Content uploaded by Alex Ning Li
Author content
All content in this area was uploaded by Alex Ning Li on Jul 14, 2021
Content may be subject to copyright.
1
HOW VOICE HELPS TEAMS COPE WITH EXOGENOUS CHANGE: DIFFERENTIAL
EFFECTS OF PROHIBITIVE AND PROMOTIVE VOICE
Alex Ning Li
Department of Management and Leadership
Texas Christian University
Subrahmaniam Tangirala
Department of Management and Organization
University of Maryland
Alex Ning Li https://orcid.org/0000-0001-6821-6332
Correspondence concerning this article should be addressed to Alex Ning Li, Department of
Management and Leadership, Neeley School of Business, Texas Christian University, Dan D.
Rogers Hall, Fortworth, TX 76109, United States. Email: ning.li@tcu.edu
2
ABSTRACT
Teams often confront exogenous events that induce discontinuous change and unsettle
existing routines. In the immediate aftermath of such events (the disruption stage), teams
experience a dip in their performance and only over time regain their previous performance
levels (in the recovery stage). We argue that prohibitive voice that allows teams to manage errors
better is instrumental for preventing performance losses in the disruption stage. Whereas,
promotive voice that helps teams innovate or improve team processes, can facilitate steeper and
more positive performance trajectories in the recovery stage. We also propose that voice is
especially functional when teams confront higher change intensity and, thereby, highlight that
voice is particularly important when change events cause greater discontinuity in the task
environment. We found general support for our theory in a correlational field study involving
172 production teams in which we examined over time trajectories in objective team
performance, and a field experiment involving 88 teams in a different production setting, where
team members were trained, incentivized, and provided opportunities to engage in voice. We
discuss the implications of our findings for literatures on voice, team adaptation and resilience.
Keywords: employee voice, team resilience, temporal dynamics, team adaptation
3
Teams need to adapt to external disruptions such as shifting consumer demand, changing
governmental regulations, mergers or acquisitions, and new technology (Reeves & Deimler,
2011; Schweiger & DeNisi, 1991; Stouten, Rousseau, & De Cremer, 2018). However, in the face
of such environmental turbulence, many teams fail to properly to reorganize or modify their
internal processes to effectively meet their new performance challenges (Alliger, Cerasoli,
Tannenbaum, & Vessey, 2015; Maynard, Kennedy, & Sommer, 2015; Stoverink, Kirkman,
Mistry, & Rosen, 2020). Scholars have referred to this as a “resilience gap” in organizations
(Hamel & Valikangas, 2003) that needs to be understood and examined.
Specifically, when adapting to exogenous change, teams deal with two key challenges
(DevRaj & Jiang, 2019; Hale, Ployhart & Shepard, 2016): (a) Teams have to stem sharp dips in
performance that typically occur in the immediate aftermath of the change event (disruption
stage) because they make errors in the unfamiliar performance environment, and (b) they need to
show an over-time improvement in performance (recovery stage) by implementing innovations
that allow them to thrive in the changed environment. Hence, resilient teams are those that
contain initial disruptive consequences of change on team performance and recover to
demonstrate steeper positive trajectories of team performance post-change.
We propose that voice—or members’ expression of ideas or concerns that constructively
challenge the status-quo (Van Dyne & LePine, 1998)—is critical for teams to demonstrate such
resilience. Through voice, members draw on their privately-held knowledge and make it public
in a manner that allows their teams to take action (Morrison, 2011). Hence, voice is crucial to
ensuring that the competence and expertise of members is put to appropriate use within teams
(Sherf, Sinha, Tangirala & Awasty, 2018). When teams have to cope with disruptions that create
novel performance environments, we argue that it becomes especially important for them to draw
4
on their members’ knowledge and insights to stem disorder caused by such disruptions as well as
innovate to thrive in the post-change environment. Hence, voice, we note, becomes an important
and proximal behavioral input that determines team success in times of change.
Here, we distinguish between prohibitive and promotive forms of voice (Liang, Farh, &
Farh, 2012). Prohibitive voice is members’ expression of concerns about practices or behaviors
that are detrimental to the team, whereas promotive voice represents their expression of novel
ideas for improving team functioning. We argue that prohibitive voice and promotive voice can
both enhance team resilience but for different reasons and at different times during the change
process. As prohibitive voice leads teams to stay vigilant toward threat and damage, it can
prevent performance losses by helping teams manage errors. Hence, prohibitive voice can be
particularly functional for teams in the disruption stage when the teams have a need to reduce
action errors in the immediate aftermath of change. In contrast, as promotive voice mobilizes
members to pursue ideal and aspirational goals, it can facilitate performance gains by enabling
members to introduce innovative changes to team processes. Hence, promotive voice is likely
particularly functional for teams in the recovery stage when teams become more capable of
seeking systemic improvements in their processes. In other words, we propose that the effects of
prohibitive and promotive voice vary across temporal phases of team adaptation (disruption stage
vs. recovery stage) to externally-imposed change. In addition, given the greater need for error
management and process innovation when the intensity of change—or the degree to which the
change event makes the existing work processes obsolete (Morgeson, Mitchell, & Liu, 2015) —
is higher, we further argue that the beneficial effects of prohibitive and promotive voice on team
resilience are stronger when teams face more intense disruptions.
5
By examining these relationships, we make several contributions. First, we contribute to
research on team adaptation and resilience. Work in this tradition has pointed out that exchange
of task-relevant information is essential for teams that are seeking to overcome change-triggered
challenges (Burke et al., 2006; Christian, Christian, Pearsall, & Long, 2017; King, Newman, &
Luthans, 2016; Stoverink, et al., 2020). However, it has not theoretically clarified the kind of
communication that needs to take place for teams to successfully adapt to change. By pointing
out that challenge-oriented communication in the form of prohibitive and promotive voice helps
teams demonstrate resilience, we add conceptual precision to and, thereby, extend this research.
Second, relatedly, this literature has noted that in the face of change, teams have the task
of minimizing and managing change-driven adversity, as well as mending capabilities post-
adversity (e.g., Alliger et al. 2015). We conceptually elaborate on this idea by delineating error
management (properly handling mistakes during enactment of new team routines) as a way
teams can minimize and manage the negative effects of change and process innovation
(improvement of team routines) as a way teams can mend over time. Importantly, we bring in the
novel insight that teams need to enact these two processes sequentially or in a temporally ordered
manner to demonstrate resilience. Hence, we identify specific processes that help teams during
periods of change as well as when these processes need to be deployed by teams.
Third, we extend emerging research on voice consequences. This research has moved
from establishing that voice generally has positive consequences for teams to specifying when it
does so. For instance, scholars have noted that voice is especially useful when all team members,
and not merely a few of them, have the chance to speak up (Sherf et al., 2018; Wooley et al.,
2010), when it freely flows to people in positions of authority in the team who can act on such
voice (Detert, Burris, Harrison, & Martin, 2013) or when the team had failed to meet previously-
6
set goals and, hence, is motivated to change its processes by acting on voice (e.g., Li, Liao,
Tangirala, & Firth, 2017). We add to this research by providing the insight that voice is
functional for teams that are dealing with externally-imposed change, and particularly so when
such change calls for a more radical departure from the past. Thereby, we highlight that, at the
between-team level, voice can demonstrate significant variability in its usefulness depending on
the intensity of change in the performance environment faced by the teams.
Fourth, in contrast to most prior studies, we take a longitudinal approach to examining
the effects of voice on team outcomes. Thereby, we highlight that the usefulness of voice varies
at the within-team level as different forms of voice are valuable to teams at different points in
time. In particular, we propose that prohibitive voice allows teams to immediately mitigate
negative consequences of disruptive change, whereas, promotive voice allows teams to recover
or bounce back from initial decline in performance by innovating in the changed environment.
Demonstrating such temporal bracketing of the validity (referred to as the validity interval;
Zaheer, Albert & Zaheer, 1999) of a variable is conceptually important; Otherwise scholars can
come to erroneous conclusions about the effects of that variable as they can mistakenly discard
or affirm theories regarding how it operates based on when, in time, its effects are sampled.
THEORY AND HYPOTHESES
Team Resilience as a Process Unfolding over Two Stages
Changes in technology or regulations, mergers and acquisitions, or new raw material can
act as exogenous triggers for internal change in teams (Stouten, et. al, 2018). These events
“reflect discrete, discontinuous “happenings,” which diverge from the stable or routine features
of the organizational environment” (Morgeson et a., 2015, p. 519). In the face of such events,
teams experience two distinct stages of change—disruption and recovery (Hale, et al., 2016).
7
Although there is likely some short-term overlap between these stages, they typically unfold
sequentially over time in the post-change environment (Devraj & Jiang, 2019; Hale, et al., 2016).
The immediate aftermath of the change event is characterized as the disruption stage. In
this stage, teams usually experience a sharp dip in performance as members find it difficult to let
go of their old habits and adopt new behaviors or cognitions (LePine, 2003). Action errors or
unintended deviations from plans or incorrect actions due to lack of knowledge (Frese & Zapf,
1994) can frequently occur at this stage. After this early period of volatility, teams enter the
recovery stage (Devaraj & Jiang, 2018; Hale, et al., 2016). In this stage, members begin to
exhibit greater fluency in enacting new behavioral routines. The initial drop in performance is
stemmed and teams start leveraging their initial experience with the changed environment to
improve their ways of working. Hence, recovery stage is characterized by steady over time
improvement in team performance. In sum, as Figure 1A depicts, following an exogenous
change event, teams tend to go through an early period of sudden performance drop (in the
disruption stage) followed by a gradual gain in performance over time (in the recovery stage).
Distinct Functions of Promotive Voice and Prohibitive Voice
Voice, or the member expression of concerns or ideas (Li et al., 2017), can help teams
prevent performance losses in the disruption stage and achieve performance gains in the recovery
stage. It represents a direct exercise of upward influence by members to change team processes
and outcomes (e.g., Tangirala & Ramanujam, 2008); because it is such a form of informal shared
leadership engaged in by the members, voice can be considered a behavioral input to team
functioning within the input-process-output (IPO) models that describe the way teams operate
(Mathieu, Maynard, Rapp, & Gilson, 2008). During change, teams need to put all resources at
their disposal to use to survive and thrive (Stoverink, et al., 2019). Via voice, employees draw on
8
their privately held knowledge or expertise and makes it publicly available in the form of ideas
and concerns that the teams can act on (Sherf et al., 2018). Voice emerges as a consequence of
other input factors such as leadership, incentives, and member personality that orient employee
to constructively contribute to the change process (Morrison, 2014). Hence, it can act as a
connective tissue that links those more distal antecedents to team processes during change.
To understand voice effects on team resilience in more depth, it is critical to differentiate
two forms that voice can take in teams: Prohibitive voice and promotive voice (Huang, Xu,
Huang, & Liu, 2018; Kakkar, et al., 2016; Liang et al., 2012; Lin & Johnson, 2015). Prohibitive
voice is directed at avoiding failure in or harm to teams and involves the expression of concern
by members about existing or impending factors that can cause a deterioration of performance
(Li et al., 2017). Hence, prohibitive voice drives teams to stay vigilant towards threat and
therefore helps them avoid damage and losses. In contrast, promotive voice refers to the
expression of new ideas or solutions for improving team functioning (Liang et al., 2019). Hence,
promotive voice can encourage teams to try out or adopt new practices and procedures.
The Effects of Prohibitive Voice at the Disruption Stage
In the disruption stage, prohibitive voice can be instrumental in preventing performance
losses in teams by facilitating error management, a team process through which members reduce
the amount, scope, and impact of errors in their teams (Rizzo, Ferrante, & Bagnara, 1994;
Kontogiannis & Malakis, 2009). Errors are the discrepancy between expected and observed
outcomes (Zapf & Reason, 1994). Error management represents a collective recognition among
members that an error is occurring or about to occur, diagnosis by them of the cause of the error,
and corrective actions taken by them as a team to remove the sources of the error (Sellen, 1994;
Rizzo, Ferrante, & Bagnara, 1994). In the context of IPO models of team functioning, whereas
9
prohibitive voice acts as an input that represents members’ expressions of concern about
problems confronting the team, error management as a team process describes how teams can
draw on that input to remain vigilant about specific errors that might emerge from those
problems, unpack the root-causes of such problems, and take steps to address them.
Immediately following the change event, team performance dips because members are
suddenly thrown out of pre-change equilibrium and often falter in properly operating new
unfamiliar routines in an error-free manner (Vallas, 2003; Waller, 1999). Hence, proper handling
of errors is critical for minimizing performance losses at this juncture. Prohibitive voice is
particularly valuable in this context. It involves members expressing concern about factors that
can cause deterioration of the status quo and giving one another a ‘‘heads-up’’ when they see a
threat looming (Liang et al., 2012). Thus, such voice allows for open airing of information on
threats to performance that might not otherwise be common knowledge across all members.
Hence, when prohibitive voice is higher, teams can develop a richer understanding of errors—a
critical condition to rectify and minimize the effects of those errors. Moreover, given its focus on
harms and losses to teams, prohibitive voice can bring in a vigilance mindset in teams (Li et al.,
2017). Therefore, it can push teams to remain on alert about any discrepancies between expected
and observed outcomes in the team. It can also make the teams motivated to elaborate on and do
root-cause analyses of such discrepancies to identify habits or behaviors that underlie them.
Thereby, it can help teams set up systems that correct those habits or behaviors and prevent
errors from reoccurring. Consequently, we propose that when prohibitive voice is higher, teams
are able to better engage in error management, which minimizes performance losses that occur in
the early aftermath of exogenously induced change. This is graphically illustrated in Figure 1B.
Hypothesis 1a: In the disruption stage, prohibitive voice is negatively related to team
10
performance losses.
Hypothesis 1b. In the disruption stage, prohibitive voice is negatively related to team
performance losses via error management.
