I Trust You, You Trust Me: understanding the effects of trust on team performance.

trust youAs noted in a previous blog post, the dominant view amongst scholars is that team trust is beneficial for team performance. Trust helps team members to suspend their uncertainty and vulnerability in relation to teammates and allows them to work more effectively and efficiently, using their energy and resources in ways that contribute to team performance.

A lack of trust results in team members losing sight of team goals and focusing instead on their own personal interests, priming defensive actions aimed at protecting themselves against possible harm by others. Well, that’s the theory in any case.

The reality is much more complex. For example, the meta-analysis by De Jong and colleagues found that the effects of team trust on team performance were, on average, positive (.30), but the variation across studies ranged from negative to positive. This suggests that lower levels of team trust may facilitate better team performance in some situations.

This is nicely illustrated in an experimental study by Thommes and Uitdewilligen (2019). Thommes and Uitdewilligen begin by reminding us that temporary teams (e.g., crisis management teams, medical teams, firefighters) are increasingly relied upon in modern society to help us deal with uncertain, challenging environments. These temporary teams are comprised of skilled experts working in an interdependent way for a finite time. In a life or death situation, having a finite time to decide and act is a challenge. Information sharing, coordination, and cooperation are critical for these teams.

As such, one might assume that high levels of team trust would support the key positive exchanges that are needed. However, as noted by Thommes and Uitdewilligen, while this might be the case in established teams that have developed trust over time—trust that is grounded in a history of learning together—in newly formed teams, high levels of trust that are imported as a given (i.e., ‘high swift trust’) may be dangerous. This is particularly the case if people simply assume that their team members are capable, reliable, and dependable, without really putting this to the test.

Learning to trust one another in a truly functional way implies pushing and pulling one another in various ways. For example, when sharing and coordinating information, team members might naturally prompt others for detailed elaborations and reinforce a systematic approach to information integration. Established teams may do this as part of their regular practice, but newly established teams might be ‘overly trusting’ and not push other team members in this way, particularly if their competence is accepted as a ‘given’.

A slightly lower level of trust might be more useful in this context, as it can prompt team members to elaborate and be more systematic when sharing and integrating information that is critical for optimal team performance.

This is exactly what Thommes and Uitdewilligen found. In a fascinating experiment, they assigned 40 teams to work on an Emergency Management Simulation (MUEMS). Team members were placed in the role of first-responders working together to minimize the impact of fires. Each team member was assigned to different roles, either as a fire commander, a chemical specialist, or a police officer. Each team member also had unique knowledge and expertise, for example, about the number of available fire trucks in each fire station (the fire commander), which roads to block in order to get to the fire stations (the police officer), and the risk of other buildings catching fire (the chemical specialist).

Each expert could calculate the cost of their individual decisions, but the distribution of knowledge across the whole group was unknown to team members prior to interaction. While each team member received training in their own role, when working together they needed to weigh and balance their various decisions to minimize costs and maximize overall team performance.

After initial training in routine scenarios, every team worked on four scenarios during the critical experimental performance phase. They worked on two routine scenarios and two non-routine scenarios. While the routine scenarios were similar to the training tasks previously practiced, the non-routine scenarios included some novelty (e.g., a bomb threat), which required new priorities and strategies to perform well on the task.

The key experimental manipulation involved activating either the ‘high swift trust’ or ‘low swift trust’ algorithm in teams. In the high trust condition, team members were told that the performance of others in previous tasks provided a good basis for the upcoming team tasks. In the low trust condition, participants were told that their own performance provided a good basis for the upcoming team tasks, BUT at least one of their team members had previously made multiple mistakes and thus may have not fully grasped their role and responsibilities.

Team interactions were recorded and coded to generate a measure of team information processing—that is, the extent to which information was (a) disseminated, (b) integrated, and (c) generated to make decisions.

Information dissemination involved sharing simple facts or decision-making without providing much context or reasoning. Information integration involved team members building upon each other’s statements (e.g., to prioritize or contextualize information). Information generation involved not only integration of team member inputs, but also the identification of problems and the generation of new ideas and strategies to address them. Naturally, while some form of ‘information processing’ is essential for team performance, Thommes and Uitdewilligen weighted team information processing scores to reflect the qualitatively different levels of information processing involved, with dissemination weighted (×1), integration (×2), and generation (×3).

The key findings of this experiment are very interesting (and understandable):

(1) Teams primed with low levels of trust engaged in significantly more information processing than did teams primed with high levels of trust. If you don’t fully trust team members, you push for more elaboration, integration, and generation to maximize the quality of information processing in the team.

(2) When the effects of information processing on team performance were examined, it was found that information processing had no effect on team performance for routine tasks. However, higher levels of information processing predicted better team performance for non-routine tasks. More elaboration, integration, and generation predicted better performance in the non-routine scenarios.

As such, we shouldn’t make any assumptions about what we mean by ‘trust’ in a teamwork context. As noted by Thommes and Uitdewilligen, trust-building activities are often recommended to enhance team performance. But trust dynamics influence performance dynamics in different ways across different contexts. As noted in a previous blog post, the effects of team trust on team performance show massive variation across studies – sometimes positive and sometimes negative. To my mind, the variation has always been more interesting than anything else, and we should pay heed to this variation when working with teams.

When temporary teams (e.g., crisis management teams, medical teams, firefighters) are working on non-routine tasks and when different team members have complementary knowledge and skills that need to be coordinated, and when these coordinations can be optimized to produce the most effective (and cost-effective) solutions, prompting team members with a ‘healthy suspicion’ might be valuable.

In order to see this as a team facilitator, you need to monitor and track the dynamics in a group. If team members are primed and prompted in a way that supports a ‘healthy suspicion’ when suspicion is warranted, they may expend more effort sharing and integrating information and generating new ideas in order to develop better solutions. High levels of trust and lower suspicion may arise over time, AND be coupled with sustained high-quality information processing in teams – but this may take years of working together closely with team members.

However, there is a danger in our move to more mobile, temporary, team-based working arrangements that some teams will fall foul to ‘high swift trust’. If team members wholeheartedly embrace every utterance of their expert teammates, without seeking to elaborate, integrate, and generate new ideas, the team as a whole may fall foul to a lack of methodical creativity in those critical situations where they come face-to-face with the non-routine scenarios. We must expect the unexpected, and these non-routine scenarios are more prevalent than we might think. We can never fall asleep at the wheel.


Featured Paper:

Thommes, M. S., & Uitdewilligen, S. (2019). Healthy suspicion: The value of low swift trust for information processing and performance of temporary teams. Group Dynamics: Theory, Research, and Practice, 23(2), 124-139.