

––––––––––––––––––––
Subscribe to the Legal News!
http://legalnews.com/Home/Subscription
Full access to public notices, articles, columns, archives, statistics, calendar and more
Day Pass Only $4.95!
One-County $80/year
Three-County & Full Pass also available