May 2010
Sun Mon Tue Wed Thu Fri Sat
            1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31          

Authors' Committee

Chair:

Matt Blackwell (Gov)

Members:

Martin Andersen (HealthPol)
Kevin Bartz (Stats)
Deirdre Bloome (Social Policy)
John Graves (HealthPol)
Rich Nielsen (Gov)
Maya Sen (Gov)
Gary King (Gov)

Weekly Research Workshop Sponsors

Alberto Abadie, Lee Fleming, Adam Glynn, Guido Imbens, Gary King, Arthur Spirling, Jamie Robins, Don Rubin, Chris Winship

Weekly Workshop Schedule

Recent Comments

Recent Entries

Categories

Blogroll

SMR Blog
Brad DeLong
Cognitive Daily
Complexity & Social Networks
Developing Intelligence
EconLog
The Education Wonks
Empirical Legal Studies
Free Exchange
Freakonomics
Health Care Economist
Junk Charts
Language Log
Law & Econ Prof Blog
Machine Learning (Theory)
Marginal Revolution
Mixing Memory
Mystery Pollster
New Economist
Political Arithmetik
Political Science Methods
Pure Pedantry
Science & Law Blog
Simon Jackman
Social Science++
Statistical modeling, causal inference, and social science

Archives

Notification

Powered by
Movable Type 4.24-en


« May 19, 2010 | Main | August 27, 2010 »

28 May 2010

Jaywalking allowed?

David Brooks has a column in today's New York Times about our difficulties assessing risk, in light of the oil leak in the Gulf of Mexico. One tendency he highlights is our excessive faith in devices aimed at minimizing risk. Although his general point is probably correct, I think he makes a statistical error in the example he uses to illustrate this tendency. Brooks seems to confuse numbers with rates.

He writes: "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." I think that more pedestrians die in crosswalks than when jaywalking because more pedestrians cross the street in crosswalks than anywhere else. If the rate of dying is higher for pedestrians in crosswalks than when jaywalking, we might attribute the difference to our overconfidence in safety devices. However, the rate of dying need not be higher in crosswalks than elsewhere for more pedestrians to die in crosswalks. As long as many more people cross in crosswalks, it is likely that more will die in crosswalks. Brooks' point about risk assessment would have been stronger if he had considered not only the numerator (the number of deaths) but also the denominator (the number of people at risk of death) of the rate of pedestrian deaths that he's interested in.

Posted by Deirdre Bloome at 3:03 PM