In previous posts (here and here), I’ve been writing about the different ways that groups can be created for class projects, and whether the method of group assignment has an impact on learning. In this post, I want to focus on one argument in favor of allowing students to select their own groups: self-selected student groups tend to be more cohesive than other kinds of groups. Group cohesion refers to the extent to which individual group members feel strong connections with the group and share a unity of purpose, and if we set aside the concern that having students select their own groups has the potential to be an unpleasant and stressful experience for some students, it does make sense to assume that on average self-selected groups should be more cohesive. When given the option, students are likely going to choose to work with others whom they already know and like.

Greater cohesion can be useful because it can increase goal commitment (Klein and Mulvey 1995) and decrease social loafing (Karau and Hart 1998), and several meta-analyses of the research on group cohesion suggest that there is a positive relationship between the cohesiveness of a group and the group’s performance on tasks (Beal et al. 2003; Gully, Devine, and Whitney 2012; Mullen and Copper 1994). Thus, it’s easy to take this to mean that greater cohesion would be a good thing for student learning groups.

However, as is my way, I’m going to play the contrarian for a moment. Those previously mentioned meta-analyses look at group performance, not learning, and they are aggregating across a wide array of different operationalizations of performance. A sizeable portion of the research on group dynamics comes from the fields of organizational psychology, business, and management, which means that some of these measures of performance have little relevance to learning. For example, Beal et al. (2003) report on performance such as the total monthly sales by a sales team and the win-loss ratio for athletics teams. Research showing that a cohesive sales team can sell more widgets in a fiscal quarter than a non-cohesive sales team shouldn’t naturally lead us to assume that a cohesive group would learn more from a collaborative assignment.

Further, much of the research on group cohesion is correlational, comparing the performance of existing groups that differ in level of cohesiveness. At least one of the meta-analyses I mentioned above suggests that it is possible that the biggest driver of the relationship between cohesion and performance is actually the direct path from performance to cohesion. In other words, part of the reason correlational research shows that cohesive groups tend to perform better is because a history of success tends to bring a group together (Mullen and Copper 1994).

With those caveats out of the way, I will admit that it does make conceptual sense that cohesion would be beneficial for collaborative assignments. When groups are cohesive, we might expect greater motivation, satisfaction, and persistence. Additionally, the concept of cohesion is itself similar to the concept of positive interdependence, which Johnson and Johnson (n.d.) argue is a necessary component of collaborative learning. Their argument is that the benefit of collaborative learning comes from students exchanging ideas, asking each other questions, brainstorming, problem solving, and providing feedback to one another, and for this to occur, students have to feel as though they have a shared purpose or goal (i.e., interdependence).

The relevance of group cohesion to collaborative learning is further supported by a meta-analysis that examined the strength of the cohesion-performance relationship across different types of group tasks (Gully, Devine, and Whitney 2012). Some tasks involve very little interdependence, such as when each individual simply contributes their independent work to some whole product at the end. In these tasks, the performance of the group depends on each member, but there is little actual interaction, and in these cases, the relationship between group cohesion and performance is weak. On the other end of the spectrum, though, are tasks that involve group members engaging in multiple stages of back-and-forth communication. This ongoing and reciprocal exchange is more characteristic of what we would hope to see in collaborative learning groups, and the meta-analysis showed that the cohesion-performance relationship was strongest for these kinds of tasks.

In a related meta-analysis, Chung and colleagues (2018) examined the research comparing the performance of friend-groups relative to acquaintance-groups, and they found somewhat different results. Generally, friend groups were shown to perform better than acquaintance groups, but task interdependence was not a significant moderator of the effect. They also found that the positive effect of being in a friend group became more pronounced when the group was larger. This makes sense when you consider that social loafing becomes more common in larger groups, and it can be mitigated by cohesion.

One additional finding from the Chung et al. (2018) meta-analysis had to do with how performance was assessed on tasks. Some tasks, which they called maximizing tasks, measure performance in terms of quantities or durations (e.g., a brainstorming task that involves coming up with as many ideas as possible). In such cases, the tasks may benefit from the increased motivation and trust that we can experience from being in cohesive groups. Other types of tasks, which they called “optimizing tasks,” are assessed in terms of the quality of the performance (e.g., when a group is tasked with coming up with the best possible solution to a problem). As an instructor, this distinction seems important because I don’t typically give students an assignment that asks them to create the longest presentation they can or to write as many pages as they can. While friend groups were found to be more productive than acquaintance groups on maximizing tasks, there was no difference between the two types of groups when it came to performance on optimizing tasks. Chung and colleagues’ argument for why this might be the case is that friendship groups likely get a boost to motivation that can help with persistence, but they might also be less inclined to critique themselves. Maintaining positive social relationships between group members may take precedence, and behaviors that could be interpreted as conflict (e.g., providing critical feedback or debating the best way to do something) may be less likely to occur.