The Effects of Promotive Voice at the Recovery Stage
In the recovery stage, the team shifts focus from merely identifying and correcting errors
to demonstrating positive growth in the post-change environment (Hales et al., 2016). Promotive
voice can facilitate performance gains in this stage by enhancing process innovation, or the
implementation of creative changes to team routines (De Dreu & West, 2001). In the context of
IPO models of team functioning, whereas promotive voice acts as an input that represents
expressions of work ideas, process innovation as a team process represents how teams
collectively elaborate on those ideas and find ways to put them to actual use over repeated trials.
Promotive voice, by bringing forth novel work ideas from members (Liang et al., 2012),
can stimulate teams to think about new possibilities that have not been considered previously.
Hence, it can trigger exploration-oriented actions such as experimentation that can lead to
innovation within teams (Liang et al., 2019; Edmondson & Lei, 2014). Moreover, promotive
voice by drawing out members' diverse ideas, can allow for their creative synthesis within teams
(Harvey, 2014). Thus, it can help teams come up with and implement novel modifications to
work processes that can improve team performance. Additionally, as promotive voice
emphasizes the pursuit of aspiration, accomplishment, and ideal future states, it can help foster
promotion-focused mindsets in teams (Li et al., 2017), which can further motivate them to depart
from conventional solutions and integrate divergent lines of thought to improve team processes.
Consequently, we propose that when promotive voice is higher, teams are able to better engage
11
in process innovation, which can enhance over time performance gains in the recovery stage.
This phenomenon is illustrated in Figure 1C.
Hypothesis 2a: In the recovery phase, promotive voice is positively related to team performance
gains over time.
Hypothesis 2b. In the recovery phase, promotive voice is positively related to team performance
gains over time via process innovation.
We do not hypothesize effects of promotive voice on performance losses in the disruption
stage, or of prohibitive voice on performance gains in the recovery stage. In the disruption stage,
teams are in the midst of a negative performance spiral, which primes attention toward
preventing loss (Idson, Liberman, & Higgins, 2000). Hence, teams in this stage are likely more
motivated to act on prohibitive voice, which due to its focus on minimizing harm, fits their
regulatory focus (Cesario, Higgins, & Scholer, 2008). Moreover, confronted with errors of
execution, teams in this stage might be less capable of further modifying processes that are still
unfamiliar to them. Hence, promotive voice that encourages adoption of new and novel ways of
operation (Li et al., 2017), might be less appealing to teams in the disruption stage. Therefore, as
illustrated in Figure 1C, although promotive voice should enhance performance gains in the
recovery stage, it is less likely to impact the performance of teams in the disruption stage.
In the recovery stage, teams have potentially emerged from the instability that comes in
the immediate wake of the change event; they would have developed an initial understanding of
the changed environment based on their experiences in the disruption stage (Devaraj & Jiang,
2018; Hale et al., 2016). This should allow them shift attention from mere error correction to the
more ambitious task of improving their processes (Argyris & Schon, 1978). In this context,
prohibitive voice that is focused on merely minimizing errors (Li et al., 2017) should be less
12
appealing and a poorer-fit to the needs of the teams (Idson, et al., 2000). Prohibitive voice that is
targeted toward reducing behavioral deviations in teams (Li et al., 2017) is also incompatible
with experimentation that can lead to process innovation that is so critical during recovery stage.
Therefore, as illustrated in Figure 1B, although prohibitive voice should prevent performance
losses in the disruption stage, it is less likely to impact performance gains in the recovery stage.
The Moderating Role of Change Intensity
When the post-change environment is more divergent from the pre-change environment,
teams need to cognitively and behaviorally adapt more to the change event (Morgeson et al.,
2015). In those circumstances, teams should find a greater utility for voice, which can help teams
manage and minimize the greater disruption caused by the event as well as address the stronger
need for recovery and mending post such performance disruption. Thus, we propose that change
intensity acts as a moderator of the effects of voice on team outcomes. Change intensity refers to
the extent to which the change event makes prevailing practices and structures obsolete
(Morgeson et al., 2015). When confronting more intense change, teams have to deal with more
radically different post-change task requirements. Such teams find their previously acquired task
knowledge and methods of coordination less applicable to the new environment, and are
compelled to more fundamentally alter how they carry out their tasks. Hence, in the face of
intense change, teams are required to make more significant, concurrent, and large-scale
adjustments to their cognitions, structures, and routines to reestablish an alignment between their
internal team processes and their performance environment (Ledford & Mohrman, 1993).
In this context, input factors such as prohibitive and promotive voice that can enable key
team processes (error management and process innovation) can be even more useful in
facilitating resilience. Specifically, when change intensity is higher, teams face the possibility of
13
a steeper decline in their immediate post-change performance due to higher occurrence of errors
in a more unfamiliar environment (e.g., Frese & Keith, 2015). In that context, when members are
willing to point out existing or looming problems via prohibitive voice, teams might be better
able to minimize or manage errors caused by those problems. Thereby, they might better stem
performance losses in the disruption stage. Similarly, when change intensity is higher and teams
confront a more novel environment, initial plans for operating might need to be adjusted even
more within teams (Argote & Eppel, 1990; Edmondson & Lei, 2014). In that context, when
members contribute more new ideas via promotive voice, teams are better prepared to meet the
demand for greater process innovation and thereby set out on a steeper performance trajectory in
the recovery stage. In sum, the effects of prohibitive voice (promotive voice) should be stronger
on error management (process innovation) and, hence, on performance outcomes when change
events are more intense. Figure 2 illustrates our overall theoretical model.
Hypothesis 3a. In the disruption phase, the effect of prohibitive voice on team performance
losses is stronger when change intensity is higher.
Hypothesis 3b. In the disruption phase, the indirect effect of prohibitive voice on team
performance losses via error management is stronger when change intensity is higher.
Hypothesis 4a. In the recovery phase, the effect of promotive voice on team performance gains is
stronger when change intensity is higher.
Hypothesis 4b. In the recovery phase, the indirect effect of promotive voice on team performance
gains via process innovation is stronger when change intensity is higher.
OVERVIEW OF STUDIES
We tested our theoretical model in two studies. The first study was a correlational field
study involving 172 production teams that were naturally experiencing an exogenous change
14
event to which they had to adapt to. Using archival objective data on how teams met their
production targets, we were able to capture team performance trajectories multiple months pre-
and post- the disruptive event. We administered surveys during this time horizon to measure our
key variables. We supplemented this correlational study with a field experiment involving 88
teams in a different production setting that were also experiencing externally induced change. In
different conditions, the members were trained and incentivized to engage in different forms of
voice. Team resilience in those voice conditions were compared to that in a control condition to
better establish causality in the relationship between voice and team resilience.
STUDY 1: METHOD
Study 1 was approved by the Texas Christian University IRB (Protocol Number: 1806-
104-1807). We collected data from a China-based global leader in specialty chemicals that
produces more than 300 varieties of products used in textile, printing, and petrochemical
industries. The study was conducted during a period in which the company underwent a business
process reengineering. The company was struggling to hold its position in the coating segments.
A reputable external consulting firm was brought in to help with the problem. The consultants
helped the company launch an automation initiative to improve operational efficiency. As a part
of this initiative, the company introduced 4 new product-administration platforms to replace
existing ones that had been in use for years. As a result, production teams needed to work with
new hardware and software to manage their production lines. The company also changed key
suppliers to ensure that the raw material used in the production lines was of consistent quality.
This initiative was fully visualized and driven top-down and represented an exogenous event for
the production teams, which were at one point in time informed of the change and provided new
templates to work off by external consultants. This change was discontinuous as the teams had to
15
substantively modify their existing practices, structures, and workflow.
Additionally, given that automation needs differed across production teams, the use of
new platforms affected some teams more drastically than others. That is, although all production
teams which constituted our sample confronted discontinuous change, the intensity of change
faced by them varied. Some teams had to operate completely new and unfamiliar processes
whereas others had to face less novelty as some of the new processes were not so fundamentally
different from prior manual ones that they were used to. The change predominantly focused on
the technical aspects of team functioning and left other aspects intact. For instance, team
membership remained consistent pre- and post-change. Hence, we were able to eliminate
confounding factors such as varying team composition that may account for change in team
performance over time. Members performed interdependent tasks that involved reciprocal
coordination of action. Although the teams needed to achieve production goals and qualification
criteria set top-down by the management, they had a great deal of autonomy in carrying out team
tasks; the teams developed their own plans to meet production goals and adjusted internal team
processes to ensure that members’ actions are aligned with external demands. These teams were
in existence for several years prior to the change and had stable membership.
Procedure and Sample
We sought to minimize common source biases (Podsakoff, MacKenzie, Lee, &
Podsakoff, 2003) by collecting data from different sources, including archival records, leaders,
and members. Once the decision for change was made by top management, an announcement
was made, and the teams were notified that the change would occur at a pre-set time point. We
identified 245 teams influenced by the change and solicitated aggregate demographic
information about them from the company. These 245 teams had a total of around 1300 members
16
(82.05% male and average age of 33.42 years). Immediately following this announcement, team
leaders, who were charged with leading the implementation within their teams, began
discussions with the consultants about the new processes that their teams would need to adopt.
Figure 3a provides our data collection timelines for Study 1. We captured team
performance from company records. We had 8 measurements of team performance. We had 4
measurements exactly 1 month apart pre-change (T1-T4) and 4 measures exactly 1 month apart
post-change (T5-T8). The change event occurred shortly after T4 and the time lag between the
change event and T5 was one month. Approximately 2-3 weeks prior to change, we captured our
predictors and moderator via first survey administration (Survey 1). By this time, the team
leaders have developed a good understanding of the scope of change and were able to report on
the intensity of the such change. At this time, members provided ratings of the typical extent to
which voice is expressed in their teams. This allowed us to capture between team differences in
the expression of voice and temporally separate the measurement of voice from that of our
mediators. To appropriately time the measurement of our mediators, we closely worked with the
company. According to the real-time data obtained by the company, team performance began to
show signs of stabilization (the variation in key operational metrics started reducing) at the end
of 3 weeks after the change event. Hence, via our second survey administration (Survey 2), we
measured error management in the following week (before T5) during the disruption stage. In
addition, most teams recovered 65% of performance losses by the end of 3 months after the
change event. Hence, via our third survey administration (Survey 3), we measured process
innovation in the following month during the recovery stage (before T8).
We administered our surveys in Chinese that were created by two bilingual scholars with
doctorate using Brislin’s (1970) translation-back-translation procedure. Disagreement was
17
resolved through discussion. Items were further modified and contextualized based on the
feedback from several managers from the company. Participants completed the surveys during
work hours or lunch breaks and were assured of confidentiality. They returned their surveys in
sealed envelopes directly to us. To avoid any hint of coercion in responding, we did not solicit
any individually identifying information from them. Instead, they only provided us with their
team identity. All participants received a small gift in exchange for their time. We were able to
match data for 172 teams across all time periods. Hence, we had a response rate of 70.2 % at the
team level (172 out of 245 teams). These 172 teams had 904 members in total or 5.26 members
on average (range: 4 to 7). For Survey 1 and Survey 2, average within-team response rates
(ranging from 60%-100%) for this final sample were 80.36% and 79.91%, respectively.
Measures
Unless specified otherwise, all measures were anchored on 7-point Likert scale (1=
“Strongly disagree” to 7 = “Strongly agree”). All our measures are listed in Online Appendix A.
Team voice. We measured prohibitive and promotive voice from members’ using 5-item
scales from Liang et al., (2012) modified to make team the referent. The mean rwg(j) for
prohibitive voice was .88, ICC(1) was .24 (F = 2.24, p < .01) and ICC(2) was .55. The mean rwg(j)
for promotive voice was .86, ICC(1) was .22 (F = 2.14, p < .01) and ICC(2) was .53.
Furthermore, F-test was significant (p < .01) indicating between-team differences in mean levels
of both measures. Although our ICC(2) values were lower than the conventional cutoffs, other
aggregations statistics, such as ICC(1) and rwg(j), were within acceptable ranges (LeBreton &
Senter, 2008). In such circumstances, aggregation can be justified on a conceptual basis, which
exists for voice based on prior validation studies of that construct at the team level (Podsakoff,
Maynes, Whiting, & Podsakoff, 2015). Hence, we felt that there was sufficient evidence for the
18
aggregation of voice scales to the team level.
Change intensity. We measured change intensity using a 3-item scale from team leaders’
perspective. This scale was adapted from a measure of radical innovation (Subramaniam &
Youndt, 2005) that focused on the extent to which a new technology is different from the old
one. The scale was especially appropriate as our research context also focused on technological
upgrades. We modified the scale to fit our study setting.
Error management. We measured error management from team members using 6-items
relevant to the study context from the 10-item scale by Kontogiannis (1999). We directed team
members to focus on the period after the change (which constituted the disruption stage). The
mean rwg(j) was .84, ICC(1) was .17, and ICC(2) was .45. Furthermore, F-test was significant (p
< .01) indicating between-team differences in the mean level of the measure. Although our
ICC(2) was lower than the conventional cutoff, other aggregations statistics, such as ICC(1) and
rwg(j), were acceptable (LeBreton & Senter, 2008). Moreover, the variable has been often
conceptualized at the collective level (e.g., van Dyck, Frese, Baer, & Sonnentag, 2005), Hence,
we felt that there was sufficient empirical and theoretical justification for its aggregation to the
team level.
Process innovation. We used a 3-item scale (Farh, Lee, & Farh, 2010) to measure
process innovation from team leaders. The team leaders performed part of team tasks and
supervised members’ job activities on daily basis. As a result, they were well placed to report on
process innovation in their teams.
Team performance. The company relied on an objective metric—the percentage of team
outputs that met preset quantity or output standards—to measure and track performance of
teams. Given that team performance represents the extent to which the teams are able to meet the
19
expectations or demands of the organization (Pritchard, Jones, Roth, Stuebing, & Ekeberg,
1988), percentage scores of team performance can directly indicate how well the teams were able
to deliver on their promises. Additionally, the output standards did not change before and after
the change event, which made our metric comparable across the time span of the study. We
captured team performance at 8 measurement occasions—the first 4 measurements represented
pre-change performance and the later 4 measurements represented post-change performance.
