Here is the amazing tale of one of the biggest, longest, most important social science experiments ever undertaken. It’s a 60-year odyssey of data collection and analysis that eventually comes to a very surprising conclusion. That’s Joan McCord, up above, the hero of this saga and the first-ever female president of the American Society of Criminology (photo: Geoff Sayre-McCord). OK, yeah, at first glance Ms. McCord does not seem like she’d be the star of “the most exciting” of anything. But read on, data fans, read on.
Big, big thanks to Freakonomics Radio for bringing us this story in their podcast “When Helping Hurts.” You can listen to it on their site, or right here:
The “When Helping Hurts” podcast page includes a written transcript and a long list of related references and resources.
Here’s the set up… The Cambridge-Somerville Youth Study was one of the first large-scale longitudinal randomized social experiments. Back in the 1930s, Dr. Richard Clarke Cabot took a group of 250 at-risk Boston suburban youth and paired them up with similar kids who were doing well. He then randomly assigned half of the pairs to be the treatment group, and half to be the control group. For six years the treatment group received mentoring, counseling visits, summer camp, and other interventions designed to assist them. The control group received no interventions. The kids were monitored for progress. The data was analyzed in 1948 and again in the 1950s, and no treatment effect was found. That is, the mentoring and other assists had no measurable effect on the kids.
The bulk of the podcast is about Joan McCord’s re-analysis of the data in the 1970s, when new and more sophisticated techniques became available. She compared the participants on length of life, criminal record, mental and physical health, and job and marriage satisfaction.
Ha! I am not going to spoil the suspense, or a good story well told, by revealing the surprise ending. You’ve got to listen to the podcast, or (as I prefer) read the transcript. But be assured that this is one data analysis adventure with some very thought-provoking results. Why do we measure? To find out what works and what doesn’t. This great tale of measurement and analysis is both an inspiration and a caution. ∞
