Bryan Caplan  

How to Fix Group Projects

The "Courage Campaign's" False... Tabarrok on Innovation...
When I was a student, I always hated group projects.  As a professor, I never assign them.  The source of my antipathy: Group projects provide terrible incentives.  Since everyone gets the same grade, the lazy and incompetent free ride off the hard-working and able.  The free riders often add insult to injury by deliberately aggravating the people doing the work - and loudly protesting, "You can't boss me around."

Still, group projects are not without their defenders.  Their best argument: People work in groups in the real world, so they might as well get used to the process now.  Sure, group projects have built-in problems.  But it's important for students to learn how to overcome these problems.

So far, so good.  But this defense of group projects conveniently overlooks massive differences between the typical school group project and the typical work group project.  Namely: In school projects, students are equals.  Even if there's a "team captain," he's selected democratically, and every team member gets the same grade.  In the workplace, in contrast, there's an externally appointed boss.  The boss gets rewarded for group success - and hands out rewards to team members based on their contribution.

If we really wanted to use group projects to prepare students for the world of work, then, we'd totally change the incentive structure.  Away with equal status, equal rewards, and democracy!  Instead:

1. The teacher would begin by selecting the best students to be team leaders. 

2. Team leaders' grades would be based on their group's performance - but with an average grade well above the average grade for the class as a whole.  If the average student gets a B-, say, the average team leader would get an A-.

3. The team leader, not the teacher, would grade his own team members, using a budget of points based on his group's overall performance.

For example, suppose there are four teams with five students each.  The teacher hands out the following grades to each group:

Team #1: C+
Team #2: B
Team #3: A-
Team #4: D

The teacher's judgment would have two effects.  First, it implies the exact grades of the group leaders.  Suppose leaders get a +1 grade bonus:

Leader #1: B+
Leader #2: A
Leader #3: A+
Leader #4: C

Second, the teacher's judgment implies the average grade group leaders are allowed to give their team members.  The leader of Team #1, for example, has to give his team members an average grade of C+ (2.33 grade points per student).  He could hand out four grades of C+.  But he could also give {an F, a C, a B+, and an A} - which also averages out to 2.33 grade points per student.  Since the team leader wants to maximize group performance, he has a strong incentive to reward performance and punish its absence.

To improve the system further, the teacher would demote underperforming leaders to the ranks, promote the best-performing non-leaders to leadership roles, and allow quits and firing.

Many people will object that team leaders might "play favorites."  Clearly some would.  But at least they'd pay the price: Team leaders who reward incompetents will get lower grades themselves - and have trouble retaining talent (and their leadership positions!) if there's repeated play.  In any case, if we're trying to teach people about the real world, isn't learning how to handle and cope with favoritism a vital skill? 

Another objection might be that team leaders would be uncomfortable giving unequal grades to fellow students.  Fair enough.  But at least leaders would pay the price for their own squeamishness.  And once again, they learn a vital skill: to put your feelings aside and judge people on their merits.

The main problem with my proposal, I'm afraid, is that there's less than zero incentive for teachers to adopt it.  Even I'm not planning to experiment with my own proposal.  Yes, if students went to school to learn real-world skills, they'd thank me.  But in practice, the reward for my effort would be a lot of angry student resistance - and rock-bottom evaluations.

Comments and Sharing

COMMENTS (18 to date)
Andreas Moser writes:

I don't like group projects.
I best work alone, like James Bond.

Fabio Rojas writes:


You and I have had this argument a number of times. I am the unnamed defender in the posts.

My response to the bad incentives problem is simply to allow each team to eject free riders. Every semester when I have group projects, I require students to write down what they did and this must be approved by other students. Students who were described as non-participators get an F. If a students insists that they were unfairly treated are given a chance to do their own project.

There is the remaining issue of variation in performance. My view is that the errors introduced by assigning a group grade are small in comparison to the monitoring costs. Arguing over whether a C+ student should really be a B- is too much effort. A simple lesson on the price of information.

And this makes sense. In jobs, the reward for being a better team players is rarely an immediate increase in pay. In industries where group work is the norm, you just have to live with some inaccurate immediate pay-offs.