Confirmatory factor analysis (CFA). We had asked members to provide their responses
that were identified only by their team numbers. Via this approach, we were able to connect
responses at the team-level over time but were unable to do the same with individual member
responses. In that context, we sought to establish the discriminant validity of our measures in
multiple ways. To begin with, as prohibitive and promotive voice were rated at the same time,
we conducted a CFA on these two measures. Results indicated that the 2-factor model fit the data
reasonably well, χ2 (34) = 100.54, p < .01, CFI = .98, RMSEA = .05, SRMR = .02. When the two
forms of voice were loaded on a single factor, the model fit substantially deteriorated χ2 (35) =
657.93, CFI= .81, RMSEA = .16, SRMR = .08. Additionally, we used all measures at the team
level (prohibitive voice, promotive voice, change intensity, error management, and process
innovation) to conduct another CFA. In the case of measures that were reported by members, we
created indicators in our measurement model, each of which represented the aggregated scores of
all the members on individual items of the measures. Results indicated that the 5-factor model fit
the data reasonably well, 2 (N = 172; df = 199) = 246.30, p < .05, CFI = .98, TLI = .97, RMSEA
= .04, SRMR = .04. Alternative nested models achieved significantly poorer fit. Hence, our
measures showed evidence of discriminant validity
1
.
1
We conducted a content validity study following methods recommended by Colquitt, Sabey, Rodell, and Hill
(2019) to assess whether prohibitive voice is distinct from error management and whether promotive voice is
20
STUDY 1: ANALYTICAL APPROACH
We employed discontinuous growth modeling (Singer & Willett, 2003) to test our
hypotheses
2
. Discontinuous growth modeling identifies patterns underlying repeated measures of
the dependent variable (team performance in our case) that are punctuated by a discontinuous
event (Bliese & Lang 2016). This approach helps track the dependent variable’s trajectory before
the event (pre-change stage of stability), in the immediate aftermath of the event (post-change
disruption stage), and, finally, in the later time following the event (post-change recovery stage).
It also allows for modeling predictors (prohibitive and promotive voice, error management,
process management, and change intensity in our case) of the dependent variable’s trajectory at
any particular stage while factoring in the correlated error structure of nested observations.
Appendix A summarizes the way we coded time, disruption and recovery in our analyses.
Two types of coding can be utilized in discontinuous growth models—relative coding and
absolute coding (See Bliese & Lang, 2016). In relative coding, time is coded as linearly
increasing over the focal period, whereas, in absolute coding, time is coded as linearly increasing
till the point of change but as a constant afterwards. In relative coding, the coefficients for
disruption and recovery can be understood as dip and bounce-back in the dependent variable,
respectively, relative to the projected trajectory of the dependent variable in the focal period.
Whereas in absolute coding, the coefficients for disruption and recovery are interpreted as the
extent of dip and bounce-back in the dependent variable, respectively, from the point of the
distinct from process innovation. We recruited 260 working adults from Prolific Academy online panel. The details
of the study are available in Online Appendix C on Open Science Framework. The results indicated that our scales
demonstrated definitional correspondence—items showed fidelity to definitions of the constructs—and –
definitional distinctiveness–items of one construct were distinct from those of the other such that the participants
were able to distinguish between items of prohibitive voice and error management and those of promotive voice and
process improvement. This further bolstered our confidence in the uniqueness of these constructs.
2
Data and syntax used in our analysis are available on Open Science Framework: bit.ly/2D84mCq
21
change without reference to the projected trajectory of that dependent variable. Given our focus
on the post-change period, we utilized absolute coding. However, we reproduced our results with
relative coding as robustness analyses to show that our coding choices did not affect our results.
We adapted the discontinuous growth modeling codes in R from Bliese and Lang (2016)
to the context of our model. Appendix A lists the models that we tested using those R codes.
These codes allowed us to index and predict initial performance losses and subsequent
performance recovery using our independent variables and mediators. The results from this set of
analyses is presented in Table 2. For the test of indirect effects, we captured estimates of the
immediate performance losses and subsequent performance gains for all the teams in our sample
from the output of the R codes and conducted a team-level path analyses (Table 3) for indirect
effects using MPLUS 8.1 (Muthén & Muthén, 1998-2017).
STUDY 1: RESULTS AND DISCUSSION
Table 2 displays the means, reliabilities, and correlations. Confirming the default pattern
of team performance expected in Figure 1A, Model 1 of Table 2 indicated that, in general, there
was a dip in performance in the immediate aftermath of change (disruption, b = -6.15, df=1201, p
< .01) and a positive slope as time elapsed indicating an over-time bounce back over time
(recovery, b = 2.02, df=1201, p < .01); this is represented in the scatter plot of team performance
over time in Figure 4. Hypothesis 1a predicted that prohibitive voice would reduce team
performance losses in the disruption stage. This prediction is tested by the term representing the
interaction between disruption and prohibitive voice in Table 2, Model 2 (Equation 2 in
Appendix A). Prohibitive voice was not significantly related to team performance losses (b = .18,
p > .05), failing to support Hypothesis 1a. Hypothesis 1b predicted that prohibitive voice would
indirectly reduce team performance losses via its positive effects on error management. As Table
3, Model 1 indicates, prohibitive voice was positively related with error management (b = .34, p
22
< .01). Additionally, as Table 3, Model 5 indicates, error management was significantly related
to immediate post-change team performance losses (b = .12, p < .01; note that positive effects of
error management here indicate that it reduces the negative impact of the change event on team
performance). This result is also reaffirmed by the significant interaction between error
management and disruption in Model 3 of Table 2 of the discontinuous growth model (b = .45, p
< .01). Results based on 10000 bootstrap draws suggested that the indirect effect of prohibitive
voice on disruption via error management was significant (.06, 95% CI: .03, .10) such that
prohibitive voice reduced the negative association between change transition and team
performance, supporting Hypothesis 1b. Index of mediation (Preacher & Kelley, 2011), a
measure of mediation effect size, was .08, indicating that performance losses decreased by .08
SD due to error management in the disruption stage caused by 1 SD increase in prohibitive voice.
Hypothesis 2a predicted that promotive voice would increase the pace of team
performance gains in the recovery stage. This prediction is tested by the term representing the
interaction between recovery and promotive voice in Table 2, Model 2 (See Equation 2 in
Appendix A). Indeed, promotive voice was positively related to team performance gains (b = .21,
p < .01), supporting Hypothesis 2a. Hypothesis 2b predicted that promotive voice would
indirectly enhance team performance gains via process innovation. As Model 3 of Table 3
indicates, promotive voice was positively related with process innovation (b = .63, p < .05). As
Table 3, Model 6 indicates, process innovation was related to team performance gains (b = .11, p
< .01). This result is also reaffirmed by the significant interaction between process management
and recovery in Model 3 of Table 2 of the discontinuous growth model (b = .16, p < .01). Results
based on 10000 bootstrap draws suggested that the indirect effect of promotive voice on team
performance gains via process innovation was significant (.07, 95% CI: .04, .12) such that
23
promotive voice enhanced team performance gains in the recovery period, supporting Hypothesis
2b. Index of mediation (Preacher & Kelley, 2011), a measure of mediation effect size, was .14,
which indicated that performance gains increased by .14 SD due to process innovation in the
recovery stage caused by 1 SD increase in promotive voice
Hypothesis 3a postulated that change intensity moderates the relationship between
prohibitive voice and team performance losses in the disruption stage. As reported in Model 4 of
Table 2 (see Equation 3 in Appendix A), the term representing the interaction between
disruption, prohibitive voice and change intensity was significant (b = .37, p < .01). Simple
slopes analysis indicated that when change intensity was low (-1SD), the relationship between
prohibitive voice and team performance losses was negative (b =-.13, p < .05); When change
intensity was high (+1SD), the relationship was positive (b = .14, p < .05). This interaction is
graphically illustrated in Figure 6 (with y-axis reverse coded to facilitate interpretation).
Hypothesis 3b postulated that the indirect relationship between prohibitive voice and team
performance losses via error management should be conditional on change intensity. As seen in
Model 2 of Table 3, the relationship between prohibitive voice and error management was
moderated by change intensity (b = .29, p < .01). Simple slopes analysis suggested that when
change intensity was low (-1SD), prohibitive voice was not related to error management (b = .01,
p > .05); When change intensity was high (+1SD), the relationship was positive (b = .53, p < .01;
see Figure 7). Analyses utilizing 10000 bootstrap draws suggested a non-significant indirect
effect when change intensity was low (-1SD, .00, 95% CI: -.04, .03), but a significant one when
change intensity was high (+1SD, .09, 95% CI: .04, .15). These indirect effects were
significantly different (estimate = .08, 95% CI: .03, .15), supporting Hypothesis 3b.
Hypothesis 4a postulated that change intensity moderates the relationship between
24
promotive voice and team performance gains in the recovery stage. As reported in Model 4 of
Table 2, (see Equation 3 in Appendix A), the term representing the interaction between recovery,
promotive voice and change intensity was not significant (b = .02, p > .05), failing to support
Hypothesis 4a. Hypothesis 4b postulated that the indirect relationship between promotive voice
and team performance gains via process innovation is conditional on change intensity. As seen in
Model 4 of Table 3, the relationship between promotive voice and process innovation was
moderated by change intensity (b = .29, p < .05). Simple slopes analysis suggested that when
change intensity was low (-1SD), promotive voice was related to process innovation (b = .39, p
< .05); When change intensity was high (+1SD), the positive relationship was stronger (B = .92,
p < .01; See Figure 8). Analyses utilizing 10000 bootstrap draws suggested a significant indirect
effect when change intensity was low (-1SD, .05, 95% CI: .01, .09), and a stronger one when
change intensity was high (+1SD, .11, 95% CI: .06, .17). These indirect effects were
significantly different (.06, 95% CI: .01, .12), supporting Hypothesis 4b.
We performed several robustness checks. First, we examined whether our results held
when controlling for the nonlinear trends of team performance over the 8 performance periods
(including the 4 pre-change performance periods and the 4 post-change performance periods)
and the nonlinear trends of performance recovery over the 4 post-change performance periods.
The pattern of our findings remained the same even after controlling for the two quadratic
functions in our models. Second, we examined whether our findings are robust to using relative
coding of time (see Analytical Approach section and Appendix A), The pattern of our results and
the support for our hypotheses was unchanged. Given that tests using relative coding compare
the effects of predictors on post-change periods relative to pre-change periods (Bliese & Lang,
2016), general support for our theoretical model using relative coding indicates that prohibitive
25
and promotive voice were more likely to impact team performance under conditions of change
(i.e., in the post-change disruption and recovery stages) than under conditions of stability (i.e., in
the pre-change stability stage). Hence, in our data, voice was more useful to teams when they
confronted change in the environment than when they faced stability. We discuss the
implications of these issues in greater details in the discussion section.
In Study 1, we tested our theoretical model in a sample of 172 production teams adapting
to discontinuous change. Our theory was largely affirmed (with the exception of Hypotheses 1a
& 4a). In particular, our results indicated that especially under conditions of higher change
intensity, prohibitive voice helped teams stem performance losses in the disruption stage via its
positive effects on error management and, promotive voice facilitated stronger performance
gains over time in the recovery stage by enhancing process innovation. Although findings from
Study 1 were informative, the correlational research design did not allow for strong inferences of
causality. Moreover, some key variables (e.g., prohibitive voice and error management) were
measured using surveys from the same sources that could have led to percept-percept inflation.
To overcome these limitations, we conducted a field experiment design (Study 2).
STUDY 2: METHODS
Study 2 was approved by the Texas Christian University IRB (Protocol Number: 1920-
213). The field experiment took place in a company in China (different from Study 1) that
produces electric and electronic devices such as power transformers. Our data collection focused
on a period during which the company had to produce a specially designed product based on
market demand. As a result, there was a top-down push to majorly reconfigure production. This
represented an exogenous change in task demands for production teams. Figure 3b provides data
collection timelines of the study. We identified 106 teams that were affected by the change. In
26
general, the change was challenging. Due to this externally imposed change in the manufacturing
process, workers had to grasp new technical details which were more complex than what they
were accustomed to. Although engineering blueprints were provided, workers still had to figure
out operational steps such as the order in which to place washers and spacers when assembling
their work. Because of the nature of the change, some teams faced more radical transition than
others. For instance, wiring teams were impacted more strongly as they, prior to the change, used
machines to cut wires of uniform length and bundled them with clamps; post-change, the wires
were to be cut in different lengths and routed through sleeves with the application of a fabric
tape. In contrast, given that the size and shape of frames only changed slightly, molding teams
were not as severely impacted by the change.
We randomly assigned the teams (using a random number table) to 1 of 3 experimental
conditions: prohibitive voice (n = 36), promotive voice (n = 35), and control (n = 35). We had
usable data from a (final) sample of 88 teams (28 in prohibitive condition, 31 in promotive
condition, and 29 for the control condition). We administrated data collection in 3 waves. We
introduced the manipulation and measured change intensity from leaders at T0, approximately 1
week prior to the change event. After the change event, we measured error management and
process innovation from members at T1 (1 week after the change event, which represented the
disruption stage as per our conversation with company managers) and T2 (2.5 weeks after T1,
which represented the recovery stage according to company managers). We captured team
performance from skip-level managers who supervised these teams at T0, T1, and T2. The 2
forms of voice, as manipulation checks, were measured at all the 3 time points. Average within-
team response rates (ranging from 60%-100%) for T0, T1, and T2 were 81.12%, 82.71%, and
79.68%. In the final sample, average team size was 3.93 (range: 3 to 6), average age was 38.62
27
years, and 54.53% was male. There were no significant differences among conditions on any of
these demographic variables. Surveys were administered during work hours. Participants
returned the surveys in sealed envelopes to us. To avoid coercion, we did not solicit any
individually identifying information, but only team identity. All participants received a small gift
in exchange for their time. Surveys were administrated in Chinese and created by two bilingual
scholars with doctorate using Brislin’s (1970) translation-back-translation procedure (any
disagreement about translation was resolved through discussion). The items were further
modified and contextualized based on the feedback from several managers from the company.
