How Can We Save Ourselves From Ourselves?

John List and Uri Gneezy have appeared on our blog many times. This guest post is part a series adapted from their new book The Why Axis: Hidden Motives and the Undiscovered Economics of Everyday Life. List also appeared in our recent podcast “How to Raise Money Without Killing a Kitten.”

It’s a late-September afternoon in 2009 and the students of Fenger High School on Chicago’s South Side are crossing a vacant concrete lot. Some live in the Altgeld Greens housing project. Others live in a part of Chicago’s rough Roseland neighborhood (also called “The Ville”).  Some of the students from these areas have developed fierce antipathies toward each other, though the groups are more like cliques than gangs.

As the teenagers cross the lot, a fight breaks out. Someone pulls out a cell phone and starts recording a video of 15 to 20 kids fighting. Around a minute into the video, someone discovers a couple of two-by-fours lying in the empty lot. Eugene Riley, sporting a red motorcycle jacket, takes one of the big pieces of wood from a pal and swings it like a baseball bat into the back of 16-year-old honor student Derrion Albert’s head.

“Dannnggg!” someone exclaims. Screaming and shouting, the kids start running—some towards the shouting, others away from it. Derrion tries to get to his feet but he is punched and kicked as someone shouts, “Oh my god, you guys!” Derrion attempts to protect his head. There’s a distraction, but eventually the camera pans back to Derrion. He’s on the ground, defenseless, staring blankly at the camera and at that moment he’s attacked again. Eventually his assailants flee and immediately the cameraman and others run up to Derrion.

Someone says, “Get up, son.” His friends pick him up and bring him into a community center adjacent to the empty lot. His friends scream his name, desperate for him to respond. Two minutes into the video, you finally hear a siren. Derrion died hours later.

It didn’t take long for the three-minute video of Derrion’s murder to reach the public—quickly becoming national news. The calls for action were fierce. Desperate to try something new to stop the violence Chicago Public School (CPS) students experience every year, the CEO of CPS, Ron Huberman, devised a plan to stop the killing and asked us to help.  

His plan was to match the most at-risk students with a highly-paid advocate who, in Ron’s words, would “act as a part mentor, part truant officer, and part role model to the youths.” To get the project started, Ron asked the following question:  “Out of 700 schools and 400,000-plus students, how do we figure out who is most likely to run into trouble?” Together with Dana Chandler and Steve Levitt, we quickly developed a way to predict the CPS students most at-risk.  With the help of our list, Ron Huberman chose the 250 students most at risk.  These teens were the recipients of their own “Secret Service” agent for the rest of the school year.

After the school year, we found that at-risk students in the program were 1.41 percentage points less likely to be a victim of a shooting. The students that just missed the cut-off for our program? They too were less likely to be a victim of a shooting, but only by 0.1 percentage points. The design wasn’t truly random—after all, the goal was to predict the students most in need of help—so we can’t be sure that the effect of the program was robust, but the evidence is strongly suggestive.  Most importantly, the story shows that the economic science can go a long way if there is a willing partner to listen. 

If you want to explore our world further, take the Why Axis Challenge: visit, post a photo of your copy of The Why Axis, and be entered to win prizes, including a meeting with Uri, John, and Freakonomics author Steven Levitt! Be sure to stay tuned for more posts to come, which will give a glimpse into more “undiscovered economics.”


Instead of simply promoting your book why not share the actual name of the program and how people can participate.

Eric W

1.4% doesn't seem like a lot of difference. What was the cost of this program? How likely were they to be shot overall?

Enter your name...

1.4 percent is not a lot. 1.4 percentage points, however, is not the same thing.

If you have a 1.5% chance of getting shot, then a reduction of 1.4 percent means that you still have a 1.479% chance of getting shot. However, a reduction of 1.4 percentage points means that you have only a 0.1% chance of getting shot. In this example, a 1.4 percentage point reduction is a 93.3 percent reduction.

I don't know what starting numbers they were working with, but the principle is the same no matter what the numbers are: a big relative reduction on a small baseline is still a small number.

Steve Cebalt

Some of the stastics, if I read them correctly, are shocking in and of themselves: roughly speaking and rounding, between 1% to 3% of students in these groups are victims of a shooting in a given semester, do I have that right? I appluad the use of behavioral economics to try to make a difference, although here, it appears to have made no real discernable difference -- or inclusive at any rate. Changing social behavior is a monumental task, not always subject to programmed interventions; but we have to try and test, as was done here.

