Friday, November 6, 2009

Psychological Safety in Work Teams: Is it Group Trust?

 

In her 1999 article, Psychological Safety and Learning Behavior in Work Teams, Amy Edmonson introduces the construct of team psychological safety, which she defines as “a shared belief held by members of a team that the team is safe for interpersonal risk taking.”  She defines interpersonal risk-taking as, “a sense of confidence that others will not embarrass, reject or punish someone for speaking up.”  What differs psychological safety from trust, as it is defined in the literature, is that the former is an inter-group phenomenon, and the latter is (in the literature) interpersonal.   But the definitions and mechanisms described are quite similar. 

 

Rousseau et al. (1998) argue most scholars can agree on a fundamental definition of trust. They noted that "confident expectations" and a "willingness to be vulnerable" are critical components of most definitions of trust (Morrow, Hansen, Pearson, 2004). Consistent with this view, Morrow, Hansen, Pearson (2004) defined trust as the extent to which one believes that others will not act to exploit one's vulnerabilities (Barney and Hansen, 1994; Mayer et al., 1995; McAllister, 1995). 

 

McAllister (1995), for example, explores the nature and functioning of relationships of interpersonal trust among managers and professionals in organizations, the factors influencing trust's development, and shows a positive effect on performance when trust was present in these relationships in an empirical study.  Building on that and other research, Morrow, Hansen, Pearson (2004) argue that trust evolves from a pattern of careful, rational thinking (cognitive-based), coupled with an examination of one's feelings, instincts and intuition (affect-based). 

 

Similarly, Edmonson (1999) writes that a group “will conclude that making a mistake does not lead to rejection when they have had a team experience in which appreciation and interest are expressed in response to discussion of their own and others’ mistakes.” Such shared experiences, occurring over time, create the tacit belief that that the team is a psychologically safe place. The phenomenon Edmonson (1999) describes as psychological safety could arguably be called “group trust,” which would be a development in the trust literature, but it may also be more than that. 

 

Edmonson (1999) observes that psychological safety leads to improved performance through enhanced learning, not necessarily through the combination of individual trust and/or affect.  In her research, learning behavior mediates between team psychological safety and team performance.

 

Her results support an integrative perspective in which context support/team leader coaching and shared beliefs together shape team performance outcomes.  As the group feels more confident to share information, more information is exchanged and therefore all collective information is more likely to be shared.

 

Edmonson (1999) Psychological Safety Model:

Cognitive reminders to encourage sharing context/coaching ->  Multiple experiences of sharing without consequence -> Psychological safety -> Willingness to share -> Inter-group learning -> Improved performance

 

Trust literatures generally describe the benefits of cognitive trust as increased voluntary collaboration and affective trust as increased information sharing (McAllister,1995). (Note also that trust in economics literatures has also been found through empirical studies to reduce transaction costs by avoiding costly negotiations and contracting (Dyer, 1997; Sako, 1992)).

 

McAllister (1995), Morrow, Hansen, Pearson (2004) Trust Model:

Interactions and cues with other agent provide data, which are processed and the output is:

 

1) Careful, rational thought about the competence of the other and structures in place to protect the self-> Cognitive-based trust -> Trust the evaluated person and actively engage in collaborative work and seek knowledge from him/her

 

2) An examination of one's feelings, instincts and intuitions about the interactions -> Affect-based trust -> Open sharing of sensitive personal information, knowledge sharing, idea sharing

 

Edmonson’s psychological safety model, it could be argued, describes a method for the creation of affective trust, which therefore encourages information sharing, which then leads to increased group performance.  Through that analogy, one may argue that missing from her model is the cognitive-based trust mechanisms.  However, this may be a feature of interpersonal trust which is not able to be generalized when moving from micro to meso trust.  It would be interesting to observe groups that employ psychological safety pre and post employment.  Will there be increases in levels of cognitive trust, which would manifest itself in increased collaboration, in addition to the increased information sharing?

 

Based on Edmonson’s work, the UVAHS could benefit from employing psychological safety into its groups.  We would like to propose a grounded study in which we and observe and report on current group interaction, introduce Edmonson’s psychological safety methods, and observe to see if improvements occur and (if they do) whether the improvements come from increased learning through affect, and if we are able to observe any improvements in cognitive trust as well.  If we observe measures of both cognitive and affective in a collective “group trust” we may bring the field one step closer to solving the mystery of local versus generalized trust as posed by Kramer and Cook (2004) – why do people generally trust their doctor, but not doctors in general?  Or, in the context of one of the health system’s problems, “Why do doctors trust their department, but not the health system in general?”