Manipulation of prohibitive and promotive voice
The teams were randomly assigned to 3 conditions—promotive voice, prohibitive voice
and control. In particular, we followed the best practices in conducting field experiments in
identifying a situation where the company was naturally working on a plan for enhancing voice
(employee participation) and we helped out by providing outside technical assistance (Grant &
Wall, 2009). In particular, in the voice conditions, voice was manipulated using 3 closely related
activities. First, we provided training materials that were used by the company managers to help
team members understand the importance of voice and highlight how voice can be expressed in a
respectful, thoughtful, constructive, and clear manner during team interactions (Details of these
training materials can be found in Online Appendix B). These materials were reviewed by
members collectively during the daily operational review in the first week following the change
event, and at least once each week for the rest of the focal period.
Second, the company provided a monetary incentive to encourage members to speak out.
Specifically, for the month, which represented focal period of the study, the company modified
how monthly performance-based bonuses for members were assessed. In general, the members
28
had monthly bonuses that were tied to a set of criteria regarded as important by the company.
Voice was introduced as one of the criteria for this bonus in the study period. In particular, voice
carried 15% weight in the monthly bonus and made up an average of 4.5% of total monthly pay
(varying from 1% to 6.5%). This bonus was based on leaders’ evaluation of members’ voice on a
5-point scale (1=Extremely dissatisfactory, 2=Less than satisfactory, 3=Satisfactory, 4=More
than satisfactory, 5=Extremely satisfactory). Importantly, the company did not change any other
criteria that already been used but merely rolled back the weight of them proportionally such that
the overall bonus levels for teams did not change; only voice was introduced as a new criteria for
existing bonus allocations. This change did not affect employees’ base pay or any other benefits.
Third, time was set aside for voice (a minimum of 5 minutes) in the daily operational
team reviews where all members were present. Members could use this opportunity to express
voice or recognize teammates’ voice. In this forum, members were encouraged to express their
voice in an unrushed way in case they were not able to communicate their thoughts or opinions
during the previous day as the team was enacting its routines. Members were also urged to
applaud or recognize any instances where they noticed teammate(s) who spoke out the previous
day. Team leaders were asked to take notes and keep a log for members’ voice.
Although the above 3 activities similarly took place in both voice conditions, different
content of voice (Liang et al., 2012) was emphasized in the promotive voice and the prohibitive
voice conditions. In the prohibitive condition, the members were asked to focus on bringing up
concerns about how the team as a whole can avoid deviating from prescribed guidelines and
procedures in the modified manufacturing processes. In the promotive condition, the members
were asked to focus on expressing suggestions on how the team as a whole can change internal
team processes for the better. Hence, in the prohibitive condition, members were trained to and
29
rewarded for reporting existing or potential problems that make the status quo unsatisfactory or
worse by raising timely alarms that can hurt the team. In contrast, in the promotive condition,
members were trained to and rewarded for bringing up creative suggestions for improving
current work practices. The teams in the control condition did not receive any of these voice
treatments but only technical training pertaining to change that was common across all the
conditions. To avoid any confounding, that the overall bonus levels and material resources (e.g.,
equipment, leadership support, technical support from quality control) received by the teams in
the control condition were the same as those received by the teams in the voice conditions. We
had no reports of any treatment diffusion across conditions.
Measures
The measures of promotive and prohibitive voice (which acted as manipulation checks
for our interventions), change intensity (moderator), error management, and process innovation
(mediators) were identical to those in Study1 (see Online Appendix A). Team performance was
measured using a 5-item scale (Van & Bunderson, 2005) to that captured collective success on 5
dimensions of work efficiency, quality, productivity, goal achievement, and overall effectiveness
on a 9-point Likert scale (1 "far below established standards", to 5 "meet established standards ",
to 9 "far above established standards "). For measures of voice and team performance at Time 0,
the participants focused on the month before the change event. For all measures at Times 1 and
2, instructions directed participants to focus on the time period after the last wave of survey.
Control variables. Although we randomly assigned teams to conditions that limited the
need for controls, we sought to establish the incremental validity of our voice manipulations over
related inputs available to teams. Hence, we controlled for several variables captured at T0. Our
results remained substantially the same with and without these controls. We controlled for team
30
size that represents a resource for teams (Duffy, Shaw, & Stark, 2000) and for transformational
leadership that can impact team resilience (Dimas et al., 2018) using a validated short 7-item
measure (Carless, Wearing, & Mann, 2000). We also controlled for human capital available to
teams in form of member characteristics (Gucciardi et al., 2018). In particular, member
personality traits of openness to experience and conscientiousness can predict how teams adapt
to change (LePine, Colquitt, & Erez, 2000); Thus, we controlled for validated short-form scales
of conscientiousness and openness from a Big-5 Personality battery (Mini-IPIP, Donnellan,
Oswald, Baird, & Lucas, 2006). We used additive form of aggregation to represent
conscientiousness and openness at the team level as averages of member scores (LePine, 2003).
Similarly, members gain job knowledge as well as work skills over their job tenure (Schmidt,
Hunter & Outerbridge, 1986); hence, we controlled for average member job tenure as another
aspect of human capital.
Aggregation statistics: For promotive voice (manipulation check), the mean rwg(j)
was .85, ICC(1) was .30 (F = 2.38, p < .01) and ICC(2) was .58 at T0; the mean rwg(j) was .86,
ICC(1) was .49 (F = 4.17, p < .01) and ICC(2) was .76 at T1; the mean rwg(j) was .88, ICC(1)
was .65 (F = 6.99, p < .01) and ICC(2) was .85 at T2. For prohibitive voice (manipulation
check), the mean rwg(j) was .88, ICC(1) was .37 (F = 2.89, p < .01) and ICC(2) was .65 at T0; the
mean rwg(j) was .87, ICC(1) was .46 (F = 3.76, p < .01) and ICC(2) was .73 at T1; the mean rwg(j)
was .89, ICC(1) was .57 (F = 5.26, p < .01) and ICC(2) was .80 at T2. For transformational
leadership, the mean rwg(j) was .87, ICC(1) was .35 (F = 2.71, p < .01) and ICC(2) was .63. For
error management, the mean rwg(j) was .89, ICC(1) was .51 (F = 4.41, p < .01) and ICC(2)
was .77 at T1; the mean rwg(j) was .88, ICC(1) was .44 (F = 3.53, p < .01) and ICC(2) was .71 at
T2. For process innovation, the mean rwg(j) was .85, ICC(1) was .50 (F = 4.33, p < .01) and
31
ICC(2) was .76 at T1; the mean rwg(j) was .86, ICC(1) was .51 (F = 4.34, p < .01) and ICC(2)
was .76 at T2. Hence, there was support for the aggregation of these variables to the team level.
Confirmatory factor analysis (CFA). Members survey responses were identifiable only
by their assigned team numbers. Hence, we were able to connect responses at the team-level over
time but were unable to do at the individual level. In that context, we sought to establish the
discriminant validity of our measures at each time point. At T0, results indicated that the 5-factor
model fit the data well, 2 (N = 278; df =265) = 439.40, p < .01, CFI = .94, TLI = .93, RMSEA
= .05, SRMR = .04. Alternative nested models achieved poorer fit. For instance, an alternative 4-
factor model that combined indicators of prohibitive and promotive voice fit data worse,
2 (N =
278; df = 269) = 605.90, p < .01, CFI = .88, TLI = .87, RMSEA = .07, SRMR = .06. At T1,
results indicated that the 4-factor model fit the data well, 2 (N = 284; df = 146) = 344.56, p
< .01, CFI = .95, TLI = .94, RMSEA = .07, SRMR = .04. Alternative nested models achieved
poorer fit. For instance, an alternative 3-factor model that combined indicators of prohibitive
voice and error management fit data worse,
2 (N = 284; df = 149) = 955.31, p < .01, CFI = .78,
TLI = .75, RMSEA = .14, SRMR = .11. At T2, results indicated that the 4-factor model fit the
data well, 2 (N = 274; df = 146) = 246.53, p < .01, CFI = .97, TLI = .97, RMSEA = .05, SRMR
= .04. Alternative nested models achieved poorer fit. For instance, an alternative 3-factor model
that combined indicators of prohibitive voice and error management fit data worse,
2 (N = 274;
df = 149) = 775.79, p < .01, CFI = .81, TLI = .79, RMSEA = .12, SRMR = .12. Together, these
results indicated discriminant validity of our measures.
Analytical approach. We created 2 dummy variables to index the 3 experimental
conditions and used these dummies as independent variables in team-level path analyses in
MPLUS 8.1 (Muthen & Muthen, 1998-2017). To establish how our manipulations and mediators
32
led to a change in team performance, we controlled for prior team performance levels. Hence,
the coefficients for the voice conditions and error management in our regression models at T1
indicated their impact on team performance losses vis-à-vis baseline team performance at T0.
The coefficients for the voice conditions and process innovation in our regression models at T2
indicated team performance gains vis-à-vis team performance at T1. Results remained
substantially the same with or without such controls for prior levels of performance.
Voice Manipulation checks. We manipulated voice at T0 and expected these
manipulations to increase voice at early phases of change (T1) and sustain it into later phases of
change (T2). Hence, we measured voice at T0, T1 and T2 as manipulation checks. An analysis of
variance (ANOVA) with repeated measures indicated a significant interaction between time and
condition for prohibitive voice (F [4, 170] = 8.85, p < .01, η2 = .17) and promotive voice (F [4,
170] = 13.66, p < .01, η2 = .24). As reported in Table 4a, post hoc comparisons indicated that in
the prohibitive voice condition, prohibitive voice was higher at T1 (p <.01) and at T2 (p <.01)
than at T0 (baseline); there were no differences between T1 and T2 levels (p >.05). This
indicated that the prohibitive voice manipulation increased prohibitive voice, which remained at
consistent levels post-change. In the promotive voice condition, promotive voice was higher at
T1 (p <.01) and at T2 (p <.01) than at T0 (baseline); there were no differences between T1 and
T2 levels (p >.05). This indicated that the promotive voice manipulation increased promotive
voice, which remained at consistent levels post-change. There were no concomitant increases in
the post-intervention period in either forms of voice in the control condition (p >.05). Given
commonalities between promotive and prohibitive forms of voice (both being expressions of
challenge to the status-quo; Liang et al., 2012), not surprisingly, prohibitive voice manipulation
enhanced promotive voice and vice versa (p <.01). However, importantly, prohibitive voice was
33
higher in the prohibitive condition than in (a) the promotive condition both at T1 and T2 (d
= .81, S.E. = .22, p <.01; d = .71, S.E. = .25, p <.01, respectively), and (b) control condition both
at T1 and T2 (d = 1.03, S.E. = .22, p <.01; d = 1.06, S.E. = .25, p <.01, respectively). Similarly,
promotive voice was higher in the promotive condition than in (a) the prohibitive condition both
at T1 and T2 (d = .56, S.E. = .23, p <.05; d = .69, S.E. = .26, p <.01, respectively), and (b)
control condition both at T1 and T2 (d = .99, S.E. = .23, p <.01; d = 1.11, S.E. = .25, p <.01,
respectively). This indicated that each form of our manipulation disproportionally enhanced the
corresponding form of voice, thus lending support to the effectiveness of our intervention.
STUDY 2: RESULTS AND DISCUSSION
Table 5 displays the descriptive statistics for Study 2. Table 4b provides mean by
experimental condition for team performance, error management and process innovation. First,
we examined the pattern of team performance across T0, T1, and T2. The teams experienced a
decline in team performance from T0 (M = 6.54, SD = 1.16) to T1 (M = 4.42, SD = 1.40; t (87)
=20.09, p < .01) and then recovered at T2 (M = 5.48, SD = 1.49; t (87) = 10.28, p < .01); this is
represented in the scatter plot representing team performance over time in Figure 5. This pattern
indicated that performance across the three time points showed the expected V-shape pattern as it
did in Study 1 and supported our premise that the period between T0 and T1 represented the
disruption stage and the period between T1 and T2 represented the recovery stage.
Hypothesis 1a predicted that teams in prohibitive voice condition, compared with teams
in control condition, would experience less team performance losses in the disruption stage. As
reported in Model 2 of Table 6, prohibitive voice was related to team performance at T1 (b =
1.02, p < .01), supporting Hypothesis 1a. Hypothesis 1b predicted that error management
mediates the effects of prohibitive voice on team performance. As Table 7, Model 2 indicates,
34
prohibitive voice was positively related with error management at T1 (b = .51, p < .05). As Table
6, Model 4 indicates, error management was related to performance at T1 (b = .37, p < .01).
Results based on 10000 bootstrap draws suggested that the indirect effect of prohibitive voice
condition on performance losses via error management was significant (.19, 95% CI: .03, .42),
supporting Hypothesis 1b. Index of mediation was .13, which indicated that performance losses
differed by .13 SD of performance at T1 between prohibitive voice condition and control
condition due to the effect of prohibitive voice on error management.
Hypothesis 2a predicted that teams in promotive voice condition, compared with teams in
control condition, would experience more team performance gains in the recovery stage. As
Table 6, Model 6 indicates, the promotive voice condition was not related to team performance
gains (b = .19, p > .05), failing to support Hypothesis 2a. Hypothesis 2b predicted that the
promotive voice condition would lead to team performance gains in the recovery stage due to its
positive effect on process innovation. As Table 8, Model 5 indicates, promotive voice was
positively related with process innovation at T2 (b = .63, p < .01). Additionally, as Table 6,
Model 8 indicates, process innovation was related to performance at T2 (b = .30, p < .05).
Results based on 10000 bootstrap draws suggested that the indirect effect of the promotive voice
condition on team performance via process innovation was significant (.20, 95% CI: .04, .49),
supporting Hypothesis 2b. The index of mediation was .12, suggesting that performance gains
differed by .12 SD of performance at T2 between promotive voice condition and control
condition due to the effect of promotive voice on process innovation.
