Friday, May 28, 2010

There goes David Brooks again....

When David Brooks cites a stat for didactic purposes, hit the factcheck button.

Today, Brooks undertakes to explain why Bad Accidents Happen to Good Workers, citing several allegedly ineluctable laws of human nature alleged to explain why we're biased toward minimizing risk. Among them:
Third, people have a tendency to place elaborate faith in backup systems and safety devices. More pedestrians die in crosswalks than when jay-walking. That’s because they have a false sense of security in crosswalks and are less likely to look both ways.
As an inveterate New York diagonal jaywalker, I took some satisfaction in this apparent factoid. But then I began to think. Aren't there a lot more people in the crosswalks than outside them? Ten times as many, maybe? Okay, in New York, five times as many?

Google and Wikipedia steer casual inquirers on this question to Tom Vanderbilt's In Defense of Jaywalking, published in Slate last November.  There, lo, I found half my NY conjecture vindicated -- minus any figures as to what percentage of street crossings in the city are outside the lines:

In New York City the Post, quoting numbers from the DoT, said 50 jaywalkers were killed annually; this is a high number to be sure, but just one-fourth the total pedestrian death toll.
I would venture to speculate, though, that well more than one-fourth of NYC street crossings are in crosswalks. Vanderbilt's analysis provides further statistical complexity:

Yet a closer look at national data shows that 59 percent of pedestrian deaths for which location information was recorded happened in places where pedestrians had no convenient access to a crosswalk. While jaywalking is often cited as a cause of pedestrian accidents, less than 20 percent of fatalities occurred where a pedestrian was crossing outside an easily available crosswalk.

The second sentence might provide some basis for Brooks' assertion -- but again,only if more than 20 percent of pedestrians jaywalk within range of an easily available crosswalk.

State and city DMV statistics for pedestrian accidents are hard to apply to Brooks' dictum because they track multiple accident causes on all kinds of roads, most of which don't fit into the jaywalk/crosswalk dichotomy. New York State DMV stats for 2007, for example, break down as follows: Crossing with Signal - 4,142; Crossing Against Signal, - 1,657; Crossing, No Signal, Marked Crosswalk - 1,091; Crossing, No Signal or Crosswalk - 3,352; Other Actions in Roadway - 1,186; Unknown - 1,840; and several other assorted causes with fewer victims. 

A Toronto Transportation Services statistical survey of pedestrian collisions offers similar ratios (and ambiguities). Here are the five-year averages (2002-06) for various causes of collisions: crossing with right-of-way - 791; crossing without right-of-way - 375; crossing - no traffic control - 175; crossing ped crossover - 74; crossing marked crosswalk without right-of-way - 31; on sidewalk or shoulder - 165; running onto roadway - 142; other -308; unknown or blank - 115; and a several other causes in single or double digits. 

In short, this vintage Brooks stat shot with a tidy conclusion attached tells us very little.

More broadly, Brooks' "rules" as to why risk management tends to fail highlight various facets of inbuilt human optimism, but they do not address the core question of incentives. If our bias is to minimize risk, institutional procedures have to push against the grain. A few proposed principles below.
  1. Don't put people with bottom-line responsibility in charge of risk management. Attention, governments: don't put those is charge of revenue collection in charge of inspection or enforcement.
  2. Don't ghettoize risk managers and compliance officers.
  3. Don't penalize workers at any level who take a stand in favor of a risk-mitigating course of action
  4. Institute a "tie goes to the safety advocate" presumption in any dispute over whether to choose a more- or less-risky course of action.
  5. Don't incentivize undue risk-taking (okay, this one is hard. Businesses do have to reward some forms of risk-taking. The banking industry is struggling now with how to make employee compensation bring risk home to the employee.  At a bare minimum, though, it's fair to say "don't incentivize risks to human life and limb."

No comments:

Post a Comment