STUDY OF VIOLENCE COMMITS SUICIDE WITH METHODOLOGY

Date: Saturday, January 8, 2005

In December I wrote a column on the effects of welfare reform, and a couple of days later, I heard from Neil Wollman. He's a professor of psychology at Manchester College in Indiana and senior fellow of the college's Peace Studies Institute. He told me the institute conducts an annual study called the National Index of Violence and Harm, and the most recent one, using 2002 figures, had just been released. Could he send it to me? he asked and I agreed.

With no little trepidation, I must say. I'm as well-disposed toward peace as anyone, but I have noticed that organizations that make it part of their name probably aren't referring to the same concept I am. More along the lines of "Peace in our time" or "Peace at any price" than "Peace through strength."

And indeed, the study has a notable ideological bent, but there's nothing wrong with that. What makes it totally useless is its preposterous methodology, which consists of picking 19 unrelated statistics (several of them subdivided), converting them to percentages -- in this case, percent increase or decrease compared with 1995 -- and averaging the percentages. You can see the result at www.manchester.edu/links/violenceindex/ if you want to bother.

You should have learned in elementary school why averaging percentages makes no mathematical sense unless the absolute numbers are the same. Average 10 percent of 100 with 50 percent of 20 and you get 30 -- but it's not a percent of anything relevant.

The individual indicators are fairly reasonable in themselves, they're just not comparable. They were chosen, somewhat tendentiously, on the basis of a definition of violence as "an action or structural arrangement that results in physical or nonphysical harm to one or more persons" even if the harm isn't perceived. That leaves room for all kinds of ideological mischief, though I make no claim that's the intent.

It's just the way people see the world. In his summary of the study, Wollman notes that income disparities -- in class, race, gender and age -- "are not good news for a society that holds equality as a primary value." Maybe he believes that equality is a primary value, but in my view it is clearly secondary to liberty. The road to equality runs through tyranny, because you can't force people to turn out equal except by squashing them flat.

The index of personal harm is divided into things people do to other people - homicide, sexual offenses, battery, robbery and reckless behavior (which includes carrying or possessing weapons illegally, and driving under the influence), and things people do to themselves, including suicide and deaths from substance abuse. That index is almost 15 points better than in 1995, though as noted the actual numbers are nearly meaningless.

The index of societal harm has four parts. Government violence, in the sense above, results from law enforcement and from the correctional system, which gives equal weight to 71 cases of legal executions and 609,000 cases of incarceration of nonviolent criminals.

Corporate violence includes air pollution and effects on consumers and employees. Family violence includes domestic violence and child abuse and neglect. And fourth there's a catchall category called structural violence, which results from "the structure or hierarchies of United States society."

This category includes something called social negligence, which means lack of health insurance, hunger, homelessness and the high school dropout rate. The rest are infant mortality and life expectancy, hate crimes, employment discrimination, poverty disparity and gang membership.

What does it mean to average such disparate measures? Not much, but this index is slightly down too, and for a most interesting reason. The category for hunger had included only requests for emergency food, and, as the authors candidly admit, their source previously omitted "cities experiencing no change or a decrease in emergency requests."

But that indicator was going up so fast it had "a disproportionate impact" on the rest of the index. So they cut the weight of that indicator in half, subdividing the category and adding another indicator called "food insecurity" calculated by the U.S. Department of Agriculture (which the USDA itself says is not a measure of hunger).

When results are so readily manipulated, they aren't worth much -- but by now, you already know that.

The summary notes that a much larger share of the black population lives in poverty, compared to whites, but fails to mention the principal reason, which is that a much larger share of the black population consists of never-married women with children. The authors acknowledge that many people would believe that abortion ought to figure in the index, and direct them to sources for the data. Fine, but leaving abortion out while putting capital punishment in reveals that the whole enterprise is essentially political.

You know, there's a reason the phrase "academic exercise" is not a compliment.