Hypothesis 3a predicted that change intensity moderates the relationship between
prohibitive voice and team performance losses in the disruption stage. However, as Table 6,
Model 3 indicates, the interaction between the prohibitive voice condition and change intensity
35
was not significant (b = .05, p > .05), failing to support Hypothesis 3a. Hypothesis 3b predicted
that the indirect relationship between prohibitive voice and team performance losses via error
management should be conditional on change intensity. As Table 7, Model 3 indicates, the
relationship between prohibitive voice and error management was not moderated by change
intensity (b = -.02, p > .05), failing to support Hypothesis 3b.
Hypothesis 4a predicted that change intensity moderates the relationship between
promotive voice and team performance gains in the recovery stage. As Table 6, Model 7
indicates, the interaction between promotive voice condition and change intensity was significant
(b = .72, p < .01), supporting Hypothesis 4a. Simple slopes analysis indicated that when change
intensity was low (-1SD), there was a negative association between promotive voice and
performance gains (b =-.69, p < .05, See Figure 9). However, when change intensity was high
(+1SD), there was a positive association between promotive voice and performance gains (b
= .98, p < .01). Hypothesis 4b postulated that the indirect relationship between promotive voice
and team performance gains via process innovation is conditional on change intensity. As Table
8, Model 6 indicates, the relationship between promotive voice and process innovation was
moderated by change intensity (b = .52, p < .01). Simple slopes analysis suggested that when
change intensity was low (-1SD), promotive voice was not related to process innovation (b = .01,
p > .05; See Figure 10); However, when change intensity was high (+1SD), the relationship was
positive (b = 1.22, p < .01). Analyses utilizing 10000 bootstrap draws suggested a nonsignificant
indirect effect when change intensity was low (-1SD, .00, 95% CI: -.20, .22), but a significant
one when change intensity was high (+1SD,.38, 95% CI: .09, .82). These indirect effects were
significantly different (estimate = .38, 95% CI: .06, .89), supporting Hypothesis 4b.
36
Via a field experiment, we provide supporting evidence that prohibitive and promotive
voice play distinct roles in facilitating team resilience. Our results showed that (a) prohibitive
voice helped stem performance losses in the immediate aftermath of change event by enhancing
error management, (b) promotive voice contributed to performance gains in the recovery stage
by enhancing process innovation when change intensity was higher, and (c) the effects of
promotive voice were contingent on the change intensity faced by the teams.
GENERAL DISCUSSION
We draw on theories of team adaptation and resilience to explicate how voice can make
teams more effective during discontinuous change. We tested our theory in two studies: A
correlational field study involving 172 production teams in which we examined team adaptation
as indexed by objective team performance, and a field experiment involving 88 teams in a
different production setting, where team members were trained, given opportunities and
incentivized to engage in voice. In general support of our theory, consistently across the two
studies, we found that (a) prohibitive voice reduced performance losses in teams in the disruption
phase of change by enabling error management, (b) promotive voice enhanced performance
gains in the recovery phase of change by enabling process innovation, and (c) the indirect effects
of promotive voice on performance gains in the recovery phase via process innovation were
stronger when the teams faced more intense disruptions. We found mixed support for the
prediction that the effects of prohibitive voice were conditional on disruption intensity (i.e., in
Study 1 but not Study 2; see Table 9). Our findings have several conceptual implications.
Theoretical Implications
Contribution to the literature on team adaptation and resilience. First, we explicate
team processes that explain how teams cope with change. In general, team adaptation and
37
resilience have been conceptualized in multiple ways (Burke et al., 2006; Christian, et al., 2017;
Stoverink, et al., 2020). For instance, team adaptation has been visualized as the ability of teams
to maintain prior performance levels in the post-change environment (LePine, 2003). Similarly,
resilience is often seen as the capability to remain immune to performance decrements in the face
of stressors (see Cheng, King & Oswald, 2020). By contrast, in this manuscript, rather than
merely view team adaptation and resilience as inherent team capabilities that are represented in
the ultimate post-change outcomes achieved by teams (their final post-change equilibrium levels
of performance; LePine, 2003), we take a process-oriented approach to explicating adaptation
and resilience. On that front, we build on recent discussions in the literature that teams go
through distinct phases of disruption and recovery in a post-change context (e.g., DevRaj &
Jiang, 2019; Hale, et al., 2016) and highlight that error management and process innovation are,
previously unexamined but important, sequentially unfolding team processes that underlie team
adaptation and resilience. In doing so, we contribute to research by identifying specific processes
that differentiate resilient and adaptive teams from those that are brittle.
Second, we identify voice as a key predictor of those team processes. Research on team
adaptation and resilience has recognized that, in the turbulence that characterizes current
business environments, effective teams are not merely those that do well during periods of
stability but rather those that can successfully cope with and thrive during change events (e.g.,
Alliger, et al., 2015; Burke, Stagl, Salas, Pierce, & Kendell, 2006; LePine, 2003; Marks, Zaccaro
& Mathieu, 2000; Stoverink, et al., 2020). And, hence, that studying facilitators of team
resilience is a crucial endeavor. Prior conceptual discussions in this research had noted that, in
general, within-team communication is critical for teams to adapt to change and remain resilient
(e.g., Stoverink, et al., 2020). We provide theoretical nuance to and extend these discussions by
38
delineating how particular forms of communication that involve constructively questioning the
status-quo (prohibitive voice and promotive voice) are critical inputs that help teams manage
immediate errors due to change events as well as, over time, seize innovation opportunities
presented by those events to set out on a path of enhanced performance.
Third, empirically (Study 2), we demonstrate that voice holds incremental validity over
other inputs available to teams such as transformational leadership and human capital (team
composition). Voice represents a direct way in which teams can publicly access privately-held
expertise and knowledge of members and put it to use in the team (Sherf et al., 2018). Voice
emerges from (Morrison, 2014) and, thereby, likely connects more distal inputs such as rewards
or leadership to team processes and outcomes. In fact, in our voice interventions in the field
experiment, we leverage training, coaching and incentives to encourage members to speak up in
different forms to help teams remain adaptive and resilient. Hence, we show how voice can be a
proximal antecedent that scholars of team adaptation and resilience need to pay attention to.
Contribution to the voice literature. First, we demonstrate how voice has a stronger
impact on team performance when teams are confronted with changes of higher intensity.
Although existing research has indicated that voice-team performance relationship is conditional
on various factors (e.g., the nature of distribution or flow of voice within teams; e.g., Detert et
al., 2013; Sherf et al., 2018), it has not adequately recognized how, at the between-team level,
the performance context that the teams operate under can impact the effects of voice on teams.
We highlight that voice is especially beneficial to collective effectiveness when teams are in
greater flux—i.e., when they face a greater need to break up with the existing equilibrium and
forge a new one. In a related vein, our robustness analyses in Study 1 using relative coding of
time indicated that voice had a stronger relative impact on team performance in the post-change
39
environment (when team adaption was critical) than in the pre-change environment (when teams
experienced greater stability). This further underscored how voice is particularly critical in
periods of environmental flux during which teams are attentive to threats and opportunities that
get surfaced via voice. Voice is likely to get less attention under the condition of stability during
which teams were likely more comfortable with existing and well-tested routines and practices.
In sum, we suggest that the effects of voice on team functioning are sensitive to the context,
which, we argue, should be more explicitly incorporated in theory building on voice.
Second, we add to emerging work that has sought to delineate how voice comes in
different forms—prohibitive and promotive voice (Kakkar, et al., 2016; Liang et al., 2012; Lin &
Johnson, 2015; Wei, et al., 2015). This work has debated the comparative importance of these
two voice types on individual and team outcomes (Chamberlin, et al., 2017; Li et al., 2017). We
demonstrate that the relative value of prohibitive and promotive voice varies dynamically over
time. That is, prohibitive voice is especially beneficial to teams immediately following an
exogenous jolt as it helps teams manage errors in enactment of planned actions. However, later
over time, promotive voice becomes a particularly important facilitator of team effectiveness as
it enables members to explore and implement novel routines or practices that can lead to
systemic innovation. Hence, our study emphasizes the need to consider time when examining
how prohibitive and promotive voice differentially impact team outcomes. In doing so, we depart
from prior research that has predominantly focused on comparing these two voice forms at one
point in time and highlight how the validity of voice changes at the within-team level.
Third, we designed an evidence-based field intervention (Study 2) to enhance voice. The
elements of the intervention were designed keeping in mind a long tradition of research that has
indicated that voice cannot be merely presumed to be an extra-role behavior that emerges
40
unsolicited at the workplace; but, rather increases when members embrace it as expected
behavior at work because of training, incentives, and structural opportunities to speak up (Parke,
Tangirala, & Hussain, 2020; Sherf, Tangirala & Venkataramani, 2018; Tangirala, Kamdar,
Venkataramani, & Parke, 2013). By designing this intervention, we highlight how conceptual
ideas regarding the antecedents of voice from the literature can be translated into specific
interventions that can encourage and cultivate voice within teams.
Practical Implications
Our study has several managerial implications. First, it is critical for team leaders to not
only appreciate the importance of voice but also understand when it is particularly valuable. We
suggest that it is worthwhile to encourage a high level of voice when teams need to deal with
major exogenous change because voice enables teams to adapt quickly and effectively. In this
sense, voice can separate resilient teams from brittle ones. The former swiftly make necessary
adjustments to perform well in unstable performance environments; whereas, the latter struggle
and even collapse in such contexts. Second, team leaders should encourage both promotive and
prohibitive forms of voice during change. Leaders in general are said to react more positively to
promotive than to prohibitive voice because the latter can come across as more interpersonally
critical (e.g., Chamberlin et al., 2018). This can cause leaders to pay attention to novel
suggestions from members while more readily dismissing their concerns about deviant behaviors
within the teams. However, our findings indicate that both forms of voice are indispensable in
their own way to team success during periods of change. Hence, leaders need to overcome any
bias against prohibitive voice. Third, we provide clear guidance for how leaders can enhance
voice. In our field experiment, by training members to speaking up, incentivizing them for doing
so, and setting aside explicit “voice time” in team meetings to encourage and recognize voice,
41
we were able to successfully increase voice in teams. Such practices can be followed by
managers, especially in circumstances where the task environment of the teams is in flux and
employee participation in the form of voice is critical for team success.
Limitations and Directions for Future Research
Our research has several limitations. First, in Study 1, ICC(2) values for some of our
variables (e.g., voice) were on the lower side. We had proceeded with aggregating these
variables to the team level based on strong theory that these constructs exist at that level as well
as supporting evidence from other aggregation indices (e.g., ICC (1)). However, this constitutes a
limitation because these low ICC2 values, by reducing the reliability of group-means, might have
affected our ability to estimate true relationships among our variables in that study. Although this
was less of a concern in our Study 2, where our ICC2 values were stronger and we replicated key
relationships, we urge caution in interpreting results from Study 1. Second, we examined team
resilience in terms of how teams were able to sustain their performance levels after exposure to
an exogenous change event. We did not examine other team outcomes, such as team viability or
the degree to which members are committed to remain in the team (Mathieu et al., 2008), which
are also important for organizations. Scholars have noted that it is likely that teams that remain
successful in terms of achieving their goals during adversity, might also sustain their team spirit
over time (Stoverink, et al., 2020). However, this remains an assumption that future studies need
to explicitly test. Those studies might need to consider other mediators (rather than error
management and process innovation) that might underlie effects of voice on team viability
during disruptive change (e.g., greater satisfaction with team functioning due to enhanced sense
of agency in the team or the “voice process effect;” Greenberg & Folger, 1983).
42
Third, future research needs to resolve some inconsistencies in results across our studies
(See Table 9 for a summary). Consistently in the studies, we found that (a) by enabling error
management, prohibitive voice stemmed performance losses in the disruption phase of change
and (b) by enabling process innovation, promotive voice enhanced performance gains in the
recovery phase of change; Moreover, such indirect effects of promotive voice were stronger
when the teams faced more intense disruptions. At the same time, we found mixed support for
the idea that prohibitive voice via error management has stronger effects when change disruption
intensity is higher (supported in Study 1 but not Study 2). Interaction effects, especially those
involving categorical variables such as the experimental dummies used in Study 2, need a high-
powered sample (Aguinis, Boik, & Pierce, 2001), which was likely missing in Study 2 (N=88
teams). Studies using larger sample sizes in combination with field experimental designs should
seek to replicate our results and more affirmatively establish whether or not the prohibitive voice
effects are conditional on the intensity of disruptions faced by teams.
Fourth, in Study 1, 68 out of 172 teams showed higher post-change performance levels
(at T8 or the final measurement wave) than pre-change performance levels (the average of the
first 4 waves of measurement—T1-T4). In Study 2, 10 teams out of 88 teams had done so (when
their performance pre-change, T0, was compared with that in the recovery stage, T2). This
indicated that although some teams were able to exceed their pre-change performance equilibria
others had failed to do so within the period of our observation. It is possible that, had we
continued observation for longer periods, we would have seen more teams surpass their pre-
change performance levels (Zaheer, et al., 1999). However, at the same time, these findings do
tentatively suggest that although some teams do show “post-traumatic growth,”—that is, become
stronger as a result of change (Maitlis, 2020)—others do not, perhaps because they lack adequate
43
levels of voice. Future studies should follow teams experiencing disruptive change for longer
periods of time than we did in our studies to establish baselines on how many of them set new
post-change equilibria higher than those they were accustomed to in the pre-change era.
Finally, we examined error management and process innovation as parallel processes that
impact team outcomes at different stages of the post-change adaptation process. The correlations
amongst these processes were low in our data (see Tables 1 and 5). Moreover, exploratory
analysis indicated that these two processes did not interact with each other to predict post-change
performance in either of our studies (p > .05), suggesting that they potentially only had additive
effects on team outcomes in the settings that we examined in our studies. However, we urge
future research to develop and test theory about how these processes might combine to impact
teams. For instance, on the one hand, it is possible that the prevention focus involved in
managing errors can interfere with the promotion focus involved in the innovation process;
hence, teams that are seeking to simultaneously enact both these processes can become distracted
and less effective. On the other hand, it is possible that error management acts as a guard rail that
helps teams engaging in process innovation avoid catastrophic failures when experimenting with
new ideas or practices; thereby, error management can help bolster the effects of process
innovation on team outcomes. Examining such potential interactive effects can help studies
further unpack how teams remain resilient in the face of exogenous change events.
