essentially tried to extend the M&M experiment to the scale of a metropolitan school system. He tested several different incentive programs in public schools—offering bonuses to teachers if they improved their classes’ test results; offering incentives like cell-phone minutes to students if they improved their own test results; offering families financial incentives if their children did better. The experiments were painstaking and carefully run—and the results have been almost uniformly disappointing. There are a couple of bright spots in the data—in Dallas, a program that paid young kids for each book they read seems to have contributed to better reading scores for English-speaking students. But for the most part, the programs were a bust. The biggest experiment, which offered incentives to teachers in New York City, cost seventy-five million dollars and took three years to conduct. And in the spring of 2011, Fryer reported that it had produced no positive results at all.
7. The Coding-Speed Test
This is the problem with trying to motivate people: No one really knows how to do it well. It is precisely why we have such a booming industry in inspirational posters and self-help books and motivational speakers: what motivates us is often hard to explain and hard to measure.
Part of the complexity is that different personality types respond to different motivations. We know this because of a series of experiments undertaken in 2006 by Carmit Segal, then a postdoctoral student in the Harvard economics department and now a professor at a university in Zurich. Segal wanted to test how personality and incentives interacted, and she chose as her vehicle one of the easiest tests imaginable, an evaluation of basic clerical skills called the coding-speed test. It is a very straightforward test. First, participants are given an answer key in which a variety of simple words are each assigned a four-digit identifying number. The list looks something like this:
| game | 2715 |
| chin | 3231 |
| house | 4232 |
| hat | 4568 |
| room | 2864 |
And then a little lower on the page is a multiple-choice test that offers five four-digit numbers as the potential correct answer for each word.
| - | A | B | C | D | E |
|---|---|---|---|---|---|
| 1. hat | 2715 | 4232 | 4568 | 3231 | 2864 |
| 2. house | 4232 | 2715 | 4568 | 3231 | 2864 |
| 3. chin | 4232 | 2715 | 3231 | 4568 | 2864 |
All you have to do is find the right number from the key above and then check that box (1C, 2A, 3C, etc.). It’s a snap, if a somewhat mind-numbing one.
Segal located two large pools of data that included scores from thousands of young people on both the coding-speed test and a standard cognitive-skills test. One pool was the National Longitudinal Survey of Youth, or NLSY, a huge survey that began tracking a cohort of more than twelve thousand young people in 1979. The other was a group of military recruits who took the coding exam as part of a range of tests they had to pass in order to be accepted into the U.S. Armed Forces. The high-school and college students who were part of the NLSY had no real incentive to exert themselves on the tests—the scores were for research purposes only and didn’t have any bearing on their academic records. For the recruits, though, the tests mattered very much; bad scores could keep them out of the military.
When Segal compared the scores of the two groups on each test, she found that on average, the high-school and college kids did better than the recruits on the cognitive tests. But on the coding-speed test, it was the recruits who did better. Now, that might have been because the kind of young person who chose to enlist in the armed forces was naturally gifted at matching numbers with words, but that didn’t seem too likely. What the coding-speed test really measured, Segal realized, was something more fundamental than clerical skill: the test takers’ inclination and ability to force themselves to care about the world’s most boring test. The recruits, who had more at stake, put
