Abstract
The upward voicing of ideas is vital to organizational performance. Yet power differences between voicers and those with authority may result in valuable ideas being overlooked. In this ethnographic, 31-month longitudinal study of a multi-disciplinary team in the healthcare sector, we examine how upwardly voiced ideas can endure to reach implementation. Of 208 upwardly voiced ideas, most were rejected in the moment, but 49 reached implementation despite appearing to be initially rejected. These ideas were kept alive by other team members who later drew upon and revived the initial ideas through what we call the voice cultivation process. We detail this process and describe five pathways through which voiced ideas stayed alive to reach implementation by overcoming different forms of resistance. We illustrate how the allyship of others can help voice live on beyond its initial utterance to reach implementation and generate change, even when the person who initially spoke up is no longer on the team or advocating for the idea. By reconceptualizing voice as a collective, interactional process rather than a one-time dyadic event, this paper develops new theory on how employees can help one another’s voice be heard to positively impact their teams and organizations.
Keywords
Upward voice—employees’ discretionary offering of constructive ideas for improving organizational or unit functioning to those with authority—is a vital pathway for valuable ideas to reach decision makers who might otherwise remain in the dark (e.g., Hirschman, 1970; LePine and Van Dyne, 1988; Detert and Burris, 2007). Harnessing improvement ideas voiced by employees at all organizational levels is critical for achieving a range of outcomes (e.g., Ashford, Sutcliffe, and Christianson, 2009; Morrison, 2011), from better organizational decision making and error detection (e.g., Morrison and Milliken, 2000) to improved implementation of new practices (Edmondson, 2003), increased employee satisfaction and motivation, and reduced turnover intent (Parker, 1993; for a review see Bashshur and Oc, 2015).
Despite the benefits of upward voice to organizations, employee voice remains rare and elusive (e.g., Morrison and Milliken, 2000). Scholars have identified numerous barriers that make speaking up both risky and difficult (e.g., Morrison and Phelps, 1999; Milliken, Morrison, and Hewlin, 2003; Premeaux and Bedeian, 2003; Grant and Ashford, 2008; Sherf, Parke, and Isaakyan, 2020). Power differences between voicers and those with the authority to implement their ideas often lead voicers to fear speaking up because of the potential for negative career consequences, such as being viewed negatively (e.g., Milliken, Morrison, and Hewlin, 2003) and receiving poor performance evaluations (e.g., Burris, Detert, and Romney, 2013).
A vast body of scholarship on upward voice has identified many characteristics and behaviors of authority holders and individual voicers that promote speaking up and reduce the likelihood of voicers being penalized. Authority holders such as those in supervisory or leadership roles can foster voice among employees by exhibiting openness (Detert and Burris, 2007), approachability (Milliken, Morrison, and Hewlin, 2003), and a willingness to take action (Edmondson, 2003), among other characteristics. These leaders are thought to play a critical role in encouraging voice by influencing employees’ perceptions of the benefits and consequences of speaking up (e.g., Edmondson, 1999; Edmondson and Nembhard, 2006; Srivistava, Bartol, and Locke, 2006; Sherf, Tangirala, and Venkataramani, 2019). Leader behaviors and leader–subordinate relationships can also create climates that inspire self-confidence and feelings of commitment that engender employee contributions (Keller and Dansereau, 1995; Kirkman and Rosen, 1999; Siegall and Gardner, 2000; Edwards and Collinson, 2002; Seibert, Wang, and Courtright, 2011).
Research on voicers has identified a host of characteristics associated with individuals choosing to speak up, such as conscientiousness, level of organizational identification, and autonomy (for reviews see Morrison, 2011; Morrison, 2014; Bashshur and Oc, 2015). More recent work has examined voicer behaviors, finding that when voicers frame ideas in ways considered more palatable to authority holders, such as framing ideas as opportunities for the organization or unit to improve (i.e., promotive voice) rather than as problems that are harming the organization or unit (i.e., prohibitive voice), they can mitigate negative career and job performance risks associated with voicing (e.g., Chamberlain, Newton, and LePine, 2017; Li et al., 2017).
This scholarship has made great strides in uncovering the antecedents and consequences of upward voice in organizations by identifying the factors that promote initial moments of voice and enable their frequency, as well as describing the consequences for voicers, supervisors, and unit performance. Yet without examining what happens to a voiced idea over time, the role that others aside from voicers and authority holders play in resisting and supporting it, and how and whether that voiced idea ultimately reaches implementation, our understanding of when and how upward voice makes real, material change in teams and organizations is incomplete.