CONCLUSION
We proposed that voice more positively impacts team effectiveness when teams face
discontinuous change. Indeed, our findings indicated that prohibitive voice helped teams stem
performance losses in the immediate aftermath of change by allowing them to manage errors
better, and promotive voice facilitated performance gains in the later phases of change by
44
enhancing process innovation. We hope that these results spur further research on how and when
voice can make teams more adaptive and resilient in the face of exogenous change.
45
REFERENCES
Alliger, G. M., Cerasoli, C. P., Tannenbaum, S. I., & Vessey, W. B. 2015. Team resilience: How
teams flourish under pressure. Organizational Dynamics, 44: 176-184.
Aguinis, H., Boik, R. J., & Pierce, C. A. 2001. A generalized solution for approximating the
power to detect effects of categorical moderator variables using multiple regression.
Organizational Research Methods, 4(4), 291-323.
Argote, L., & Epple, D. 1990. Learning curves in manufacturing. Science, 247(4945): 920–924.
Argyris, C., & Schon, D. (1978). Organizational learning: A theory of action perspective.
Reading, MA: Addison-Wesley.
Bliese, P. D., & Lang, J. W. B. 2016. Understanding Relative and Absolute Change in
Discontinuous Growth Models: Coding Alternatives and Implications for Hypothesis
Testing. Organizational Research Methods, 19(4): 562–592.
Brislin, R. W. 1970. Back-translation for cross-cultural research. Journal of Cross-Cultural
Psychology, 1(3): 185–216.
Burke, C. S., Stagl, K. C., Salas, E., Pierce, L., & Kendall, D. 2006. Understanding team
adaptation: A conceptual analysis and model. Journal of Applied Psychology, 91(6): 1189-
1207.
Carless, S. A., Wearing, A. J., & Mann, L. 2000. A short measure of transformational leadership.
Journal of Business and Psychology, 14(3), 389-405.
Cesario, J., Higgins, E. T., & Scholer, A. A. 2008. Regulatory Fit and Persuasion: Basic
Principles and Remaining Questions. Social and Personality Psychology Compass, 2(1):
444-463.
Chamberlin, M., Newton, D. W., & LePine, J. A. 2017. A Meta-Analysis of Voice and Its
Promotive and Prohibitive Forms: Identification of Key Associations, Distinctions, and
Future Research Directions. Personnel Psychology, 70(1): 11–71.
Cheng, S. K., King, D. D., & Oswald, F. L. 2020. Understanding How Resilience is Measured in
the Organizational Sciences. Human Performance, 1-34.
Christian, J. S., Christian, M. S., Pearsall, M. J., & Long, E. C. 2017. Team adaptation in
context: An integrated conceptual model and meta-analytic review. Organizational
Behavior and Human Decision Processes, 140: 62–89.
De Dreu, C. K. W., & West, M. A. 2001. Minority dissent and team innovation: The importance
of participation in decision making. Journal of Applied Psychology, 86(6): 1191-1201.
Detert, J. R., Burris, E. R., Harrison, D. A., & Martin, S. R. 2013. Voice Flows to and around
Leaders: Understanding When Units Are Helped or Hurt by Employee Voice.
Administrative Science Quarterly, 58(4): 624–668.
Devaraj, S., & Jiang, K. 2019. It’s about time - A longitudinal adaptation model of high-
performance work teams. Journal of Applied Psychology, 104(3): 433-447.
Dimas, I. D., Rebelo, T., Lourenço, P. R., & Pessoa, C. I. P. (2018). Bouncing back from
setbacks: On the mediating role of team resilience in the relationship between
transformational leadership and team effectiveness. Journal of Psychology, 152(6), 358-
372.
Duffy, M. K., Shaw, J. D., & Stark, E. M. 2000. Performance and satisfaction in conflicted
interdependent groups: When and how does self-esteem make a difference? Academy of
Management Journal, 43, 772–782.
46
Donnellan, M. B., Oswald, F. L., Baird, B. M., & Lucas, R. E. (2006). The mini-IPIP scales:
tiny-yet-effective measures of the Big Five factors of personality. Psychological
Assessment, 18(2), 192-203.
Edmondson, A. C., & Lei, Z. 2014. Psychological Safety: The History, Renaissance, and Future
of an Interpersonal Construct. Annual Review of Organizational Psychology and
Organizational Behavior, 1(1): 23–43.
Farh, J. L., Lee, C., & Farh, C. I. C. 2010. Task Conflict and Team Creativity: A Question of
How Much and When. Journal of Applied Psychology. 95(6): 1173-1180.
Frese, M., & Zapf, D. 1994. Action as the core of work psychology: A German approach.
Handbook of industrial and organizational psychology, Vol. 4 (2nd ed.). Palo
Alto, CA, US: Consulting Psychologists Press.
Grant, A. M., & Wall, T. D. 2009. The neglected science and art of quasi-experimentation: Why-
to, when-to, and how-to advice for organizational researchers. Organizational Research
Methods, 12(4), 653-686.
Greenberg, J., & Folger, R. 1983. Procedural justice, participation, and the fair process effect in
groups and organizations. In Basic group processes (pp. 235-256). Springer, New York,
NY.
Gucciardi, D. F., Crane, M., Ntoumanis, N., Parker, S. K., Thøgersen-Ntoumani, C., et al. 2018.
The emergence of team resilience: A multilevel conceptual model of facilitating factors.
Journal of Occupational and Organizational Psychology, 91(4): 729–768.
Hale, D., Ployhart, R. E., & Shepherd, W. 2016. A two-phase longitudinal model of a turnover
event: Disruption, recovery rates, and moderators of collective performance. Academy of
Management Journal, 59(3): 906–929.
Hamel, G., & Välikangas, L. 2003. The Quest for Resilience. Harvard Business Review, 81(9).
Harvey, S. 2014. Creative synthesis: Exploring the process of extraordinary group creativity.
Academy of Management Review, 39(3): 324–343.
Huang, X., Xu, E., Huang, L., & Liu, W. 2018. Nonlinear consequences of promotive and
prohibitive voice for managers’ responses: The roles of voice frequency and LMX. Journal
of Applied Psychology, 103(10): 1101–1120.
Idson, L. C., Liberman, N., & Higgins, E. T. 2000. Distinguishing gains from nonlosses and
losses from nongains: A regulatory focus perspective on hedonic intensity. Journal of
Experimental Social Psychology, 36(3): 252–274.
Kakkar, H., Tangirala, S., Srivastava, N. K., & Kamdar, D. 2016. The dispositional antecedents
of promotive and prohibitive voice. Journal of Applied Psychology. 101(9): 1342-1351.
King, D. D., Newman, A., & Luthans, F. 2016. Not if, but when we need resilience in the
workplace. Journal of Organizational Behavior, 37(5): 782–786.
Kontogiannis, T. 1999. User strategies in recovering from errors in man-machine systems. Safety
Science, 32(1): 49–68.
Kontogiannis, T., & Malakis, S. 2009. A proactive approach to human error detection and
identification in aviation and air traffic control. Safety Science, 47(5): 693-706.
LeBreton, J. M., & Senter, J. L. (2008). Answers to 20 questions about interrater reliability and
interrater agreement. Organizational research methods, 11(4), 815-852.
Ledford, G. E., & Mohrman, S. A. 1993. Self-Design for High Involvement: A Large-Scale
Organizational Change. Human Relations, 46(2): 143–173.
47
LePine, J. A. 2003. Team adaptation and postchange performance: Effects of team composition
in terms of members’ cognitive ability and personality. Journal of Applied Psychology,
88(1): 27-39.
LePine, J. A., Colquitt, J. A., & Erez, A. (2000). Adaptability to changing task contexts: Effects
of general cognitive ability, conscientiousness, and openness to experience. Personnel
Psychology, 53, 563–593
Li, A. N., Liao, H., Tangirala, S., & Firth, B. M. 2017. The content of the message matters: The
differential effects of promotive and prohibitive team voice on team productivity and safety
performance gains. Journal of Applied Psychology, 102(8): 1259-1270.
Liang, J., Shu, R., & Farh, C. I. C. 2019. Differential implications of team member promotive
and prohibitive voice on innovation performance in research and development project
teams: A dialectic perspective. Journal of Organizational Behavior, 40(1): 91–104.
Liang, J., Farh, C. I. C., & Farh, J. L. 2012. Psychological antecedents of promotive and
prohibitive Voice: A two-wave examination. Academy of Management Journal, 55(1): 71–
92.
Lin, S. H., & Johnson, R. E. 2015. A suggestion to improve a day keeps your depletion away:
Examining promotive and prohibitive voice behaviors within a regulatory focus and ego
depletion framework. Journal of Applied Psychology, 100(5): 1381–1397.
Mackenzie, S. B., Podsakoff, P. M., & Podsakoff, N. P. 2011. Challenge-oriented organizational
citizenship behaviors and organizational effectiveness: Do challenge-oriented behaviors
really have an impact on the organization’s bottom line? Personnel Psychology, 64(3):
559–592.
Maitlis, S. (2020). Posttraumatic growth at work. Annual Review of Organizational Psychology
and Organizational Behavior, 7, 395-419.
Marks, M. A., Zaccaro, S. J., & Mathieu, J. E. 2000. Performance implications of leader
briefings and team-interaction training for team adaptation to novel environments. Journal
of Applied Psychology, 85(6): 971–986.
Maynard, M. T., Kennedy, D. M., & Sommer, S. A. 2015. Team adaptation: A fifteen-year
synthesis (1998–2013) and framework for how this literature needs to “adapt” going
forward. European Journal of Work and Organizational Psychology, 24(5): 652–677.
Morgeson, F. P., Mitchell, T. R., & Liu, D. 2015. Event system theory: An event-oriented
approach to the organizational sciences. Academy of Management Review. 40(4): 515-537.
Morrison, E. W. 2014. Employee Voice and Silence. Annual Review of Organizational
Psychology and Organizational Behavior, 1(1): 173–197.
Morrison, E. W. 2011. Employee voice behavior: Integration and directions for future research.
Academy of Management Annals, 5(1): 373–412.
Parke, M. R., Tangirala, S., & Hussain, I. (2020). Creating organizational citizens: How and
when supervisor- versus peer-led role interventions change organizational citizenship
behavior. Journal of Applied Psychology. Advance online publication.
https://doi.org/10.1037/apl0000848
Podsakoff, N. P., Maynes, T. D., Whiting, S. W., & Podsakoff, P. M. 2015. One (rating) from
many (observations): Factors affecting the individual assessment of voice behavior in
groups. Journal of Applied Psychology. 100(4): 1189-1202.
Podsakoff, P. M., MacKenzie, S. B., Lee, J. Y., & Podsakoff, N. P. 2003. Common Method
Biases in Behavioral Research: A Critical Review of the Literature and Recommended
Remedies. Journal of Applied Psychology, 88(5): 879–903.
48
Preacher, K. J., & Kelley, K. (2011). Effect size measures for mediation models: quantitative
strategies for communicating indirect effects. Psychological methods, 16(2), 93.
Pritchard, R. D., Jones, S. D., Roth, P. L., Stuebing, K. K., & Ekeberg, S. E. 1988. Effects of
Group Feedback, Goal Setting, and Incentives on Organizational Productivity. Journal of
Applied Psychology, 73(2): 337–358.
Reeves, M., & Deimler, M. 2011, January 2. Adaptability: The new competitive advantage.
Harvard Business Review.
Rizzo, A., Ferrante, D., Bagnara, S., 1994. Handling human error. In: Hoc, J.M.,Cacciabue, P.C.,
Hollnagel, E. (Eds.), Expertise and Technology: Cognition &Human Computer
Interaction. Lawrence Erlbaum Associates, New Jersey, pp. 195–212.
Schmidt, F. L., Hunter, J. E., & Outerbridge, A. N. (1986). The impact of job experience and
ability on job knowledge, work sample performance, and supervisory ratings of job
performance. Journal of Applied Psychology, 71, 432–439.
Schweiger, D. M., & Denisi, A. S. 1991. Communication with Employees Following a Merger:
A Longitudinal Field Experiment. Academy of Management Journal, 34(1): 110–135.
Sellen, A. J. 1994. Detection of Everyday Errors. Applied Psychology, 43(4): 475–498.
Sherf, E. N., Sinha, R., Tangirala, S., & Awasty, N. 2018. Centralization of member voice in
teams: Its effects on expertise utilization and team performance. Journal of Applied
Psychology, 103(8): 813-827.
Stoverink, A. C., Kirkman, B. L., Mistry, S., & Rosen, B. (2020). Bouncing back together:
Toward a theoretical model of work team resilience. Academy of Management Review,
45(2), 395-422.
Stouten, J., Rousseau, D. M., & De Cremer, D. 2018. Successful organizational change:
Integrating the management practice and scholarly literatures. Academy of Management
Annals, 12(2): 752–788.
Subramaniam, M., & Youndt, M. A. 2005. The influence of intellectual capital on the types of
innovative capabilities. Academy of Management Journal, 48(3): 450–463.
Tangirala, S., & Ramanujam, R. (2008). Exploring nonlinearity in employee voice: The effects
of personal control and organizational identification. Academy of Management Journal,
51(6), 1189-1203.
Vallas, S. P. 2003. Why teamwork fails: Obstacles to workplace change in four manufacturing
plants. American Sociological Review, 68(2): 223–250.
Van Der Vegt, G. S., & Bunderson, J. S. (2005). Learning and performance in multidisciplinary
teams: The importance of collective team identification. Academy of management Journal,
48(3), 532-547.
Van Dyck, C., Frese, M., Baer, M., & Sonnentag, S. (2005). Organizational error management
culture and its impact on performance: a two-study replication. Journal of applied
psychology, 90(6), 1228.
Van Dyne, L., & LePine, J. A. 1998. Helping and voice extra-role behaviors: Evidence of
construct and predictive validity. Academy of Management Journal, 41(1): 108–119.
