(by Christopher Allen with Elyn Andersson and Shannon Appelcline)
Two years ago, the Bainbridge Graduate Institute (www.BGI.edu) faculty gathered to radically reinvent their sustainable business curriculum for the next decade. Our goal was not only to update course content, but also to significantly update how the material was taught. We wished to make our teaching process (our pedagogy) more interactive and also more effective for students graduating into a 21st-century work environment, where people increasingly work in teams-both online and offline.
As a specialist in group interactions, I was asked by the faculty whether formal graduate student study groups (called "Study Buddies") should consist of two people or of three. I did not have an easy answer to this question.
My expertise with group dynamics comes from professional experience as an entrepreneur and from considerable experience both building online communities and helping others to do the same. Through these processes, I've seen how groups of people act differently at different sizes. As I discussed in my previous blog post on Group Thresholds [Allen], there are pros and cons associated with different group sizes. However, the smallest group that I spoke of in that article was the 'working team size', which is a group of four to nine members (but ideally about seven). I didn't talk about groups consisting of less than four members-specifically dyads (a group of two people) and triads (a group of three people)-and that was what the BGIedu faculty was now interested in learning more about.
My first inclination when considering BGIedu's request was to say that triads would be the best size for study groups, but I wasn't sure. I knew from my past research on group size that a dyad has a significantly different behavior than any other small groups. However, I wasn't certain if triads should also be considered on their own, or if they were just the low end of one of those 'work teams', which I'd always thought of as containing four to nine members. Triads were a hole in my research.
I decided to dig deeper into this question by engaging my teaching assistant Elyn Andersson and my writing partner Shannon Appelcline to help me investigate further. This post is the fruit of our research and our discussions about dyads and triads. We hope it might help you decide whether a dyad or a triad is the better team size for groups that you are working in.
Before I share with you the details of our investigation, I want to be clear that dyads, triads, and small 'working team' groups can all be effective. Any challenge faced by groups of these sizes can be overcome by a good process, teamwork, training, or just positive energy. However, my thesis is that certain sizes of groups are better for certain purposes-and when a group size is selected appropriately it will require less energy and thus be more likely to be productive when the inevitable group challenges occur. When choosing among a dyad, a triad, or a working team, the important thing to focus on is thus the intent in forming the group.
For the groups being considered by BGIedu, the faculty felt that the students taking the new curriculum needed three things:
More formal support for studying (i.e. not ad hoc study teams);
The inclusion of emotional support in that support structure; and
A system where the members of a study group would help each other to keep their commitments to study.
The last requirement - that members of these study teams would be accountable to each other without one student just taking up the slack of another - was the toughest one.
Clearly, working teams of 4-9 people were too large for our purposes. Teams this size are rarely intimate enough to support the needs of individuals. In addition, research shows that they are poor at supporting accountability. That's because working teams of 4-9 are very vulnerable to social loafing (i.e., slackers)-except when the tasks assigned to these working teams are highly independent (i.e. an entrepreneurial team that has different members focused on different topics, such as management, development, operations, marketing and sales) or when they require high diversity. An upcoming blog will cover this topic in more depth.
Removing working teams left the question of dyad vs. triad - which our research would show were both very different from working teams. Triads ultimately fit BGIedu's particular need better. That was primarily due to advantages of accountability and cohesion, combined with some emotional support. However, some simple numerical properties of triads also prove useful: they have elemental diversity because they're up to 50% more diverse than dyads; and if someone misses a team meeting the other two-thirds of the team can still benefit. For all of these reasons, my recommendation to the BGIedu faculty was that these formal study teams be formed in triads.
Dyads, however, are quite appropriate for a variety of other purposes, and should not be dismissed. I'll be talking of some of their advantages before proceeding on to the lessons learned about triads themselves.
Everyone has had experience with dyads. The first social experience of anyone's life is a dyadic relationship - that between a mother and her unborn child. This parent-child relationship continues once the child is born. Overall, dyads are the most common social group; they are seen everywhere, in personal, academic and business relationships.
To decide if a dyad is the best group size for your needs, it is again important to focus on the purpose of the group. A number of interesting characteristics of dyads - some benefits and some limitations - make them unique from triads and other larger group sizes, among them:
There have been some academic studies [Taylor] that suggest that it's easier to be open and share within dyads. The research has shown that people feel safer about revealing personal and intimate information in a dyad than they do when a third person is added to the group. As a result, disclosure rates increase when people are put into dyads as compared to triads. Furthermore, adding a third person to a group causes its members to be more socially conscious, making it harder for members to listen to the conversation and at the same time be aware of their own thoughts.
Together it's this trust level (allowing for disclosure) and this focus level (allowing for better listening) that make dyads much stronger for emotional support than triads or larger teams. If the intent of a group is to have an intimate and fully open space, or if the work of the group is emotional in nature, this might make dyads a better choice.