First, focusing on voice as an individual’s dichotomous choice about whether or not to speak up prevents an understanding of whether and how that voice is heard and acted upon to make change, which is the fundamental purpose of voice (e.g., Hirschman, 1970). It is possible that voiced ideas that appear to be rejected in the moment may actually live on to positively impact the team and organization at a later point in time, even despite negative career consequences for the voicer (e.g., Fast, Burris, and Bartel, 2014). It is also possible that even if employees engage in upward voice, they still may not be listened to, especially given that lower-power people’s ideas are more likely to be ignored (e.g., Berdalh and Martorana, 2006; Magee and Galinsky, 2008; Fast, Burris, and Bartel, 2014; DiBenigno, 2018, 2020). The extant literature’s inability to explain whether and how voiced ideas reach implementation to effect the change they initially sought is a major omission and may be due to methodological limitations. Nearly all studies of voice use the employee rather than the voiced idea as the unit of analysis and are cross-sectional rather than longitudinal (Morrison, 2011). Most use surveys that examine employees’ and managers’ perceptions of whether an employee engages in voice generally and then link them to managers’ perceptions of that employee, employee turnover, and job or unit performance. Because there is reason to believe that voice unfolds as a process, as initially suggested by Dyck and Starke (1999), supplementing cross-sectional designs with longitudinal empirical investigations is vital to a full understanding of the voice phenomenon. Without longitudinal data, scholarship will remain focused on the factors that promote voicing and the immediate consequences. In so doing, we may neglect to uncover the important practices and processes through which a voiced idea can gain traction to achieve the ultimate purpose of voicing: to make change toward improving organizational and unit functioning.
Second, by focusing on voice as a dyadic exchange between voicers and the authority holders to whom their voice is directed, the existing literature has largely overlooked the possibility that others outside this dyad may affect whether and how voiced ideas get implemented. Given that nearly 80 percent of modern workers spend their day in collaborative activities with other group or team members (Cross, Rebele, and Grant, 2016), it is paramount to consider how the behavior of those outside the voicer–authority holder dyad, such as fellow group and team members, may enable or discourage voiced ideas from reaching implementation to bring about change. While research has documented the importance of considering voice at the group level, showing that group-level voice climates and structures emerge and impact individual voicer behaviors and team outcomes (Morrison, Wheeler-Smith, and Kamdar, 2011; Sherf et al., 2018), this research has not explored how group processes affect the trajectory of voiced ideas over time. Yet when voiced ideas occur publicly in the presence of other group or team members, those other members might play a critical role in affecting how voiced ideas are heard and carried forward to ultimately reach implementation. Such a contextualized view of voice is critical for developing theory that accurately reflects organizational reality (e.g., Bechky, 2011), lest we risk building theories “around shallow content” (Barley, 1996).
We present one of the first ethnographic, longitudinal studies of upward voice in teams to develop a process model explaining how other team members can help keep voiced ideas alive to reach implementation and make change. Over 31 months, we observed a multi-disciplinary healthcare team’s weekly meetings and conducted interviews with team members. We identified over 200 moments of voice that we then tracked longitudinally. As the existing literature would predict, most upwardly voiced ideas were rejected or ignored in the moment. However, 49 lived on and ultimately reached implementation by materializing into a new practice or process, in some cases being formalized across the entire organization. This analysis illustrates how voice can live on beyond its initial utterance through the allyship of others, even when the person who first spoke up is no longer on the team or advocating for the idea. We detail five pathways through which this voice cultivation process occurred and describe the specific practices that team members used to help one another’s voiced ideas be heard, revived, and sustained until they reached implementation. We also analyzed ideas that failed to reach implementation, despite evidence of their value when implemented elsewhere, to address concerns that quality differences in voiced ideas could explain why some lived on to reach implementation. Based on our findings, we outline an expanded paradigm for understanding upward voice that reconceptualizes voice to include a collective and social process that may operate in the many organizational contexts in which groups and teams are used to accomplish work.
