Every once in a while a team that is sliding will have some type of player bonding activity that appears to turn the season around. A couple of quick examples would be the 2009-10 Ottawa Senators holding a fun skate in New York’s Central Park that appeared to trigger an extended winning streak and the recent fishing trip by Toronto Maple Leafs players from which the feature image for this article was taken. Such events really make for great stories and discussion points for talk radio. Should the Leafs players have gone on this team building fishing trip while in the middle of a playoff race? Did the skate in Central Park really bring the Ottawa Senators players together and allow them turn that season around? Is there any truth behind the assorted narratives presented in the media and by the players? Based on what we know from empirical research into cohesion in small groups, such as hockey teams, I am pretty confident that Tyler Bozak’s comment repeated in the tweet below is only half right.
With a big smile on his face Tyler Bozak said that team bonding and activities away from the rink helped the team.
— Shawn McKenzie (@ShawnMcKenzieSN) March 17, 2017
Social scientists commonly view cohesion as one of the most important elements of team dynamics and, ultimately, group success. Thousands of research articles mapping out the nuances of how cohesion works have already been written with about 500-600 new articles appearing in peer reviewed academic journals every year. About half of this research is done on teams in various sports including hockey. However, very little of this information appears to have trickled down to hockey management groups or into the public hockey analytics discussions that take place in social media. Teams typically take “I know it when I see it” and “gut feel” approaches to all psychosocial dynamics including cohesion, and while hockey analysts are sometimes curious about “intangibles” they typically default to sticking to what they can currently measure using public data. As a result, how teams and analysts talk about “intangibles” such as cohesion appears to be at odds with what empirical studies from academia are showing. This disconnect needs to be eliminated.
To this end, this article is a brief primer on existing empirical evidence on cohesion and how it may apply to hockey teams. It will focus on four main topics: (1) components of cohesion, (2) cohesion as a multilevel dynamic process, (3) negative effects of cohesion, and (4) cohesion and team success. Underlined words are hyperlinks that will take you to key academic articles that deal with the topic being discussed.
Components of Cohesion
Cohesion refers the bonds that form between members of a group. Groups marked with high level of cohesion share the following characteristics:
[Group members] possess a common identity, have common goals and objectives, share a common fate, exhibit structured patterns of interaction and modes of communication, hold common perceptions about group structure, are personally and instrumentally interdependent, reciprocate interpersonal attraction, and consider themselves to be a group.(Carron & Hausenblas, 1998, pp. 13-14)
Not all cohesion is the same, and over the past 30 years or so researchers have typically made a distinction between “social cohesion” and “task cohesion.” Social cohesion is about interpersonal relationships that lead to bonds between individuals on a team. An example of why this type of cohesion could matter to hockey teams is players may be more likely to sign or re-sign with a particular organization, or be willing to take a “home town discount,” if they feel personally attached to the group that is there. Players may also be more willing to ride out tough times when a team is not playing well, or if they dislike a particular head coach, if they feel connected to the other players or to the management group. Task cohesion (AKA instrumental cohesion) is more narrowly focused on building bonds through the process of working together to achieve a common objective. For NHL teams the most common objective is to win the Stanley Cup, but the discussion should not be limited to that. Other key objectives may include making the playoffs or simply improving from the previous season. On the ice social cohesion leads to players standing up for one another and supporting each other, while instrumental cohesion leads to things like buying into the system the coach has in place or filling a role that the team believes it needs.
When I presented on Rank Ordering the Intangibles at the recent Vancouver Hockey Analytics Conference I suggested that the question of whether a given intangible is a trait or a process is really important, and the answer to this question is almost always that it is a bit of both. Cohesion is no different in this regard, and although most of the literature focuses on team processes there is compelling evidence that some personality traits are linked with team cohesion and ensuing positive results. It may be a surprise to some who read this article, but there is evidence to suggest teams who add “glue guys” may be on the right track (although teams often look for the wrong characteristics, which is a different story). Researchers have consistently found that of the assorted personality traits agreeableness is typically the best predictor of team success. In a study that looked at agreeableness, cohesion, and performance across 107 project teams over the course of 4 months, Bradley, et al (2013) found that agreeableness was one of the strongest predictors of team performance while at the same time being one of the weakest predictors of individual performance. Having team members that help hold a team together, but are below average at what they do, is a common occurrence even outside of hockey.