Though dyads are clearly more effective for providing emotional support, triads can still offer that as well under the right conditions.
This intimacy and confidentiality of a dyad helps its members to build strong team dynamics relatively quickly. The two individuals not only will have formed a trust bond more easily than three or more individuals could, but they generally will have a better ratio of time to performance outcomes compared to groups of larger sizes, in part because less time is spent maintaining the team relationship itself.
In a triad or larger group, the group would need to put effort towards team work over task work and thus less time is put towards performing the task. In other words, more resources would be spent on relating to a teammate than on finishing a project.
The most unique element of dyads is that they are the only group size in which one voice is paired against another voice equally.
When was the last time you were in dyad in which you did not agree on an issue? Most likely, you and the other individual spent some considerable time trying to convince the other of your reasons, and the outcome was one of two things: you either both stuck with your opinion and walked away, or one of you changed your mind after hearing the other's ideas and opinions.
Because members can't 'vote', a dyad is either in agreement or disagreement, which forces a longer deliberation for tasks when consensus is required. There are times when you want a small group to deliberate deeply, and dyads are ideal for this.
Though dyads have several advantages, they can also bring with them a few additional problems - one of which is a power inequality of knowledge that would be detrimental within some teams. For some tasks, this could offset the benefits of having a more intimate small group. For my BGIedu groups, I was concerned that it might prevent some students from benefiting academically from the study exercise.
This topic has been addressed in a study [Day] that compared homogeneous teams that were entirely high-ability or low-ability to heterogeneous dyads of mixed abilities. The study found that while high-ability people tested better individually after learning the material in a dyad with another high-ability team mate, low-ability people did not gain much benefit from studying in a dyad with a high-ability team member.
This was somewhat surprising. A student with more knowledge of a subject should be able to mentor a student with less knowledge, therefore creating a true dyad study partner. However, this study showed that this was not the case.
The study also argued against another common conception. One might assume that two high-ability team members could learn more together, when they could bounce ideas off of each other, than they could alone. However, this was again not the case: high-ability individuals performed better after learning alone than after learning with another high-ability team member! Only homogeneous low-ability teammates performed better after being in a dyad than alone.
This study pushed me toward triads for study teams at BGIedu, as dyads would not be large enough to guarantee high performance outcomes; the teams would always consist of one higher performing student and one lower performing student.
Another disadvantage of dyads shows up in the area of accountability. The simplest version of this issue can be shown in this question: if one teammate says she did the majority of the work, while the other teammate says the work was shared 50/50, who was correct? A tougher question: if they both say that the work was shared 50/50, how is a third party able to determine if this actually was true?
Elyn, my teaching assistant, interned in a high school in Seattle while working towards her Master's in Teaching. Her experiences highlight this problem. In today's educational model, collaborative learning is highly valued, and Elyn worked to incorporate in-class group work into many of her lessons. When assessing the students' group projects, however, she found that it was difficult to determine if the students shared the work equitably or if some of the students rode on the hard work of their partners. In some cases, Elyn doubted that the work was split equally. In particular, she found that if she let the students choose their own partners, the students were more likely to say that 'both' did the work, rather than tell the teacher their friend slacked off on the assignment.
A recent study [Alkaslassy] supports Elyn's experience. In it, students working on a paired group assignment were asked to allocate the percent of work done between the pair. 86% of the students assigned equal credit between the two individuals in the group. At a surface level, one could presume that each party performed half the work and was honest in their assignment of credit. However, Elyn and I believe that in some dyadic relationships, friendship can become an overriding factor and that in those situations, students are more likely to allocate equal credit because of this friendship, not because of true equity.
In the psychological field, this relates to a well-known cognitive bias called the egocentric bias, where an individual will claim more responsibility for group action then an independent observer would credit them. Although this is most often discussed for larger groups, in dyads a member might rate their own efforts as being half of the effort when in fact is it is less. Thus in academic group projects, it can sometimes be more effective to have the students rate their peer rather than themselves.
Though group accountability can be a disadvantage for dyads, it improves when you have triads, thanks in part to a triad's inability to devolve. This, as it turns out, would be another comparative advantage that helped me choose triads over dyads for BGIedu.
Inability to Devolve
We all have experienced what happens when a party gets too busy or too noisy - the size of the conversation groups get smaller. If you are having an informal dinner for seven, sometimes the conversation will include the entire table. However while people are eating the conversation will typically devolve into smaller groups: four people at one end of the table will being having one conversation, while the remaining three will be talking at the end other. Sometimes one person will be left out of any conversation, but studies show that with small groups this is less likely to happen. We have a natural tendency not to want to exclude or reject individuals unjustly, so such a singleton will usually find a place in one conversation or another.