Voice in Organizations
Voice was first theorized as a central means through which organizational members can improve their workplaces by speaking up to authority holders rather than exiting or enduring objectionable conditions (Hirschman, 1970). Since then, over a thousand articles have explored the association between voice and individual, unit, and organizational outcomes, identifying numerous benefits of voice (e.g., Morrison, 2011, 2014; Chamberlain, Newton, and LePine, 2017). At the individual level, these include higher job performance (Thomas, Whitman, and Viswesvaran, 2010; Ng and Feldman, 2012), a greater sense of control, reduced stress, and an improved job attitude (Morrison, 2011). At the organizational and unit levels, voice has been found to improve decision making (Dooley and Fryxell, 1999; Nemeth et al., 2001), reduce quit rates (Batt, Colvin, and Keefe, 2002), improve organizational learning (Beer and Eisenstat, 1996; Barker and Camarata, 1998; Levine, 2001), and enable innovation by stimulating divergent thought (De Dreu and West, 2001; De Dreu, 2002).
Yet despite the many positive outcomes associated with upward voice in organizations, voice is difficult to foster in the presence of power asymmetries between voicers and those with the authority to implement their ideas. Higher-power authority holders such as managers and supervisors often control valuable organizational resources and can directly impact the careers of lower-power voicers both positively and negatively, making speaking up risky (e.g., Morrison and Phelps, 1999; Milliken, Morrison, and Hewlin, 2003; Premeaux and Bedeian, 2003; Grant and Ashford, 2008). These dynamics make lower-power organizational members less likely to take speaking time or openly express their opinions in group discussions (Berdahl and Martorana, 2006; Detert and Edmondson, 2011), and the interests of higher-power organizational members tend to dominate (Magee and Galinsky, 2008). Multiple studies suggest voicer reticence is warranted; upward voice can be associated with negative career consequences for voicers, from lower job performance evaluations and retaliation to reduced promotions and salary increases (e.g., Siebert, Kraimer, and Crant, 2001; Milliken, Morrison, and Hewlin, 2003; Premeaux and Bedeian, 2003; Detert and Burris, 2007; Miceli, Near, and Dworkin, 2008; Fast, Burris, and Bartel, 2014). Voicers evaluate their circumstances and weigh the costs and benefits of speaking up before doing so (Morrison and Phelps, 1999; Grant and Ashford, 2008), and given these tradeoffs, many choose not to speak up at all (Milliken, Morrison, and Hewlin, 2003; Premeaux and Bedeian, 2003). These power differences may help to explain the mixed findings regarding the impact of voice for individuals; if managers perceive voice as disloyal, hostile, or offensive, then voice may lead to negative outcomes for employees, while voice that is appreciated by managers may lead to positive outcomes (Burris, Detert, and Romney, 2013).
Given the challenge of promoting upward voice in organizations, scholarship has sought to identify the antecedents that foster voice, focusing particularly on the characteristics and behaviors of two essential parties to voice: voicers and the authority holders to whom their voice is directed (e.g., Bashshur and Oc, 2015). Much of this work has examined the effects of voicers’ individual personality characteristics, attitudes, and perceptions on the likelihood of their speaking up. Individuals are more likely to voice if they are assertive or outgoing (Maynes and Podsakoff, 2014), proactive (Detert and Burris, 2007), and oriented toward achieving (LePine and Van Dyne, 2001). They are also more likely to voice if they feel positively about their jobs (Morrison, Wheeler-Smith, and Kamdar, 2011; Burris, 2012), feel obligated to contribute to change (Liang, Farh, and Farh, 2012), perceive support from others (Liang and Gong, 2013), and feel safe to take interpersonal risks (Edmondson, 1999; Detert and Burris, 2007).
Other work has examined individual voicer behavior and found that how ideas are framed affects voicing and job performance ratings of voicers by supervisors (e.g., Liang, Farh, and Farh, 2012; Burris, Rockmann, and Kimmons, 2017). For example, in a meta-analysis of 189 independent samples from 166 studies, Chamberlain and colleagues (2017) found that voicers who frame their ideas as an opportunity to do better in the future (promotive voice) receive better performance evaluations than those who frame their ideas as redressing current problems that are causing harm (prohibitive voice) (Liang, Farh, and Farh, 2012), although whether promotive framing helps voicers’ ideas reach implementation remains empirically undemonstrated. The findings from this literature are echoed in scholarship on issue selling, which describes the implicit theories managers have about how to strategically frame and sell issues to attract the organizational attention needed to get them on the top management team’s organizational agenda, such as sustainability or gender equity, without damaging their careers (e.g., Dutton and Ashford, 1993; Dutton et al., 2001). This work similarly suggests that framing issues in ways considered more positive and less threatening may result in fewer negative ramifications for those who raise them. But it is important to note that issue-selling research focuses on managers who already have some degree of authority and clout in their organizations due to their higher-power positions compared with lower-power upward voicers whose opportunities to be heard may be more constrained (e.g., DiBenigno, 2020).