A couple of very important caveats have to be made at this point. First off, the Bradley, et al, study focuses on group success in an office environment where success is measured by pumping out reports, and findings from this type of environment may not be directly transferable to elite hockey teams. Second, when hockey teams talk about “glue guys” they often refer to weaker players who are only in the lineup because they bring a particular “intangible” dimension to the team. It is impossible to assess whether the tradeoff the team is making is a good one without tracking results over time and using a comparative method to tease out how much being a “glue guy” brings to the table with respect to long term team success. Only having on ice statistics, or only having survey results (even if they are longitudinal), is not enough for an accurate assessment of whether the tradeoff between skill and intangibles is paying off.
Cohesion as a multilevel dynamic process
The core idea of cohesion, which is that interpersonal and instrumental bonds are an important component of how small groups function, is relatively straightforward. The story becomes far more complex when you take in account that cohesion is a multilevel dynamic process. In this case “multilevel” simply means that it works both on an individual level and group level, where individual factors shape the group while the group simultaneously influences individuals factors.
Upping the complexity of the individual/group dynamics is that people form interpersonal bonds that may not extend to every group member and often organize themselves into smaller subgroups or cliques. For this reason it is important to avoid looking only at the trust-building side of cohesion and to also include a “trust-inhibiting” element where two individuals just do not click with each other. One quick example of how this plays out can be found in research done by Lusher, Kremer, and Robins (2014) on Australian Football League teams. As can be seen by this side by side comparison of two of the teams, where the nodes are the players and the arcs are the trust bonds, the trust networks formed within each team is distinct:
We can see from the visualizations that Club A has a more robust trust network and is a more cohesive unit than Club B. Every team will have a slightly different configuration of trust relationships, but if this is done often enough certain higher order patterns will emerge and it is theoretically possible to identify which pattern yields the highest level of success, or to examine how larger group dynamics impact upon key individual-level behaviors. The practical problem is that it is not currently possible to get this data for even one team, let alone enough teams to begin to make generalizations.
Further adding to the level of difficulty in understanding how cohesion in hockey actually works is that it is also a dynamic process (AKA “emergent state”), which in this instance means that it changes over time. Chiocchio and Essiembre (2009) present a very strong evidence-based argument that it takes at least four weeks for cohesion to form at a level where it impacts upon group performance, while Siebold (2006) found that cohesion tends to be “volatile” in the early stages of a group’s existence while stabilizing over time. Tales of team fishing trips and fun days prior to winning streaks late in the season make good talking points for the press, but the impact of that bonding would not be immediately apparent and it is extremely unlikely that the bonding component led to whatever short term successes immediately followed. In those cases it is far more likely that the group outing helped by easing the cognitive load of players during a stressful time.
An inherent limitation for developing cohesion within a given professional hockey team is that that the teams will often be in flux as players are switched out as a result of new signings, trades, injuries, and retirements. Although teammates will sometimes train together during the off season for the most part a given team starts to bond during training camp and likely does not become a fully cohesive group until around the time of the Christmas break or later. Of course having a stable core in place that act as anchors for the new players impacts upon how long this process takes. Regardless, the main takeaway from cohesion research should be that group building exercises such as fishing trips and skating getaways are best done early in the season in order to provide adequate time for meaningful trust bonds to form and have an impact upon achieving team goals.
Negative effects of cohesion
Although most discussions of cohesion and other psychosocial factors focus on the positive impacts and outcomes the negative side of the equation also needs to be considered. When I was a young hockey fan in Winnipeg the Jets drafted and brought in Alexei Zhamnov who was a wizard at puck handling and tried to gain the blue line with possession. The team captain at the time was Chris King, who said during an interview that Zhamnov has to learn to fit in and play the North American game which meant that he should learn to dump the puck in the corner. I will never forget how King characterized Zhamnov’s “struggles” in an interview: “he has to learn to trust his teammates and that we will go in and get the puck back for him.” Zhamnov learned to be a part of the group and dump the puck in, but given what we know today about the relative benefits of gaining the blue line with possession, changing his play in order to fit in with the group probably hurt rather than helped both his game and the team. This process is described in the literature as conformity, where people subsume parts of their identity in order to fit in with the group. The problem is that sometimes there is a pressure to conform to doing things that run counter to team success.