In the above paragraphs, I’ve pointed to research that shows the relationship between cohesion and performance is either positive or non-existent. As I’m repeatedly finding, there aren’t clear and concise answers to questions about how groups function. For example, it could be that there are instances in which certain types of conflict within a group could be beneficial (de Wit, Greer, and Jehn 2012), and to the extent that cohesion reduces the likelihood of such conflict, it could be less desirable. Even the concept of cohesion itself could be an oversimplification of a multi-faceted construct that includes social cohesion and task cohesion dimensions (Carless and De Paola 2000; Chang and Bordia 2001). However, I don’t want this to come off as some sort of pointless exercise, so I’ll try draw some summary conclusions.

Cohesion can be beneficial for groups, particularly if there is a risk of social loafing occurring among group members, for example in larger groups or on projects where individual efforts can be masked. There is some evidence that cohesive groups perform better, but it’s not always clear that the research showing links between cohesion and performance generalize to a relationship between cohesion and learning. Still, a mixture of results that range between positive and neutral would suggest that, all things being equal, cohesion isn’t a bad thing. Self-selected groups are likely to be more cohesive than other kinds of groups because they are more likely to be composed of individuals who already have relationships that connect them. But relying on preexisting relationships may not be necessary. Some of the research on group cohesion is experimental, which means the level of group cohesion is manipulated within the study. For example, in one study, asking two people to spend time discussing an issue on which they agreed strongly before beginning a task was enough to increase cohesion and reduce social loafing (Karau and Hart 1998). That means that if you put in a little work early on – either by creating groups based on shared interests or by incorporating cohesion-building activities once groups have been created – you can potentially foster group cohesion among a group of students who don’t know each other very well.


Beal, Daniel J., Robin R. Cohen, Michael J. Burke, and Christy L. McLendon. 2003. “Cohesion and Performance in Groups: A Meta-Analytic Clarification of Construct Relations.” Journal of Applied Psychology 88 (6): 989–1004.

Carless, Sally A., and Caroline De Paola. 2000. “The Measurement of Cohesion in Work Teams.” Small Group Research 31 (1): 71–88.

Chang, Artemis, and Prashant Bordia. 2001. “A Multidimensional Approach to the Group Cohesion-Group Performance Relationship.” Small Group Research 32 (4): 379–405.

Chung, Seunghoo, Robert B. Lount, Hee Man Park, Ernest S. Park. 2018. “Friends with Performance Benefits: A Meta-Analysis on the Relationship between Friendship and Group Performance.” Personality and Social Psychology Bulletin 44 (1): 63–79.

de Wit, Frank R. C., Lindred L. Greer, and Karen A. Jehn. 2012. “The Paradox of Intragroup Conflict: A Meta-Analysis.” Journal of Applied Psychology 97 (2): 360–390.

Gully, Stanley M., Dennis J. Devine, and David J. Whitney. 2012. “A Meta-Analysis of Cohesion and Performance Effects of Level of Analysis and Task Interdependence.” Small Group Research 43 (6): 702–725.

Johnson, David W., and Roger T. Johnson. (n.d.) “An Overview of Cooperative Learning.” Cooperative Learning Institute.

Karau, Steven J., and Jason W. 1998. “Group Cohesiveness and Social Loafing: Effects of a Social Interaction Manipulation on Individual Motivation within Groups.” Group Dynamics: Theory, Research, and Practice 2 (3): 185–191.

Klein, Howard J., and Paul W. Mulvey. 1995. “Two Investigations of the Relationships among Group Goals, Goal Commitment, Cohesion, and Performance.” Organizational Behavior and Human Decision Processes 61 (1): 44–53.

Mullen, Brian, and Carolyn Copper. 1994. “The Relation between Group Cohesiveness and Performance: An Integration.” Psychological Bulletin 115 (2): 210–227.

David Buck, associate professor of psychology, is the 2020-2022 Center for Engaged Learning Scholar. Dr. Buck’s CEL Scholar project focuses on collaborative projects and assignments as a high-impact practice.

How to cite this post:

Buck, David. (2020, October 13). Do Cohesive Groups Matter? [Blog Post]. Retrieved from