Waller, M. J. 1999. The timing of adaptive group responses to nonroutine events. Academy of
Management Journal, 42(2): 127–137.
Wei, X., Zhang, Z. X., & Chen, X. P. 2015. I will speak up if my voice is socially desirable: A
moderated mediating process of promotive versus prohibitive voice. Journal of Applied
Psychology, 100(5): 1641-1652.
49
Woolley, A. W., Chabris, C. F., Pentland, A., Hashmi, N., & Malone, T. W. (2010). Evidence for
a collective intelligence factor in the performance of human groups. Science, 330(6004),
686-688.
Zaheer, S., Albert, S., & Zaheer, A. 1999. Time scales and organizational theory. Academy of
Management Review, 24(4): 725–741.
Zapf, D., & Reason, J. T. 1994. Introduction: Human Errors and Error Handling. Applied
Psychology, 43(4): 427–432.
50
TABLE 1
Means, Standard Deviations, and Correlations (Study 1)
Variables
Mean
SD
1
2
3
4
5
6
7
8
9
10
1. Team size
5.26
.96
2. Pre-change performance
92.54
1.04
.13
3. Post-change performance
89.46
1.42
.14
.76**
4. Performance disruption
-6.15
.29
.09
.30**
.70**
5. Performance recovery
2.02
.30
.05
.02
.49**
.15*
6. Prohibitive voice (S1)
5.08
.59
-.05
.10
.17*
.08
.10
(.85)
7. Promotive voice (S1)
5.22
.60
-.00
.07
.14
.03
.24**
.58**
(.86)
8. Change intensity (S1)
5.37
.91
-.02
.04
.09
.05
.11
.03
.08
(.73)
9. Error management (S2)
5.31
.56
.07
.11
.22**
.31**
.01
.36**
.23**
.03
(.85)
10. Process innovation (S3)
5.11
.91
.03
.01
.20**
.08
.39**
.22**
.40**
.05
.15*
(.75)
Notes: N = 172 teams. Pre-change performance is the average team performance for the four months prior to the change event. Post-change performance is the average team
performance for the four months after the change event. Internal consistency reliabilities appear in parentheses along the diagonal. S1 = Survey1, S2 = Survey 2, & S3 = Survey 3
(see Figure 3a). * p < .05, ** p < .01.
51
TABLE 2
Discontinuous Growth Model Results on Team Performance (Study 1)
Model 1
Model 2
Model 3
Model 4
Model 5
Variables
Level 1 Predictors
Intercept
91.85**(.44)
91.83**(.44)
91.84**(.45)
91.88**(.44)
91.89**(.45)
Time
.01 (.02)
.01 (.02)
.01 (.02)
.01 (.02)
.01 (.02)
Disruption
-6.15** (.06)
-6.15** (.06)
-6.15** (.06)
-6.16** (.06)
-6.16** (.06)
Recovery
2.02** (.03)
2.02** (.03)
2.02** (.03)
2.02** (.03)
2.03** (.03)
Level 2 Predictors
Team size
.13 (.08)
.13 (.08)
.13 (.08)
.13 (.08)
.12 (.08)
Prohibitive Voice
.08 (.18)
.04 (.19)
.00 (.18)
-.01 (.19)
Promotive Voice
.15 (.17)
.17 (.18)
.17 (.17)
.19 (.18)
Change intensity
.05 (.09)
.06 (.09)
Prohibitive Voice*Change intensity
.39* (.16)
.38* (17)
Promotive Voice*Change intensity
-.16 (.16)
-.15 (.16)
Error Management
.12 (.15)
.03 (.16)
Process Innovation
-.04 (.10)
-.03 (.10)
Time*Prohibitive Voice
.06 (.05)
.06 (.05)
.06 (.05)
.06 (.05)
Time* Promotive Voice
-.08 (.05)
-.08 (.05)
-.08 (.05)
-.08 (.05)
Disruption*Prohibitive Voice
.18 (.13)
.03 (.14)
.09 (.13)
-.01 (.13)
Disruption*Promotive Voice
-.09 (.13)
-.11 (.13)
-.03 (.13)
-.05 (.13)
Disruption*Change intensity
.01 (.07)
.01 (.06)
Disruption*Error Management
.45**(.11)
.35**(.11)
Disruption*Process Innovation
.03 (.07)
.00 (.07)
Disruption*Prohibitive Voice*Change intensity
.37**(.12)
.27*(.12)
Disruption*Promotive Voice*Change intensity
.17 (.12)
.17 (.11)
Recovery*Prohibitive Voice
-.07 (.07)
-.01 (.07)
-.04 (.07)
-.00 (.07)
Recovery*Promotive Voice
.21** (.07)
.11 (.07)
.19**(.07)
.09 (.07)
Recovery*Change intensity
.04 (.04)
.04 (.04)
Recovery*Error Management
-.14*(.06)
-.12 (.06)
Recovery*Process Innovation
.16** (.04)
.17**(.04)
Recovery*Prohibitive Voice*Change intensity
-.11 (.07)
-.08 (.07)
Recovery* Promotive Voice*Change intensity
.02 (.07)
-.02 (.06)
-2 Res Log Likelihood (REML)
3449.12
3455.35
3437.53
3456.87
3447.04
df
17
25
31
34
40
AIC
3483.12
3505.35
3499.53
3524.87
3527.04
BIC
3571.92
3635.78
3661.13
3702.03
3735.30
Notes: N = 172 teams. Unstandardized coefficients are reported. REML=restricted maximum likelihood. AIC= Akaike Information Criterion; BIC= Bayesian Information
Criterion. Autocorrelation was included in all models. * p < .05, ** p < .01.
52
TABLE 3
Results of Team Level Path Analyses (Study 1)
Variables
Error management
Process innovation
Immediate Team
Performance lossesa
Subsequent Team
Performance gainsa
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Intercept
5.05**(.22)
5.14**(.22)
4.96** (.36)
5.06** (.35)
-6.86**(.26)
1.55**(.27)
Team size
.05 (.04)
.03 (.04)
.03 (.07)
.00 (.07)
.01 (.02)
.01 (.02)
Prohibitive voice
.34**(.08)
.27**(08)
-.04 (.13)
-.07 (.14)
-.03 (.04)
-.01 (.05)
Promotive voice
.02 (.08)
.05 (.08)
.63**(.13)
.66**(.13)
.00 (.04)
.06 (.05)
Change intensity
.00 (.04)
-.01 (.07)
.01 (.02)
.03 (.02)
Prohibitive voice* Change intensity
.29**(.08)
.01 (.13)
.12**(.04)
-.02 (.04)
Promotive voice* Change intensity
.01 (.08)
.29* (.12)
.06 (.04)
.02 (.04)
Error management
.12**(.04)
-.03 (.04)
Process innovation
.01 (.02)
.11**(.03)
R2
.14**
.21**
.16**
.19**
.18**
.18**
Notes: N = 172 teams. Unstandardized coefficients are reported. a The regression coefficients under these two columns are estimates of the within-team relationships that
variables coded as disruption and recovery had with team performance in our discontinuous growth models for each of the teams in the sample. By modeling how these estimates
are affected by different predictors in this team-level path analysis, we were able to derive direct and indirect effects of these predictors on how teams fared post-change. Here,
team performance losses were indexed such that the estimates are more negative when the losses are higher. Hence, a positive coefficient for any of the predictors of those
estimates implies a reduction in team performance losses. * p < .05, ** p < .01.
53
TABLE 4a
Manipulation Checks for Experimental Manipulations (Study 2)
Experimental
condition
Measured variable
Mean (s.d.) at T0
(Before intervention)
Mean (s.d.) at T1
(Early period after
intervention)
Mean (s.d.) at T2
(Later period after
intervention)
Mean differences
between T1 and T0
(T tests results)
Differences between
T2 and T0 (T tests
results)
Prohibitive condition
Prohibitive voice
4.70 (.73)
5.75 (.81)
5.71 (.91)
1.05 (8.58**)
1.00 (6.88**)
Promotive voice
4.68 (.66)
5.04 (.90)
5.19 (1.03)
.36 (2.85**)
.51 (3.06**)
Promotive Condition
Prohibitive voice
4.56 (.59)
4.93 (.79)
5.00 (.98)
.38 (3.19**)
.44 (3.02**)
Promotive voice
4.54 (.58)
5.60 (.82)
5.88 (.92)
1.06 (8.34**)
1.34 (8.55**)
Control Condition
Prohibitive voice
4.53 (.82)
4.71 (.93)
4.65 (.92)
.19 (1.75)
.12 (.89)
Promotive voice
4.55 (.82)
4.61 (.96)
4.76 (1.01)
.06 (.61)
.22 (1.86)
Notes: N = 88 teams (n=28 for prohibitive voice condition, n=31 for the promotive voice condition, and n=29 for the control condition). T0 = Time 1, T1 = Time 1, and T2 = Time
2. * p < .05, ** p < .01. TABLE 4b
Summary of Means of Key Variables by Experimental Condition (Study 2)
Experimental condition
Measured variable
Mean (s.d.) at T0 (Before
intervention)
Mean (s.d.) at T1 (Early period
after intervention)
Mean (s.d.) at T2 (Later period
after intervention)
Prohibitive condition
Error management
5.63 (.75)
5.48 (.77)
Process innovation
5.19 (.78)
5.19 (.86)
Team Performance
6.42 (1.17)
4.94 (1.32)
5.74 (1.41)
Promotive Condition
Error management
5.31 (.85)
5.25 (.89)
Process innovation
5.41 (.85)
5.43 (.87)
Performance
6.65 (1.23)
4.32 (1.49)
5.63 (1.48)
Control Condition
Error management
5.16 (.90)
4.92 (.63)
Process innovation
4.85 (.70)
4.82 (.75)
Team Performance
6.54 (1.11)
4.04 (1.28)
5.08 (1.56)
Notes: N = 88 teams (n=28 for prohibitive voice condition, n=31 for the promotive voice condition, and n=29 for the control condition). T0 = Time 1, T1 = Time 1, and T2 = Time
2. * p < .05, ** p < .01.
54
TABLE 5
Means, Standard Deviations, and Correlations (Study 2)
Variables
Mean
SD
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
1. Team size
3.93
.86
2. Member tenure (T0)
27.46
6.63
.04
3. Transformational leadership (T0)
5.24
.76
-.03
-.13
(.90)
4. Member conscientiousness (T0)
5.46
.50
-.02
.08
.21
(.77)
5. Member openness (T0)
4.83
.48
.01
.12
.20
.24*
(.81)
6. Change intensity (T0)
5.05
1.16
.15
-.02
.03
-.03
.17
(.72)
7. Prohibitive voice condition
.32
.47
.03
.09
.09
-.01
.20
.00
8. Promotive voice condition
.35
.48
-.11
-.14
-.07
.16
-.09
.00
-.50**
9. Error management (T1)
5.36
.85
.09
-.08
.03
-.02
-.03
-.01
.22*
-.05
(.89)
10. Process innovation (T1)
5.15
.81
.12
.15
.06
.12
.07
-.03
.03
.23*
.22*
(.80)
11. Error management (T2)
5.21
.80
.03
-.08
.10
.08
-.07
-.12
.23*
.03
.62**
.10
(.86)
12. Process innovation (T2)
5.16
.86
-.01
-.06
.25*
.13
.06
-.06
.03
.24*
.11
.44**
.21*
(.84)
13. Team performance (T0)
6.54
1.16
-.05
.07
.02
.02
-.03
.08
-.07
.07
-.13
-.14
-.03
-.11
(.80)
14. Team performance (T1)
4.42
1.40
.03
.06
.02
04
.04
-.12
.25*
-.06
.21*
.07
.20
-.06
.72**
(.87)
15. Team performance (T2)
5.48
1.49
-.05
-.04
.09
.18
.05
-.03
.12
.07
.13
-.06
.23*
.14
.66**
.78**
(.88)
Notes: N = 88 teams (n=28 for prohibitive voice condition, n=31 for the promotive voice condition, and n=29 for the control condition). Member tenure was on months. Internal
consistency reliabilities appear in parentheses along the diagonal. T0 = Time 1, T1 = Time 1, and T2 = Time 2. * p < .05, ** p < .01.
55
TABLE 6
Team performance as dependent variable (Study 2)
Team performance (T1)
Team performance (T2)
Variables
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Model 7
Model 8
Intercept
-2.57**
(1.64)
-2.19**
(1.47)
-1.70**
(1.47)
-5.13**
(1.51)
-1.03
(1.49)
-1.08**
(1.48)
.01 (1.45)
-2.89 (1.58)
Control variables
Team size
.10 (.12)
.10 (.11)
.15 (.10)
.04 (.10)
-.08 (.11)
-.08 (.11)
-.09 (.10)
-.09 (.11)
Member tenure
.00 (.02)
-.01 (.02)
-.01 (.01)
-.01 (.01)
-.02 (.01)
-.02 (.01)
-.03 (.01)
-.02 (.01)
Transformational leadership
-.03 (.14)
-.06 (.13)
-.07 (.12)
-.09 (.12)
.06 (.13)
.07 (.13)
.09 (.12)
-.03 (.13)
Member openness
.17 (.22)
.01 (.21)
.11 (.20)
.04 (.19)
.00 (.20)
.02 (.20)
.02 (.19)
.05 (.20)
Member conscientiousness
.04 (.21)
.07 (.20)
.00 (.19)
.07 (.18)
.45* (.19)
.41* (.20)
.49**(.18)
.39*(.19)
Team performance (T0)
.88** (.09)
.90** (.08)
.93** (.08)
.95** (.07)
.26* (.12)
.24 (.12)
.13 (.12)
.30* (.12)
Team performance (T1)
.68**(.10)
.69**(.11)
.78**(.10)
.66**(.10)
Main effects
Change intensity
-.19 (.13)
-.22 (.13)
Prohibitive voice condition
1.02**(.23)
.78 (.95)
.77**(.22)
.02 (.26)
-.89 (.92)
-.13 (.25)
Promotive voice condition
.18 (.23)
1.28 (.98)
-.02 (.22)
.19 (.23)
-3.47**(.96)
-.04 (.23)
Error management (T1)
.37** (.11)
Error management (T2)
.13 (.12)
Process innovation (T1)
.21 (.12)
Process innovation (T2)
.30*(.11)
Interactive effects
Change intensity * Prohibitive voice
condition
.05 (.18)
.16 (.18)
Change intensity * Promotive voice
condition
-.22 (.19)
.72**(.18)
R2
.53**
.62**
.67**
.69**
.67**
.67**
.72**
.70**
Notes: N = 88 teams (n=28 for prohibitive voice condition, n=31 for the promotive voice condition, and n=29 for the control condition). T0 = Time 1, T1 = Time 1, and T2 = Time
2. * p < .05, ** p < .01.