Hardy, Eys & Carron (2005) interviewed 105 athletes from across different levels and asked them whether they experienced negative effects of social or instrumental cohesion and, if so, what those effects were. Negative effects of social cohesion were much more common than those associated with instrumental cohesion, with more than half of the respondents identifying at least one negative effect. When mapped out into themes the negative effects looked like this:
Time wasting refers to things like joking around during practices to the point where the goals of the practice are not adequately met, goal-related problems refers to it sometimes being more difficult to get a buddy to buckle down and focus on a team related goal, communication problems refer to the difficulties associated with giving constructive criticism to someone you are close to, decreased focus means members of teams with high social cohesion sometimes focus on socializing with teammates to the point where it negatively impacts on performance, social isolation refers to the development of cliques within the team and the exclusion of people who do not “party” with the rest of the team, and social attachment problems refers to how teammates often spend a lot of time together as a result of games, practices, and travel, and those who are close to each other spend even more time together and can get “sick of each other” a lot faster. The striking thing about this list is that it really highlights that athletes live and experience hockey often in the same ways that we all live and experience our work.
Cohesion and team success
Existing research typically (actually, always) shows that task cohesion is more closely related to team success than social cohesion is. The key takeaway for hockey teams is that it is important to distinguish between the type of cohesion that makes players want to sign with your organization or improve team culture and the type that leads to winning more. Teams may want “glue guys” who bring leadership and help the overall team atmosphere. That is great, but the “glue guy” the team has may help social cohesion rather than task cohesion which means he is not suitable for a team that is a serious Stanley Cup contender. Conversely, a player that does not take a home town discount and does not hang out and party with his teammates may actually be excellent in fostering instrumental cohesion that actually helps a team win. Understanding how different types of cohesion work and impact on performance can help an organization fine tune their process to meet specific goals. However, I am pretty confident in saying that teams typically rely on personal assessments of players that are rooted in experience and “common wisdom” (which typically devolves into player reputation) rather than looking at understanding existing empirical evidence.
The last question I want to consider is something that comes up on social media from time to time. It is generally understood that winning leads to cohesion within teams. Some, usually in the analytics community, argue that the sense of unity experienced by players on successful teams stems directly from winning, with the point that follows being that cohesion within a team does not matter as much as NHL teams often think. There are a limited number of longitudinal studies that directly address the degree to which cohesion impacts upon winning and vice versa (often looking at performance if they are not based on sports teams) and they all tell the same general story. First off, the relationship between cohesion and performance is statistically significant, as is the relationship between performance and cohesion. This makes sense because bonding helps you perform better (or win) while performing well (or winning) helps you bond. However, when looking at effect sizes cohesion typically has almost twice as much of an impact upon success as success has upon cohesion. One recent example of this type of research is Mathieu, et al (2015) who looked at leadership, cohesion, and performance in student teams:
Once again, there may be important differences between student or work teams and professional hockey teams. My best guess is that the story will not change but the numbers might. In other words I would be shocked if cohesion did not impact on winning more than winning impacted on performance, but I am very open to the idea that in the context of professional hockey the effect sizes may not be as far apart as they are for non-sports work teams.
At the beginning of this article I referred to Tyler Bozak responding to criticism about his team going on a fishing trip in the middle of the playoff race by saying “team bonding and activities away from the rink helped the team.” Being a professional hockey player, particularly one that plays in market like Toronto, is a grind and I am sure the pressure can become oppressive at times. The team bonding activity probably helped relieve some of the stress and cognitive fatigue associated with being in a high pressure situation in a high pressure market. However, team bonding activities are a slow brew in terms of impact and are unlikely to directly lead to a big and immediate payoff.
While I am absolutely sure that team cohesion is a very large factor in the success or failure in professional hockey teams, and that hockey organizations often fail in their selection of “glue guys” that have specific interpersonal qualities they believe are important, I really have a hard time wrapping my head around how meaningful data can be collected when working as an analyst for a specific hockey team. Empirical studies into cohesion shed a great deal of light into what works in terms of helping boost performance, but such studies focus on multilevel dynamic processes. To do this, researchers often focus on dozens, if not hundreds, of teams and measure key elements of cohesion and success over a period of time. It work take including every team in the NHL just to begin to have a baseline of how cohesion specifically works in the NHL.
This should not be taken as an argument that cohesion is not important to NHL teams, because it is. To draw examples from other sciences, diseases existed before medicine was able to directly measure their effects on human cells and the billions of stars in the universe were there when the humans still thought the sun revolved around the Earth. The main takeaway from this article should be that we know a great deal about small team cohesion from existing research outside of hockey, and that robust base of knowledge can be applied to help understand what is happens in the context of professional hockey teams. Data we collect from teams should be directed toward helping us understand how a given team fits into the big picture rather than to establish what that big picture is.