Working teams also have this natural tendency to devolve when confronted with external challenges (of which noise is one of the simpler ones) or internal challenges (such as differences of opinion), only reuniting when those challenges are resolved. A team of six can devolve gracefully a number of different ways: a three and a three; a four and a two; or three groups of two. Teams of five can devolve into teams of two and three. A team of four can devolve into two teams of two. In all of these cases the devolution does not exclude any individual's voice.
However, a triad can not devolve without excluding someone. Triads thus have a lot of subtle pressure to try to keep the group together, so that no individual is explicitly rejected - except for a good reason. This leads to more accountability for the individuals in the group, compared to teams of two where the dyadic nature of the group can override liability.
A dyad's inability to vote was an advantage of that group size, and thus issues that require deeper deliberation are often better with dyads. However, you sometimes don't want as much deliberation. You want fast action.
I find that we unconsciously gauge each others' opinions in a small group very quickly, and even more so with a triad. This means that a triad seems to be able to come to at least some decision very fast. There are many cases where this can be very useful - if that is the intent of your team.
As an personal example, I saw a presentation in the early 1990s from the Apple Human Interface Group (HIG) about their survey of how users colored "folders" on the Macintosh. The result was that there was little commonality - everyone colored folders differently for different reasons. Afterward, I met with the team and suggested that that if they instead asked teams of three to decide how to color code folders that they used together, then the results from all the different teams would have much more commonality. This suggestion ended up getting me a consulting job with Apple HIG, and shows how a triad can quickly create successful performance outcomes.
Triads are not the only group size that can vote easily - all odd size groups can - but I believe the voting of a triad to be fundamentally different because the voting can be so unconscious and informal that the members are not even aware of it. This can be useful, but it can also cause problems.
When I was just starting my business, I was doing a lot of work with a company formed by two partners who had a 50/50 stock split in their company. Theirs was a good partnership; they had very different but complementary skills and worked well and profitably together. As a fellow entrepreneur, I was a peer to both them, and we were developing a new product line together.
However, when I worked with them on this product line, I saw that I'd changed their very successful dyad to a triad, and I thus became a tiebreaker in many of their business decisions - including business decisions on issues that had nothing to do with our joint product. This was caused by a shift in power: one partner might ask me a question about marketing rather than asking his partner (who knew marketing), and then it would become our opinion against the remaining partner's. Conversely, I found the marketing partner asking me to side with him on development questions.
I found this quite troubling; a more manipulative person could have controlled or broken that relationship by always breaking the ties in their own favor. In our case, we never shipped the product line and went our separate ways. I couldn't help but think it might have been because my unofficial "vote" was causing problems in their partnership.
When getting a result is the purpose of a team, a triad is better than a dyad as it allows for this voting to take place. As I mentioned to BGIedu, a triad can have a two-to-one vote, allowing a better chance at task accomplishment. Task accomplishment is also achieved because triads stay on topic better than dyads due of this voting ratio: if one member of a dyad decides to go off topic, it is easy for the other to follow his or her lead. In a a triad, two out of the three members would need to 'agree' to follow an unrelated discussion.
One final advantage of triads when compared to dyads is that the knowledge base of the group is larger. I saw this issue clearly in my recent Social Web for Social Change class at BGIedu, where I changed the team size from two to three, and immediately saw a growth in elemental diversity that benefited the groups.
In this class, I have the students participate in an exercise on personal branding. I originally had the students pair up in teams of two, effectively giving each student a 'branding buddy' to help them form their personal brand and to give and receive feedback. In the last year, however, I decided to change this dyad to a triad.
At first, these groups followed the same methods I used for dyads: one person would self-brand themselves to a second person in the group, who would then echo back what they thought they heard the first person say, using their own words. However, it is difficult for two people to really hear the similarities and differences between what they said and what they heard. The third person thus became a neutral sounding board who told both individuals that spoke what they said. Together, all three could figure out the similarities and differences between the statements. After consensus is reached, all three students switch roles. I call this the Echo Exercise, and it shows how triads can be beneficial at times when the natural characteristics of dyads don't lead to optimum results.
As it happens, this article was created by a triad. It wasn't purposeful, nor was it meant to show off the findings of this article. Instead, the triad evolved naturally because it proved the best group size for writing this piece.
I came up with the initial concept for the article and wrote out the problem and introduction. Elyn then did considerable research and wrote up a rough draft of her findings. To that point, we were a committed dyad, working together in an organic way. However, we also had an article that felt like it was in two parts. That's when I had Shannon join us. He reorganized the article, unified its voice, and produced a final draft.
Though it wasn't purposeful, you can see the advantages of triads just from this example. Most notably, our triad made great use of elemental diversity. Though I held the core ideas and concepts for the article, Elyn provided the research expertise and Shannon provided the editorial expertise that were required to complete it. Voting also came into the equation (in the unconscious way that it tends to in triads) when Shannon effectively "broke the tie" of two different writing styles.