The last paragraph of the article above is a little unclear and lacking context for non-statistical people like myself....I hope I've understood this correctly. I'd also be interested in the authors' views of futher research or implications for other interventions that may be indicated by this initiative.

Zhen Liu

"1.41% less likely to be a victim of a shooting?" Are you kidding? What's the prior,what are the uncertainties, how significant are the results, any causual relationship? Maybe I haven't heard the whole story, but for someone who is constantly bombarded by hypes and analsysis/mis-analysis from CNBC, this smells fishy.


1.4%? That seems extremely low.

Paul in Va

Lev / Dub, The "Why Axis" snippets are getting very tiresome, very fast.

First List/Gneezy don't take the advise on the book title from the readers of your blog (Who ARE their target audience), and now your subjecting us to repeated posts? Your friends wrote a book. We get it. Enough already.

Like Dan H.'s over-politicized posts, advertising your friend's book is NOT why I read the books, read the blog, and Listen to the podcast. I think you are loosing your focus... and fear you may start loosing your audience, which would be a shame. My own drivel notwithstanding, you have a great set of educated, involved, and caring commentators. Please stop taking them for granted with these posts.


Let's see, now. Highly paid advocate -> 1.31 decrease in shootings (difference in decrease between with & without the advocate). And no statistics on knifings, beatings with 2x4s, &c?

I also have to wonder about the statistical significans of that 1.41%, seeing as there were only 319 shooting incidents in the 2011-12 school year, per Chicago Tribune article here:


What all of the responses so far seem to be overlooking is that the statistic above is 1.41 percentage points. NOT 1.41 percent of the previous value. That is to say, if we take Steve's figure and assume 1-3% of students are victims of a shooting per semester (let's say 2%), times 2 semesters = 4% per year. A 1.41 percentage point drop means lowering that 4% to 2.59%, which is a quite substantial (35% of the previous shooting rate) reduction. In this case the math semantics actually make a big difference - for low probability occurrences like these, there is a huge difference between percentages of the whole, and percentage points.


Are we all overlooking this, or is the article simply not written clearly enough? If the intervention does indeed produce a 35% drop in shooting rate rather than 1.31%, wouldn't it be quite easy to actually state that?

Also, there were 113,873 high school students enrolled in the 2011-12 school year (per CPS website: ). Given the 319 shootings reported above, that's a rate of 0.28%, not the 4% you quote.

For low probability occurrences like this, random variation would seem to be a real possibility - but I Am Not A Statistician :-)

Enter your name...

Relative risks—the percent reduction rather than the absolute number or percentage point reduction—is a favored tool of pharmaceutical companies because it sounds so impressive: If you're a young adult with mild hypertension, then take our drugs! You'll halve your risk of getting a stroke. (Please don't notice that your risk of getting a stroke is tiny to begin with.)

The best possible number, which they don't provide, is NNT (number needed to treat): How many "Secret Service Agents" would they need to hire for one school year, to prevent one shooting?

Voice of Reason

What? This is getting pretty absurd. If people want body guards, they can hire them. But solving crime by literally assigning a security detail on every poor kid in the inner city is not an acceptable solution to solve crime!


You guys really should know better than to report a precise figure like a "1.41 percentage point" reduction with such a small sample size. How many of those 250 students usually get shot in a year??

steve cebalt

Hi Tim: I agree: The 1.41 percentage point reduction MAY be significant -- or not -- how do I know? These book excerpts make me appreciate the alchemy between Mr. Levitt and Mr. Dubner even more than I did, because of their ability to make complex data easily understood, with clear context. I appreciate this study on shootings, but I felt like I was left to my own devices to discern its significance and meaning...which "authors" are supposed to do for me. I don't mean to be critical -- rather, appreciative of the unique gifts of the Freakonomics team vs. others.

Julien Couvreur

On the theme of interrupting violence, readers may be interested in this TED talk on methods of slowing down escalations:
There also is a documentary on his effort called The Interrupters.

One thing that fascinates me is that, given that violence and poverty feed off each other, how do we ever come out of that cycle and culture?
Violence (aggression against people and property) is not productive in the long-term, it erodes trust and shortens horizons and makes it difficult to come out of poverty. And poverty tends to shorten horizons as well (day to day living), make better alternatives than violence less attractive or available, and increases conflicts over scarce resources.