Sunday, September 26, 2010

Research Study Shows Merit Pay Doesn't Work for Students Either

After yesterday's rant on my blog, I continued to research the issue of merit pay, or in academia-speak, "financial incentives for performance output.""  I ran into another fascinating study that I somehow missed when it came out in April 2010.  This study looked at the effectiveness of paying for standardized testing improvements as well, but this time the money went to the students instead of the teachers.

This study, conducted by Roland G. Fryer Jr. of Harvard University, EdLabs, and EBER, was quite extensive.  It was a four-year study among the four (edu-speak again) "protypically low-performing urban school districts" of Chicago, Dallas, New York City, and Washington DC (and, having lived in Washington DC, I can attest to what a desperate state the schools there are in).   It involved giving a total of $6.3 million to about 38,000 students in over 250 different school, based on their achievement within a number of different incentive systems.  The study involved students from 2nd-9th grade, including MIDDLE SCHOOLERS.   It was also much more controversial than the previous study with the teachers; not only did Fryer get thrown out of quite a number of schools, but he received hate mail and even death threats from those who say his work as "bribing students" for good performance.

Once again, there were good news/bad news result:

THE BAD NEWS:  Student merit pay didn't work in raising standardized test scores.  Or, to be more specific, offering students up to $2,000 (a significant amoutn for students, especially those from low-income families) for better grades in classes and weekly exams did nothing in terms of raising their end-of-the-year standardized test results.

THE GOOD NEWS:  However, those students who were paid for performance--that is, just showing up regularly, following directions, good behavior, etc.--did show some significant improvement in their year-end scores.  The best investment?  Paying student to read books.  That produced the largest gain in reading comprehension scores of all the incentive schemes, and at a relatively low price; the average student only earned $14 in incentive grants.  But they had much better results, not only compared to other incentive programs, but compared to other educational reforms like reducing class size and increased early education programs that cost thousands of dollars more.

To me, the two studies are both fascinating -- and related.  Once again, the result tell me that students don't WANT to have bad test scores.  You can offer what is a relative fortune to them ($2,000) to improve their scores, but they can't do it.  This suggests that low student performance comes not from "laziness" or lack of motivation, but from an inability to do any better.  On the other hand, if you give students even little sums of money to do what they can do--show up at school, be on their best conduct, read another book or do than they would have done otherwise--those small behaviorial changes can add up to a significant increase in student achievement.

In presenting this study, I don't want to argue that we should necessarily be using monetary incentives with students.  As I have said in earlier posts, I'm a great fan of Alfie Kohn, and am persuaded by his book entitled Punished by Rewards (a great interview about the contents of that work can be found at  However, I would not rule out something that some of our most struggling schools have found to be effective in raising student achievement.

But, once again, I think it demonstrates the difficulty in imposing the industrial model on education.   This study says, in my interpretation, that low test skills are not a result of lack of interest or effort or motivation by students--a problem that MIGHT be solved by economic incentives.   Rather,  if the child doesn't understand the concepts being tested, offering $2,000, $10,000, or even $1,000,000 doesnt make any difference.  The child can't pass the test.  But now, the child feels even more frustrated, even more downcast, even more worthless, than simplying failing the test without outside incentives.

Not a good idea, to my mind.


  1. On the other hand, offering small, tactful bonuses for submitting content, for example, can work well to promote collaborative projects.

  2. Agreed. But I would say that is another performance task, rather than a motivational change.

  3. I am not sure about performance-enhancement (task management) vs. motivation here. For example, people are much more likely to forward announcements about collaborative projects with money bonuses or reputation bonuses (e.g. organized by well-known entities). People are more likely to sign up for those, as well.