Problems with Performance-Based Teacher Compensation

This post is based on an email I wrote in response to an article that a friend shared with me about education:

http://www.washingtonpost.com/wp-dyn/content/article/2010/07/11/AR2010071103628.html

One of the things the article discusses is performance-based compensation for teachers and how it seems to be gaining favor among experts, philanthropists (most notably the Bill and Melinda Gates Foundation), and the government. It certainly sounds good in principle, because

  1. it’s more fair to pay people who work more and/or harder
  2. bonuses might motivate people to do better e.g. by encouraging them to put more care and effort into their work.

Based on my own experience and what I’ve learned from Dan Ariely, I’m a bit skeptical on whether the concept can be made to work in practice.

We should definitely reward people who do better, but as Wall Street has demonstrated, bonuses can backfire dramatically. In his new book, The Upside of Irrationality, Dr. Ariely describes an experiment to see how large bonuses affect people’s performance. An interesting aspect in the design of the experiment was that they chose to do it in India. This allowed them to offer substantial bonuses without breaking their research budget. All the participants were given the same set of tasks, and were told they would be awarded a bonus for doing very well on a task. The participants were randomly put into one of three groups. One group was offered 1 days’ wages as bonus. Another was offered 2 weeks’ wages. The last (and luckiest) group was offered 5 months’ wages. The surprising result was that the last group performed the worst, while there wasn’t much difference between the first two.

One explanation for this counter-intuitive result is that the low-performing group was distracted by the large bonus that they stood to gain (or perhaps in their minds, lose), and their lack of concentration prevented them from doing as well as the others. Whatever the reason, it’s clear that large bonuses can be counter-productive.

One of the big risks of a bonus system is that you displace other motivations that people have for working. In Predictably Irrational, Dr. Ariely uses a (fictional) example to illustrate the principle of displacement. Imagine, you are visiting your in-laws for Thanksgiving. At the end of the magnificent dinner, you propose to pay your mother-in-law for the wonderful meal that she has lovingly prepared. This is not likely to go over well. The problem is that your mother-in-law is not interested in money (how irrational!). The bigger problem is your obliviousness to the social context of Thanksgiving dinner, and the internal motivations that drive the those interactions.

To avoid this, rewards that are more relevant to teachers’ intrinsic motivations might be better at encouraging them to work harder + perform better. Another approach is to emphasize other types of rewards. For example, we could give “good” teacher more autonomy by imposing less bureaucracy on them.

The other part I find troubling is has to do with how we actually measure performance. This is certainly a case where the devil is in the details. I would agree with the average teacher who, as the article describes, is wary of how teacher performance is measured. It would be naive and unfair to look at the mean score that a teacher’s students receive on a standardized test. The reason for this has to do with the simple fact that every class is different. For example, one class may perform better than another simply because they started out better prepared, and the difference in test scores might not have anything to do with the quality and effort of the teachers. This is the sentiment that I’ve heard directly from teachers, and it makes alot of sense. As if it weren’t hard enough to compare classes within a school, the challenge of comparing them across schools is even greater.

In short, context is key to understanding teacher performance. In order for metrics to capture this, states could use a set of variables (such as family income level) to allow class averages to be compared with all else being equal. Not having taken any courses in statistics, I don’t claim to know how to construct such a metric, but I’m sure there are some good statisticians that have a clearer idea.

I’m not saying performance-based compensation can’t work, but we have to be mindful of a few things:

  1. We need to make sure the bonuses don’t become the reason that teachers come into work every day. Most teachers (especially good ones) aren’t motivated by money. Of course, this is not an excuse not to compensate them well. However, waving money in front of their faces could destroy what intrinsic motivation they have, leading to
    1. lower performance
    2. less job satisfaction, which could lead to more teacher turn-over
  2. We also don’t want bonuses to become a distraction.
  3. We have to get the performance metrics right.

It’s really hard to know when bonuses become “too large” and start to corrupt people’s motivation. Figuring out an appropriate amount requires great wisdom and social sensitivity. On the other hand, coming up with good metrics is probably just a matter of doing some good ol’ fashioned science. This is why I believe performance-based compensation can succeed, but without careful consideration, it can easily backfire.

Advertisements
This entry was posted in Uncategorized. Bookmark the permalink.

2 Responses to Problems with Performance-Based Teacher Compensation

  1. Alon says:

    well put. if you have the time, grab a copy of Freakonomics and read (at least) the chapter about incentives in education

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s