Research that looks beyond the individual voicer has examined the role of authority holders’ characteristics and behaviors that encourage or restrain speaking up. Authority holders influence lower-power employees’ beliefs and decisions about whether to raise issues or ideas (Ashford, Sutcliffe, and Christianson, 2009). When authority holders are perceived as open, approachable, responsive, empowering, and inclusive, employees are more likely to voice their concerns and feel psychologically safe to speak up (e.g., Nembhard and Edmondson, 2006; Srivistava, Bartol, and Locke, 2006; Detert and Burris, 2007). Leaders can also encourage employee contributions by creating climates that inspire commitment and confidence that their contributions will be valued (Keller and Dansereau, 1995; Kirkman and Rosen, 1999; Siegall and Gardner, 2000; Edwards and Collinson, 2002; Seibert, Wang, and Courtright, 2011). Yet leaders may struggle to act in these ways because of pressures on them; for example, constraints on their time and attention may lead them to send inadvertent cues that they are not open or listening (Ashford, Sutcliffe, and Christianson, 2009; Sherf, Tangirala, and Venkataramani, 2019). Moreover, when employees do speak up, leaders may ignore or reject what is said if they feel threatened due to low self-efficacy (Fast, Burris, and Bartel, 2014), have implicit negative beliefs and biases about those in lower-power roles (Morrison and Milliken, 2000), or have inflated self-assessed competence (Morrison and Rothman, 2009).
The existing literature has made notable progress advancing our understanding of the antecedents and consequences of upward voice in organizations by closely examining voicers and authority holders. Yet extant voice scholarship makes two assumptions that may curtail the development of theory on voice that both reflects the lived reality of organizational members and fully captures the dynamics of the phenomenon of voice in organizations. First, voice scholarship presumes but has not systematically examined whether and how voicing results in actual change through the implementation of voiced ideas. This is a critical omission, especially since scholarship across the social sciences suggests the ideas of lower-power people are more likely to be ignored (e.g., Anderson and Berdahl, 2003; Berdhal and Martorana, 2006; Magee and Galinsky, 2008; DiBenigno, 2018, 2020). Second, voice scholarship has assumed that only the voicer and authority holder are critical for understanding voicing, even though the majority of employees spend their days working in collaborative encounters with others in groups, and teams are a dominant form of organizing work (e.g., Wageman, Gardner, and Mortensen, 2012; Detert et al., 2013; Cross, Rebele, and Grant, 2016). While some scholars have argued for the important impact that individuals speaking up might have on the experience and outcomes of an entire group (MacKenzie, Podsakoff, and Podsakoff, 2011; McClean, Burris, and Detert, 2013), no scholarship to date has examined the role that others witnessing a voiced idea may play in helping or hindering the idea to be acted upon to make change.
Regarding the first assumption, if the purpose of engaging in voice is to improve unit or organizational functioning, this can be achieved only if a voiced idea is acted upon and implemented (e.g., Hirschman, 1970). Yet no study of voice, to our knowledge, has tracked whether and how specific voiced ideas get implemented in practice. Instead, prior work has focused on uncovering the antecedents and consequences of voice—what promotes these moments, their frequency, and how they are framed—and the attributes of and consequences for voicers, authority holders, and unit performance. These studies have primarily employed cross-sectional surveys that assess perceptions of voice frequency, organizational voice climates, and leaders’ support for speaking up and have linked these factors with supervisors’ ratings of voicing behavior, direct reports’ performance, and unit performance (Morrison, 2011). But because scholarship has predominately leveraged cross-sectional designs using perceptual point-in-time survey measures, voice research has failed to uncover whether and how upwardly voiced ideas live on to be implemented and actually make change.
Regarding the second assumption that only the voicer and authority holder are critical to understanding voice, given the salience of power differentials that discourage upward voicing and make it risky for voicers to speak up, one way lower-power people may increase their influence on decisions is through forming coalitions and gaining allies’ support (e.g., Murnighan, 1978). Yet the role of others outside the dyad of the voicer and authority holder may have been overlooked given findings that targeting voice to non-authority holders such as peers is thought to be merely ineffective venting (Detert et al., 2013). And because many voiced ideas occur in team settings rather than the imagined private exchanges portrayed in most voice scholarship, the role of team members as allies—rather than as targets of voice—has not been examined. Indeed, scholarship on voice in groups and teams overall remains limited (Bashshur and Oc, 2015). The few studies examining voice in teams have focused on voice climates in groups, showing how higher levels of perceived acceptance of speaking up are associated with a higher likelihood of individuals voicing (Morrison, Wheeler-Smith, and Kamdar, 2011), resulting in more voice at the group level as well as improved performance (Frazier and Bowler, 2015). Other work has focused on group structure, finding positive benefits to team performance when voice is evenly distributed across team members rather than coming from only a few dominant members (Sherf et al., 2018). Of these few studies of voice in groups and teams, however, none has explored the group process that may unfold as ideas are voiced, heard, and responded to within the group or team.
