Abstract:: In this review, we examine the burgeoning body of research on multiteam systems (MTSs) since the introduction of the concept in 2001. MTSs refer to networks of interdependent teams that coordinate at some level to achieve proximal and distal goals. We summarize MTS findings around three core processes and states: within- and between-team coordination processes/structures, leadership structures/processes, and cognitive and affective/motivation emergent states. Furthermore, we explore how these processes and states vary according to MTS boundary status (internal or external), component team distance (geographic, functional, cultural, and discipline), and superordinate goal type (intellectual or physical). We identify several process and state similarities across levels of these attributes, as well as highlight some important differences. We conclude with a set of propositions and future directions prompted by our review, which can serve as a guide for future MTS research.
Notes
Intro + Structure & Methodology
"MTS success is defined as the accomplishment of that distal goal" (p. 480) quotes
Earlier reviews didn't find variation in MTS attributes. Neither did this one, but review uses 3 for a lens (these are the input variables they look at)
Boundary status (internal/external)
Component team distance (CTD) (geographic, functional, cultural, discipline)
(Coded by how many of these types of diversity they exhibit)
"pattern of transition/planning processes, action/execution processes, and interpersonal processes" (p. 480, citing Marks et al. 2001) quote
Use Devine's (2002) classification for intellectual vs. physical goal types:
physical = "involves a core task, physical skills, a linear work-flow, applying existing knowledge, and a tangible product" (Devine, 2002, p. 296) definition
intellectual = "thinking as a core task, mental skills, a nonlinear work process, the derivation of new knowledge, and information as the primary work outcome" (Devine, 2002, p. 296) definition
Note - the nonlinear work process may be a key point for collaborative knowledge work.
Methods - review found 115 empirical papers. Table 1 has boundary status, task type, and CTD delineation
MTS Coordination Processes & Structures
Underlying phenomenon of MTSs: goal hierarchies, performance episodes, within- and between-team coordination (Mathieu et al., 2001)
They didn't say much about processes, but did say that "collective goals facilitate two critical processes:...action plans.. & collective coordination" (Mathieu et al., 2001, p. 298)
These are what were labeled by Marks et al. (2001) as transition and action processes
{{TODO}} Mathieu et al. (2018) gave classification of coordination elements: functions, foci, forms, phases. (Never read this!!!! Must remedy ASAP!)
coordination functions: "cognitive and behavioral processes that accompany component teams' goal accomplishment"
coordination foci: the focus or target of the coordination effort. Within a team, across to a different team, across MTS
coordination form: Who engages in coordination efforts. (Leaders, boundary spanners), and the structure of the coordination enactment (centralized vs. decentralized)
Also - this might be more mixed in H-M teams, or have tech aid leaders
coordination phase: temporal patterns and shifts as coordination is carried out
Marks et al. (2005) had found that when MTSs developed cross-team coordination plans, they lost the flexibility that is necessary with tasks that require intensive interdependence - which are complex and uncertain.
This may be relevant for H-M teams - since machines are not as good at adaptation as humans
So the solution for H-M teams that struggle with adaptation may not be clearer coordination plans
Without that flexibility, their plans (from transition phase) were not aligned with what they were doing (action phase) - so they need within and between team alignment planning, and anticipatory adaptation, to be effective as an MTS.
These will help teams balance own needs against needs of MTS.
How doH-M teamsdevelop contingency planning? H-M literature does program machines to prioritize robots to get along with humans, because long-term that's better. But that's not yet MTS focused.
Marks et al. (2005) also said need MTS processes to let teams change plans "on the fly" ("reactive strategy adjustment" Marks et al., 2001) if there are disruptions or coordination failures in conditions of intensive interdependence. This was even in low CTD, physical MTSs - this review suggests that is even more true in intellectual MTSs where information complexity is higher; and in external MTSs with high CTD and therefore stronger differentiation.
So in high CTD, intellectual MTSs it is likely critical to have the MTS process of reactive strategy adjustment.
In high CTD, intellectual MTSs where there are also H-M teams, and machines struggle with adaptation, how will those processes be carried out? MTS-AI_challenge
Note - early research didn't focus enough on within team alignment and adaptation processes - (e.g., team members knowing what the between-team interdependence needs were)
Uitdewilligen & Waller (2012) found in their study of external, physical MTSs with high CTD, that MTS adaptation needed teams sharing info about changing situation, and a core team (reps from all other teams) engage in forecasting and contingency planning; and that a key factor was the core team switching "between mono- and multidisciplinary reasoning" (Uitdewilligen & Waller, 2012, p. 386) - that the core team needed to think about how each team would see the situation and the adjustments necessary.