The incentive for good team work is in the future. You may not get an immediate boost in pay, but bosses know who to promote and who to invite to be on more important projects which do carry more pay or prestige.


drobviousso writes:

The best team projects I had were at either senior or graduate level classes where roles were assigned, and each person got graded on their role. One was a comp sci class, and we had an algorithm role (figuring out how to solve the problem) a coding role (writing the code that solves the problem) and a leader / documenter role (logistics, journaling, most of the writing of the final report). This worked well because each role had clear responsibilities.

But, before the students are able to really produce work product that means anything, group projects are hard and generally aren't useful. I learned much, much more about working in a group from Boy Scouts than I ever did in a classroom.

Eric Morey writes:

The best pedagogy is the result of a long process of adapting new designs from feedback based exclusively on "angry student resistance - and rock-bottom evaluations".

Airman Spry Shark writes:

The really aggravating group projects were always those with instructor-assigned groups. When free association is allowed, self-selection serves to sort students such that the equality with which they are evaluated is a reasonable approximation of reality.

An alternate method of enabling team leaders to be comfortable distributing unequal grades would be to precontract with members their individual difference from the group grade. Have each non-leader submit a delta at which they'll work (e.g., +0.7, -0.4); only leaders will see these numbers (after they've been adjusted such that they sum to zero over the class). The leaders then take turns drafting members (order could be determined by merit, bidding (with the leader bonuses), or lottery). At the end of the project, the leaders' bonuses (post bid, if that system is adopted) are adjusted by the opposite of their groups' net adjustment (if you've taken on all the dregs, the leader would get a larger adjustment, whereas a leader who drafts all the superstars might actually get a lower grade than the group. The strivers won't feel so put-upon, since they'll be secure in getting a bonus, while the slackers will be incentivized to exert themselves, since they'd need the group to get, for instance, a B in order to secure a C- for themselves.

Thomas F. writes:

An excellent proposal. I do hope some teachers try to improve on this idea and do their best to implement group projects in such a way. The way they were handled while I was in school was just as Caplan described - poor.

Perhaps mitigate the "angry student resistance" via using an opt-in system. Explain how the group projects would will work in advance, give grade bonuses to those who participate in the group version of the assignment, and assign smaller, frequent assignments to allow reputation to develop. Hopefully this will also lead to some level of self-selection affect - the slackers will be afraid to join (or continue participating in) a group dynamic where their low-productivity is so visible.

Pandaemoni writes:

People also work in groups in the real world outside of their employment, so there may not always be a leader appointed. In any event, the process of sorting out who will do what in the absence of a leader is in a sense just a more complex version of working on a team with an externally appointed boss.

That said, I hated group projects because of the free riding. Less aggravating, but emotionally more challenging than that was the one time I was on a group project, and two teammates sincerely wanted to contribute, but their good faith contributions were dragging down the quality of the project as a whole.

david writes:
Many people will object that team leaders might "play favorites." Clearly some would. But at least they'd pay the price: Team leaders who reward incompetents will get lower grades themselves - and have trouble retaining talent (and their leadership positions!) if there's repeated play.

Ding ding ding. If there's no repeated play, then regardless of what average grade team leaders are assigned, they have every incentive to award themselves perfect grades and everyone else zero. If there's repeated play, then this is no longer a problem to begin with:

Since everyone gets the same grade, the lazy and incompetent free ride off the hard-working and able. The free riders often add insult to injury by deliberately aggravating the people doing the work - and loudly protesting, "You can't boss me around."

since retaliatory slacking-off comes into play. Iterated Prisoner's Dilemma and all that.

FWIW, I actually had your proposed system several times in high school (some variant of selected leaders picking team members) - it's not all that uncommon, esp. when a few students are known to have the relevant skills and the the teacher wants to spread them out a bit.

Richard Allan writes:

david, you missed the point. Leaders don't assign their own grades. Also watch the UK TV show "Come Dine With Me". Everyone has an incentive to rate everyone else's contribution as a zero secretly to maximise their own chance of winning, and yet they don't.