Since then, this triad has fallen apart. It's something that was created for a specific project and with all of our other tasks, we didn't have room for it going forward. Teams breaking apart is natural. Sometimes it happens under normal conditions (as was the case with this triad), and sometimes it happens as a result of unhealthy conditions.
I did not give BGIedu any advice at the time about the termination of a team, although it will eventually come up: some teams will end badly, and members of these groups may leave with an impression that a triad is a bad group size, even though the size did not lead the group to ending in an unhealthy way. Perhaps the groups should be counseled, or perhaps the members should jump right back on the triad horse ... but that's a topic for another article.
From here, I'd like to open the conversation up to you. Have you had experiences in either dyads or triads that you would like to share? Were you one of the BGIedu students in a Study Buddy triad? Did the experiences match the studies shown here or did you have an experience that differed? We invite your comments.
Allen, Christopher (2008) Community by the Numbers, Part One: Group Thresholds Life With Alacrity (blog) (/2008/09/group-threshold.html)
Alkaslassy, E. (2011) How often do students working in two-person teams report that work was shared equitably? Assessment & Evaluation in Higher Education, 36. 367-375. (http://www.tandfonline.com/doi/abs/10.1080/02602930903428700#preview)
Baer, M. (2010) The Strength-of-Weak-Ties Perspective on Creativity: A comprehensive Examination and Extension. Journal of Applied Psychology, 95 592-601. (http://apps.olin.wustl.edu/workingpapers/pdf/2010-02-008.pdf)
Basden, B.H., Basden, D.R., Henry, S. (2000) Costs and Benefits of Collaborative Remembering. Applied Cognitive Psychology, 1. 497-507. (ftp://18.104.22.168/ck/2011-03/165/030/165/530/Research%20ArticleCosts%20and%20benefits%20of%20collaborative%20remembering.pdf)
Bertucci, A., Conte, S., Johnson, D.W., Johnson, R.T. (2010) The Impact of Size of Cooperative Group on Achievement, Social Support and Self-Esteem. Journal of General Psychology, 137(2), 256-272. (http://www.ncbi.nlm.nih.gov/pubmed/20718226_)
Blaskovich, J.L. (2008) Exploring the Effect of Distance: An Experimental Investigation of Virtual Collaboration, Social Loafing and Group Decisions. Journal of Information Systems, 22 27-46. (http://scitation.aip.org/getabs/servlet/GetabsServlet?prog=normal&id=JINFE3000022000001000027000001&idtype=cvips&gifs=yes&ref=no)
Day, E.A., Arthur, W., Bell, S.T., Edwards, B.D., Bennett, W., Mendoza, J.L., & Tubre, T.C. (2005). Ability -based pairing strategies in the team-based training of a complex skill; Does the intelligence of your training partner matter? Intelligence, 33, 39-65. (http://www.sciencedirect.com/science/article/pii/S0160289604000972)
De Cremer, D., Leonardelli, G. J. (2003). Cooperation in Social Dilemmas and the Need to Belong: The Moderating Effect of Group Size. Group Dynamics: Theory, Research and Practice, 2, 168-174 (http://www.rotman.utoronto.ca/geoffrey.leonardelli/2003GD.pdf)
Laughlin, P.R., Hatch, E.C., Silver, J.S., Boh, L. (2006) Groups perform better than the best individuals on letters-to-numbers problems: effects of group size. Journal of Personality and Social Psychology, 4. 644-651. (http://www.agileadvice.com/archives/Research%20on%20Group%20Effectiveness%20vs%20Individuals.pdf)
Markam, S.E., Dansereau, F. Jr., Alutto, J.A. (1982) Group size and Absenteeism Rates: A Longitudinal Analysis. Academy of Management Journal, 25 921-927. (http://www.jstor.org/pss/256108)
Seibold, D., Kang, P., Gailliard, B., Jahn, J.. Communication that Damages Teamwork: The Dark Side of Teams. Conference Papers - International Communication Association; 2008 Annual Meeting. 1-32. (http://citation.allacademic.com//meta/p_mla_apa_research_citation/2/3/0/5/1/pages230513/p230513-2.php)
Smith, S. and Haythorn, W.W. (1972) Effects of Compatibility, Crowding, Group Size, and leadership seniority on stress, anxiety hostility and annoyance in isolated groups. Journal of Personality and Social Psychology, 22, 67-79 (http://psycnet.apa.org/index.cfm?fa=buy.optionToBuy&id=1972-24882-001)
Taylor, R. B. De Soto, C.B., Lieb, R., Sharing Secrets: Disclosure and Discretion in Dyads and Triads. Journal of Personality and Social Psychology, 37, 1196-1203. (http://www.rbtaylor.net/pub_jpsp_1979.pdf)