12 comments:

  1. Very well thought out and really well written. I really liked how your framed this... what are the psychological safety methods that you would employ in the health care scenario? I think that the UVA health system's goal is to achieve trust and psychological safety.. so how does this get us there?

    ReplyDelete
  2. I agree with Kelly - I would have liked a bit more about your methods, but overall it is a great way to connect. However, ever since that public trust summit, I've been thinking about your last sentences. I don't actually think people trust their doctors anymore - I think they just say they do. If you measured how fast and how many times people go on WebMD.com after their appts, you could potentially take that as a lack of trust.

    ReplyDelete
  3. Great job framing this discussion(probably a bit too much for a blog, but definitely reusable in your paper, so well worth it!) I think you may have a hard time measuring "trust" through qualitative methods. I have been challenged myself to measure "collaboration", and that should be more readily apparent behavior (volunteering, idea generation, suggestions, etc.) You'll have to think hard about what might be a proxy for trust (or distrust). It might also be very difficult for you to distinguish between distrust and procedure. For example, what appear to be CYA behaviors might actually be normal protocol for health workers.

    ReplyDelete
  4. I agree with Megan's assessment! Great launching pad for a future paper; and it was a pretty sophisticated conceptual piece too, I might add.

    One project could be to really review the trust/safety/group cohesion literature and really try to tease out what things mean conceptually. There might be something new that you could add! Or, better yet, there might be some endogeneity or even an untapped latent construct to be found.

    Another would be something empirical and would involve PCA or even a factor analysis, although the claims of factor analytic results can sometimes be overstated (the math only takes us as far as the 'meaning' allows us to take it which is embedded in language).

    Great work! I REALLY hope you explore this and would love to help out in any way I can. Esp regarding group cohesion etc.

    ReplyDelete
  5. Great suggestions. I am really interested in talking to you guys about how one could design quantifiable metrics to measure the effects of implementing psychological safety. The "CYA" issue is a huge one - especially in highly bureaucratic organizations like a large health system (insert reference here). One idea that occurred to me is an anonymous drop box. One obvious thing we will can can track is performance along the lines that they are already being tracked. But we need to be able to tie those data to psychological safety and (ideally) get to affect so we can see if affect is potentially a group-level emotion and not just one-to-one, or independent of one-to-one levels of affect. Thanks for your help...

    ReplyDelete
  6. those are in many ways my thoughts in note form so I don't forget what to ask you about... If they are hard to follow please forgive me!

    ReplyDelete
  7. In an economic model, a parameter or variable is said to be endogenous when there is a correlation between the parameter or variable and the error term.

    ReplyDelete
  8. For starters we should probably ask Ryan about using sem to look at psychological safety within organizations.

    I'm really not sure on this because I haven't done work (yet) with sem, but my understanding is that we can tease out some non-individual (i.e., group) phenomena with a single data set. We can then look at different nested levels within the organization by looking, essentially, at GROUP rather than individual level variables.

    ReplyDelete
  9. What do you mean by pre and post employment? I’m not sure I see the link between introducing Edmonson’s psychological safety methods to UVAHS and bringing the field one step closer to solving the mystery of local vs. generalized trust. My theme of the week, perhaps the use of a concrete example showing how you see this link hypothetically being played out may have been helpful.

    ReplyDelete
  10. It seems like you are proposing to do too much (solve UVAHS employee identity crisis) with one tool: "Psychological Safety". What evidences suggest that UVAHS employees are lacking in psychological safety? I would rather have a qualitative study done to find out the approximate reasons of UVAHS employees' lack of identity or belongingness first before suggesting how to improve their lot.

    ReplyDelete
  11. To Brendan: pre and post employment of psychological trust... sorry for that vague reference.

    To Ishrat: We know they aren't collaborating, and we know they are pretty nasty to each other over there. I doubt it would solve everything, but it is a start potentially and the process could be a good source of data.

    ReplyDelete
  12. Good post and good conversation. I really like the conceptual work you've put in here.

    Having said that, I tend to agree with Jesse w/r/t research methods. When you are dealing with two constructs/measures as well established as trust and psych. safety, and you really want to tease them apart, I would probably engage a paper that is more about conceptual development and quantitative measurement than about inductive research. Qualitative research could be useful, but it would be more useful to push thinking in these areas to a "next level," rather than to clarify.

    ReplyDelete