56
TABLE 7
Error management as dependent variable (Study 2)
Error management (T1)
Error management (T2)
Variables
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Intercept
5.95** (1.42)
6.11** (1.38)
6.00** (1.49)
5.59** (1.28)
5.49**(1.26)
5.68** (1.35)
Control variables
Team size
.09 (.11)
.10 (.10)
.10 (.10)
.01 (.09)
.03 (.09)
.04 (.09)
Member tenure
-.01 (.01)
-.01 (.01)
-.01 (.01)
-.01 (.01)
-.01 (.01)
-.01 (.01)
Transformational leadership
.04 (.12)
.02 (.12)
.02 (.12)
.10 (.11)
.10 (.11)
.10 (.11)
Member openness
-.05 (.20)
-.12 (.19)
-.13 (.20)
-.22 (.17)
-.25 (.17)
-.23 (.18)
Member conscientiousness
-.01 (.19)
-.01 (.19)
-.02 (.19)
.14 (.17)
.12 (.17)
.12 (.17)
Team performance (T0)
-.09 (.08)
-.08 (.08)
-.07 (.08)
-.25* (.10)
-.20 (.11)
-.19 (.11)
Team performance (T1)
.26**(.08)
.20* (.09)
.20* (.09)
Main effects
Change intensity
.03 (.13)
-.06 (.12)
Prohibitive voice condition
.51*(.22)
.60 (.95)
.39 (.22)
.24 (.85)
Promotive voice condition
.17 (.22)
.61 (.98)
.27 (.20)
-.13 (.88)
Interactive effects
Change intensity * Prohibitive voice condition
-.02 (.18)
.03 (.16)
Change intensity * Promotive voice condition
-.09 (.19)
.08 (.17)
R2
.03
.09
.09
.13*
.17*
.17*
Notes: N = 88 teams (n=28 for prohibitive voice condition, n=31 for the promotive voice condition, and n=29 for the control condition). T0 = Time 1, T1 = Time 1, and T2 = Time
2. * p < .05, ** p < .01.
57
TABLE 8
Process innovation as dependent variable (Study 2)
Process innovation (T1)
Process innovation (T2)
Variables
Model 1
Model 2
Model 3
Model 4
Model 5
Model 6
Intercept
3.54** (1.32)
3.40** (1.25)
3.13* (1.34)
3.75* (1.41)
3.57** (1.35)
5.16** (.09)
Control variables
Team size
.11 (.10)
.14 (.09)
.14 (.09)
.00 (.10)
.03 (.10)
.04 (.09)
Member tenure
.02 (.01)
.02 (.01)
.02 (.01)
.00 (.01)
.00 (.01)
.00 (.01)
Transformational leadership
.07 (.11)
.10 (.10)
.11 (.11)
.26* (.12)
.29* (.12)
.31** (.11)
Member openness
.02 (.18)
.03 (.17)
.04 (.18)
-.01 (.19)
.00 (.19)
.04 (.18)
Member conscientiousness
.16 (.17)
.05 (.17)
.05 (.17)
.15 (.18)
.04 (.18)
.09 (.17)
Team performance (T0)
-.10 (.07)
-.11 (.07)
-.11 (.07)
-.10 (.11)
-.09 (.11)
-.13 (.11)
Team performance (T1)
.02 (.09)
.00 (.10)
.03 (.10)
Main effects
Change intensity
.03 (.12)
-.21 (.12)
Prohibitive voice condition
.31 (.20)
1.03 (.85)
.32 (.23)
.19 (.86)
Promotive voice condition
.64** (.20)
.71 (.88)
.63** (.21)
-2.02* (.89)
Interactive effects
Change intensity * Prohibitive voice condition
-.14 (.17)
.02 (.16)
Change intensity * Promotive voice condition
-.01 (.17)
.52** (.17)
R2
.07
.18*
.19*
.08
.17*
.27**
Notes: N = 88 teams (n=28 for prohibitive voice condition, n=31 for the promotive voice condition, and n=29 for the control condition). T0 = Time 1, T1 = Time 1, and T2 = Time
2. * p < .05, ** p < .01.
58
Table 9 Summary of Results Across Studies 1 and 2
Hypotheses
Study 1
Study 2
1a
In the disruption stage, prohibitive voice is negatively related to team
performance losses.
Not supported
Supported
1b
In the disruption stage, prohibitive voice is negatively related to team
performance losses via error management.
Supported
Supported
2a
In the recovery phase, promotive voice is positively related to team
performance gains over time.
Supported
Not supported
2b
In the recovery phase, promotive voice is positively related to team
performance gains over time via process innovation.
Supported
Supported
3a
In the disruption phase, the effect of prohibitive voice on team performance
losses is stronger when change intensity is higher.
Supported
Not supported
3b
In the disruption phase, the indirect effect of prohibitive voice on team
performance losses via error management is stronger when change
intensity is higher.
Supported
Not supported
4a
In the recovery phase, the effect of promotive voice on team performance
gains is stronger when change intensity is higher.
Not supported
Supported
4b
In the recovery phase, the indirect effect of promotive voice on team
performance gains via process innovation is stronger when change
intensity is higher.
Supported
Supported
59
Disruption
Recovery
Post-change flux
Post-change
improvement in
performance
Post-change dip in
performance
Time
FIGURE 1A
Hypothesized baseline model: How change impacts team performance trajectories
Team
Performance
Change event
Pre-change stability
60
No change in
disruption
Enhanced pace of
recovery
Time
Change event
FIGURE 1C
Hypothesized effects of promotive voice (via process innovation) on team performance trajectories
during change
Post-change flux
Team
Performance
Pre-change stability
Post-change dip in
performance
Post-change
improvement in
performance
Dotted trajectory represents change
due to high promotive voice vis-à-
vis the baseline model
Dotted trajectory represents change
due to high prohibitive voice vis-à-
vis the baseline model
No change in the
pace of recovery
Post-change flux
Post-change
improvement in
performance
Post-change dip in
performance
Time
FIGURE 1 B
Hypothesized effects of prohibitive voice (via error management) on team performance trajectories
during change
Team
Performance
Change event
Pre-change stability
Less disruption
61
FIGURE 2
Role of Voice During the Change Process
Promotive voice
Team performance losses
immediately following change
Team performance gains over
time following change
Process innovation during the
recovery phase of change
Prohibitive voice
Error management during the
disruptive phase of change
Change intensity
62
Survey 2: Error management Survey 3: Process innovation
Survey 1: Change intensity, prohibitive voice, &
promotive voice
T1
T2
T3
T4
T5
T6
T7
T8
T1-T8 represent objective measurements of team performance. T1-T4 and T4-T5 are uniformly separated one month apart. T5 was
captured exactly one month after the resumption of team operations after the change event
FIGURE 3a
Measurement Timelines (Study 1)
Change event
Survey 2&3: Prohibitive voice, promotive voice, error
management, process innovation, and performance
Survey 1: Prohibitive voice, promotive voice, performance,
change intensity, and control variables
T0
T1
T2
T0 was approximately 1 week prior to the change event. T1 was approximately 1 week after the change event. T2 was approximately 2.5
weeks after T1
FIGURE 3b
Measurement Timelines (Study 2)
Change event
63
FIGURE 4
Team Performance over Time (Study 1)
Team performance
0
Team performance
FIGURE 5
Team Performance over Time (Study 2)
1
2
64
Note: To enhance interpretability of the figure, on y-axis, expected team performance losses were reverse-coded such
that higher values represent higher team losses.
5.95
6
6.05
6.1
6.15
6.2
6.25
6.3
Lower Higher
Prohibitive voice
Team Performance
Losses
FIGURE 6
The Interactive Effect of Prohibitive Voice and Change Intensity on Team Performanance
Losses (Study 1)
Change Intensity Lower
Change Intensity Higher
65
3
4
5
6
7
Lower Higher
Prohibitive voice
FIGURE 7
The Interactive Effect of Prohibitive Voice and Change
Intensity on Error Management (Study 1)
Change
Intensity
Lower
Change
Intensity
Higher
Error management
3
4
5
6
7
Lower Higher
Promotive voice
FIGURE 8
The Interactive Effect of Promotive Voice and Change
Intensity on Process Innovation (Study 1)
Change
Intensity
Lower
Change
Intensity
Higher
Process Innovation
66
1
2
3
4
5
6
7
8
9
Control Promotive
Team Performance Gains
FIGURE 9
The Interactive Effect of Promotive Voice and Change
Intensity on Team Performanance Gains (Study 2)
Change
Intensity
Lower
Change
Intensity
Higher
1
2
3
4
5
6
7
Control Promotive
Process Innovation
FIGURE 10
The Interactive Effect of Promotive Voice and Change
Intensity on Process Innovation (Study 2)
Change
Intensity
Lower
Change
Intensity
Higher
67
APPENDIX A: DETAILS OF CODING AND ANALYTICAL MODELS OF
DISCONTINUOUS GROWTH MODELING IN STUDY 1.
CODING AND INTERPRETATION OF CHANGE-RELATED VARIABLES
Measurement occasion
Interpretations
1
2
3
4
5
6
7
8
Variables
Pre-change
period
Post-change
period
Absolute
coding of
time
Time
0
1
2
3
3
3
3
3
Linear change rate of team performance in
the pre-change period
Disruption
0
0
0
0
1
1
1
1
Immediate performance losses due to the
change event
Recovery
0
0
0
0
0
1
2
3
Rate of change in team performance in the
post-change period
Relative
coding of
time
Time
0
1
2
3
4
5
6
7
Linear change rate of team performance
Disruption
0
0
0
0
1
1
1
1
Immediate performance losses due to the
change event relative to the pre-change
trajectory
Recovery
0
0
0
0
0
1
2
3
Rate of change in team performance in the
post-change period relative to the pre-
change trajectory
ANALYTICAL MODELS
We followed recommended procedures for discontinuous growth models (e.g., Bliese &
Lang, 2016) to test our hypotheses. The models factored in autoregressive residual correlation,
which captured the within-team correlation between team performance in adjacent periods.
In equation 1 (Model 1, Table 2) β2 indexes the initial performance losses after change
(Disruption) and β3 indexes subsequent performance gains after change (Recovery).
Equation 1
Level 1
Yti = β0i+β1Timeti+β2Disruptionti+β3Recoveryti + eti
Level 2
β0i = r00 + r01TimeSizei + uoi
β1i = r10 + u1i
β2i = r20 + u2i
β3i = r30 + u3i
68
In equation 2(Model 2, Table 2), r21 (the interaction term involving prohibitive voice and
disruption) tests Hypothesis 1a, and r32 (the interaction term involving promotive voice and
recovery) tests Hypothesis 2a.
Equation 2
Level 1
Yti = β0i+β1Timeti+β2Disruptionti+β3Recoveryti + eti
Level 2
β0i = r00 + r01TimeSizei + r02ProhibitiveVoicei + r03PromotiveVoicei + uoi
β1i = r10 + r11ProhibitiveVoicei + r12PromotiveVoicei + u1i
β2i = r20 + r21ProhibitiveVoicei + r22PromotiveVoicei + u2i
β3i = r30 + r31ProhibitiveVoicei + r32PromotiveVoicei + u3i
In equation 3 (Model 4, Table 2), r24 (the interaction term involving prohibitive voice,
change intensity, and disruption) tests Hypothesis 3a, and r35 (the interaction term involving
promotive voice, change intensity, and recovery) tests Hypothesis 4a.
Equation 3
Level 1
Yti = β0i+β1Timeti+β2Disruptionti+β3Recoveryti + β4TimeSizeti + eti
Level 2
β0i = r00 + r01TimeSizei +r02ProhibitiveVoicei + r03PromotiveVoicei + r04ChangeIntensityi +
r05ProhibitiveVoicei*ChangeIntensityi + r06PromotiveVoicei*ChangeIntensityi + uoi
β1i = r10 + r11ProhibitiveVoicei + r12PromotiveVoicei + u1i
β2i = r20 +r21ProhibitiveVoicei + r22PromotiveVoicei + r23ChangeIntensityi +
r24ProhibitiveVoicei*ChangeIntensityi + r25PromotiveVoicei*ChangeIntensityi + u2i
β3i = r30 +r31ProhibitiveVoicei + r32PromotiveVoicei + r33ChangeIntensityi +
r34ProhibitiveVoicei*ChangeIntensityi + r35PromotiveVoicei*ChangeIntensityi + u3i
Equation 4 was modeled to estimate the change in team performance due to the change
event for each team in our sample. For each team, the within-team estimate of β2 was used to
index immediate performance losses experienced by that team due to the change event and the
within-team estimate of β3 was used to index subsequent performance gains by that team. These
estimates were used as dependent variables in our team-level path analyses (See Table 3) and
helped us test Hypotheses 1b, 2b,3b & 4b.
Equation 4
Level 1
Yti = β0i+β1Timeti+β2Disruptionti+β3Recoveryti + eti
Level 2
β0i = r00 + r01TimeSizei +r02ProhibitiveVoicei + r03PromotiveVoicei +uoi
β1i = r10 + r11ProhibitiveVoicei + r12PromotiveVoicei +u1i
β2i = r20+ u2i
β3i = r30+ u3i
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Applied Psychology
This content is subject to copyright. Terms and conditions apply.