Ryan Goodman writes:

I went to a high school with this exact system minus the leader bonus points. Firing rarely worked since teachers thought that the team leader wanted to get rid of the underperforming kids, since they were "underperforming". Teachers want fairness and equality, and to protect feelings. We had points allocated based on certain criteria, they were either biased by friendship, or misallocated by teachers since they couldn't observe every group.

Project based learning even systems like yours are fantastic in academic circles, but in schools, kids are biased, students are lazy, teachers decry inequality.

Projects work better in high performing groups like magnet schools, or business schools.

david writes:
david, you missed the point. Leaders don't assign their own grades. Also watch the UK TV show "Come Dine With Me". Everyone has an incentive to rate everyone else's contribution as a zero secretly to maximise their own chance of winning, and yet they don't.

Repeated play, altruistic norms, etc. In the absence of which there's no difference whether team leaders assign their own grades or not; the leader has zero regard for what grades other people get in exactly the same situations the free-rider problem is present (i.e., where norms formed by repeated play do not hold). And team members know this. The logical result is therefore exactly identical to the free-rider problem.

GlibFighter writes:

Group projects are vehicles for free riding and collective punishment. The optimal solution is to never require that a group have more than one member.

Tracy W writes:

I tend to think that the main problem with group projects is that most teachers have very little conscious knowledge about how to effectively manage a group, so how can they teach others? Generally students are left to sink or swim.

This seems to be a general problem with many educational reforms. If teachers haven't been taught something themselves, how many of them can teach it well in their turn? Particularly as teaching well requires a larger set of background knowledge than doing well - a good teacher anticipates where their students are likely to go wrong.

Willem writes:

My alma mater has (or at least had) this system in place for several curricula. They also had quasi-random assignment to groups based on recent performance. The best students get the best groupmates, to make sure they get enough stimuli.

One thing I would have added would have been peer-review of your teammates and public discussion of these results. I read about a professor doing this once, and think that in groups not too large or dynamic this would be the best way to improve all students, based on feedback.

david writes:

Teachers should be in the business of facilitating learning, not assigning grades.

Lloyd Lofthouse writes:


I was doing this with my students decades ago. I did not like or respect group work where everyone earned the same grade.

That seldom happened in my classes.

I taught English and journalism in the public schools mostly at the high school level for thirty years, and I was doing almost exactly what you describe here. One difference though. Each group could vote to fire members that did not work and ended up dividing points earned for a five member group among four members.

In addition, each group was allowed to negotiate the division of points which often resulted in someone having an A+ and another student having a D- with other grades in between.

Few earned FAILING grades unless they were fired from a team. Anyone that was fired, was assigned independent work from me and those assignments were more demanding than the group work.

Steve writes:

I went back to school in the 90's in my mid 30's. Group projects seemed to be at that time, kind of "all the rage." I basically didn't like them because the dynamics enforced were not real. If you could pick your on team you could do pretty well, but if not, you were kind of stuck. However, many of the younger students had no real "team" experience, and here, the goal is always the grade. What was troubling to me is that no teacher or prof ever attended any of the group meetings or functions so had no real insight into how the group dynamics worked or were functioning. So they didn't know who was working and who was slacking or who was obstructing. I know teachers figure that the output of the group shows the result of the team dynamics but they really taught nothing about how a group project was to proceed or how a team should handle itself, deal with problem members (since no one was in charge) etc.

And if you ever had a contrary viewpoint, you were doomed on the team, as group-think ruled the day. I was on one such team, made up of my friends or people I would prefer to work with, but they group was trying to give the teacher what they thought he wanted....a mistake where this teacher was concerned. We got a poor grade as he saw through our output. But I really doubt anyone else learned anything from that experience. I relearned that most groups like this are very dysfunctional and wonder why anyone would start them or use them even in a work environment. I agree there should be a boss to account to, but that boss has to be involved.

I object to meaningless group projects also, but have taught senior design classes where group projects were the point of the class (much of the first quarter was spent teaching about group formation and group management). There is no reason for group members to all be given the same grade, but if there is a group leader, he or she should be given the lowest grade of the group, not the highest.

Our groups were required to write a charter that included an explicit mechanism for ejecting non-functioning members of the group.

See more discussion at

Comments for this entry have been closed
Return to top