This sounds like perspective-taking, in fancy language
Can machines engage in this? How effectively could such core leaders engage in perspective-taking for machines? Communicate with them, and tell them to realign? MTS-AI_challenge
A big part of MTS Adaptation is learning from AARs - although this isn't mentioned much in previous conceptual frameworks.
(Uitdewilling & Waller (2012) say this is particularly important for temp crisis MTSs - parent organizations need to learn, so they can teach future temp MTSs responding to similar crises)
MTS coordination structures - the SNA (Social Network Analysis) perspective of who works with who
Centralized seems more instrumental (both during action & transition phases) in simpler (internal, low CTD, physical) MTSs.
List a lot of studies that found dif. things - gets a little hard to follow the pattern (that's part of their point). First batch of studies here, all physical.
Davison et al. (2012) - horizontal within team, vertical between team positive for MTS performance; horizontal between team negative for performance.
Lanaj et al. (2013) found that decentralized planning was better for motivation and from there MTS performance; but also had excessive risk-taking and higher rates of action phase coordination failures (hurting overall performance).
Other studies also found horizontal planning helpful for performance in some conditions: high intrapersonal functional diversity (team members know a lot of dif things, are functionally broad, when paired with:
vertical coordination with lead team (De Vries et al., 2016)
lack of leadership or integration team (Pilny et al., 2014)
Note - this seems to conflict with above
especially in larger MTS (Schecter 2017)
Pilny et al. (2014), between-team coordination most important when inaccurate info comes up
For coordination structures in high complexity physical MTSs (external and/or high CTD):
centralized is effective for workflow management (frequent comm from lead to all teams, and between all teams) (O'Sullivan, 2003)
Low between-team, with low centralization = low info flow, hurting between-team sensemaking, and inhibiting critical communication transmission (Schipper & Gerrits, 2017)
Other studies with similar results (Alison et al. 2015, Fodor & Flestea 2016)
Contextual factors and performance phase can influence relationship between communication structures and MTS performance:
More communication among teams necessary in design phase (transition phase), where it mitigated inaccuracy and incompleteness of teams that were not well synchronized (Xie et al. 2010)
(in other words - if don't have clear plans, have to keep talking explicitly to make sure things get done)
Intellectual MTSs, coordination structure patterns, and MTS effectiveness:
Moderately dense good for idea generation (transition phase), less during implementation phase (action phase) Kratzner et al. (2008)
Also, comparing expected interaction to informal communication patterns - inverse U shape between misalignment and MTS creativity - frequent interteam communication helps creativity by making sure info gets shared; but too much "suppress(es) individual autonomy, criticism, and leads to a collective distraction and creativity blocking" (Kratzner et al. 2008, p. 545).
me Is this also true if tech handles a lot of this misalignment, bypassing the need for explicit communication? tech_helps
Depends on boundary status - so need lots of between-team interactions for project success, but more for external MTSs, and more as project progressed through its stages (Lavallée & Robillard, 2018)
(note - external MTS, no clear leadership team)
Communication changes from horizontal to vertical when a proximal goal is ambiguous (Scheere & Kude, 2014)
(note - internal MTS, leadership team)
Suggestion: No leadership and external, makes need denser communication structure when project gets disrupted.
Summary of coordination processes and structures
Across MTS types and forms, transition/action/interpersonal processes predict success.
strategic models = shared understanding of MTS mission, and links between what the teams are capable of and accomplishing the distal goal (Mathieu et al. 2001)
Detail:
Shared interaction mental models are important for all types of MTSs. Examples:
MTS cognitive models should have accurate problem models, and shared understanding of how teams are connected and respond to different situations (Anania et al. 2017)
Could build a library of potential disconnects, perhaps even in direct communication between people coming from different functional backgrounds, so that the tech (AI here, really) can predict when something might get misinterpreted, and highlight that and give what the term means to the person sending the message.
In a group meeting, it can listen in and pop up the meaning of difficult words or acronyms, or yin-yang words that could mean different things to different people there
Likely to be more important to have transactive memory systems in external MTSs, especially those with high CTD, where are likely to have different org knowledge, functional expertise, disciplinary expertise
(Studies where TMS were important were almost all external and moderate to high CTD: Caldwell 2005, Healey et al. 2009, Henry et al. 2016, Liang & Jin 2010)
Physical vs. intellectual MTSs - will need different cognitive models:
Intellectual, these strategic models are about a shared understanding of the abstract problem space and its idea elements (they're problem models); and the shared interaction mental models are about which team works on which part of the problem, and the processes used to combine information. (Plus obv. the transactive memory of who is expert in what.)
So the tech to support shared mental models would need to be very different for physical and intellectual MTSs. Obv.
Affective/Motivational States
Overview:
goal commitment = shared willingness of team to allocate effort towards distal goal (DeChurch & Zaccaro 2010)
MTS-AI_challenge This will mean something different for machines. How to program them? - support what the team desires, or what's best for the MTS distal goal? Which of these is more important?
cohesion = Strength of bonds holding team to MTS (Gross & Martin 1952)
collective efficacy or efficacy = shared confidence that teams can work together and achieve distal goal (Bandura 1997)
trust = Shared, felt reliability that interactions and behaviors will occur. Especially in ambiguous or threatening situations (Lewicki & Bunker 1996)
psychological safety = climate that promotes willingness to speak up within the collective (Edmondson 2003)
internal, moderate CTD, physical MTS linked to positive outcomes *Hoegl et al. 2004)
external, high CTD, intellectual MTS - willingness to exhibit / grant leadership to others; and that focusing on team goals over MTS hurt overall performance (Carter, 2016)
training together increased task cohesion and overall goal commitment (McGuire, 2016).)
Group boundaries around team and MTS will each foster identity. Develop multilevel identification with both team & MTS
can be challenging for external, high CTD MTSs where members more likely to identify with team than with MTS
MTS identity, or team identity?
Lack of identity with MTS hurts effectiveness - "particularly by inhibiting information sharing and communication processes and/or increasing between-team conflict (e.g.,, Cianciolo & DeCostanza 2012, Cuijpers et al. 2016, Wijnmaalen et al., 2018)" (p. 492 quote)
On the other hand - study with internal, low CTD, physical MTS argued that identifying with MTS increased ambiguity and uncertainty for component teams; thereby raising cognitive loads and using resources otherwise needed for goal attainment; thereby hurting MTS performance. Especially when task complexity was high. (Porck et al. 2019)
tech_helps If lack of identity hurts MTS effectiveness through mediators, can tech help address those mediators? - it sounds like it's helpful to have team-level identity, but that it hurts via mediators.
In other words - team vs. MTS identity tension leads to countervailing forces (DeChurch & Zaccaro, 2013)
Could build a library of potential disconnects, perhaps even in direct communication between people coming from different functional backgrounds, so that the tech (AI here, really) can predict when something might get misinterpreted, and highlight that and give what the term means to the person sending the message.
In a group meeting, it can listen in and pop up the meaning of difficult words or acronyms, or yin-yang words that could mean different things to different people there
But having MTS identity, might deplete team resources, hurting MTS performance
(Again -
tech_helps If lack of identity hurts MTS effectiveness through mediators, can tech help address those mediators? - it sounds like it's helpful to have team-level identity, but that it hurts via mediators.
So this needs more research to figure out the best balance of team vs. MTS identity across MTS types.
Practical Implications and Recommendations
MTSs need to map their MTS more effectively. Tetrick et al. (2016) have a tool for much of this. But an MTS-TA should add a few more elements: key_element
level of CTD among teams
physical vs. intellectual work within and between teams (can vary within MTS)
MTS mapping identify "nexuses and variance in levels of interdependence among component teams" (pg. 493) - which gives focus for MTS intervention: MTS charters and MTS building
Conclusions, Future Directions
Have a set of recommendations - but they're tenacious, because many studies don't provide key info about their MTS! Some things they should include: MTS-TAkey_element
size of MTS - # of teams and members
boundary status, including # of teams from external orgs
goal type (intellectual vs. physical)
characteristics relevant to diversity or differentiation (geographic, functional, cultural, discipline)
(Note - discussion with Elisa on December 4th, 2020, these were just representative of things that matter.)
No one is looking at appointed vs. emergent - but it should matter!
quote "For example, emergent MTS teams may either be compelled to work together due to an impending or recent crisis or may choose to partner to accomplish a jointly desired goal" (p. 495)
Used old classification of teams, physical vs. intellectual from Devine (2002). Need a taxonomy of MTS goal or task types. MTSresearch_gap
We know communication needs change in response to event severity, and in response to different types of events or situations. There isn't a good classification of situations or events - multiple fields are calling for this. MTS research needs it to. MTSresearch_gap
Need to think about the network of interdependence, and its effect of emergent states.
The same pattern can mean different things for AI-human tie vs. human-human tie MTS-AI_challenge
Might even need to redefine some piece of interdependence types for MTSs with H-M teams. (Although more for intensive than for sequential or reciprocal.)
Leadership transitions may drive MTS effectiveness in dynamic environments - but we don't know much. Maybe some mental models for when the transition should happen, others for how to do shared leadership.