Because voice in teams can occur publicly with other team members present, it may be inherently different from voice that occurs privately and is of critical theoretical significance for understanding whether and how voice is implemented. When an idea is voiced publicly, it becomes “common knowledge,” defined as when “everyone knows that everyone knows that everyone knows” what was said or done (e.g., Chwe, 2001; Adut, 2005; Centola, Willer, and Macy, 2005). Publicly voiced ideas cannot be unheard and so may have the potential to resurface later thanks to those who witnessed them. Actions or statements made at one point in time may endure as “traces” that can be drawn upon or recalled during others’ future interactions (Soderstrom and Weber, 2020). This suggests the common knowledge created by voiced ideas in a public, group, or team context may create the opportunity for team member allies to refer back to and support these ideas during subsequent team interactions. We build on these ideas to develop grounded theory that identifies how team members can help one another’s voiced ideas overcome resistance and live on beyond their initial voicing to improve team and organization functioning through what we call the voice cultivation process.
Data Collection and Analysis
Over 31 months from 2012 to 2014, we studied a multi-disciplinary team formed to improve care at an outpatient primary care clinic in a large healthcare delivery organization (“PeopleHealth”) in the northeastern region of the United States. The clinic served approximately 13,500 patients per year and was run by a staff of approximately 40 doctors (mostly part-time attendings, residents, and interns), managers, faculty members, nurses, medical assistants, and receptionists. To explore the voice process over time, we collected and analyzed transcripts from weekly meetings that the first author observed over 21 months. The first author also conducted formal and informal interviews with all team members, shadowed them during their work hours, attended team social and formal events, and observed PeopleHealth leadership team meetings. Throughout, we iterated among ongoing data collection, analysis, writing, and reflection (Corbin and Strauss, 2008). We began by qualitatively coding our data to identify moments of upward voice and to track those voiced ideas over time, from the initial voicing and response through the end of the study period to determine whether an idea was implemented. To analyze the vast amount of data generated by each voiced idea, we iterated between qualitatively inducing codes and quantifying those codes to observe their patterns across voiced ideas and over time (Creswell, 1998). Through this iterative approach, we induced a process model explaining the practices team members can use to help one another’s voice be heard and live on to reach implementation (Van de Ven and Poole, 1995; Langley et al., 2013).
Setting
Healthcare provides a rich context for studying upward voice because it is a field characterized by a distinct hierarchy of occupational roles differentiated across expertise and power (Nembhard and Edmondson, 2006; DiBenigno and Kellogg, 2014). Consistent with scholarship that conceptualizes upward voice as being directed to the team or individuals on the team that have the authority to implement a voiced idea, we define a higher-power role as one with formal authority and decision rights over whether voiced ideas receive resources, support, time dedicated to them, and approval needed to reach implementation (e.g., Hardy, 1996; Milliken, Morrison, and Hewlin, 2003). In our setting, doctors, faculty members, and managers had decision-making authority over clinic staff (e.g., nurses, medical assistants, and receptionists), services, and protocols, and hence they were authority holders with the power to implement voiced ideas on the team studied. No idea reached implementation without a doctor, faculty member, or manager first giving their approval. This conceptualization of power differences by role is also consistent with past work in the healthcare context in which doctors, faculty, and managers generally have markedly greater power than nurses, medical assistants, and receptionists (e.g., Friedson, 1970; Abbott, 1988; Harrison et al., 1992; Bate, 2000), and patient team members have the least power because they are considered organizational outsiders who lack clinical expertise and organizational resources (Institute of Medicine, 2001). See Table 1 for a description of power by role among the team members we studied.
Team Member Power by Role
Manager1 and Manager3 occasionally attended meetings as guests.
There was substantial support for the existence and importance of these well-established power differences by role in our data. For example, in the first year of the observation period, a nurse described the taken-for-granted power dynamic in clinic teams: “I like having Dr. C do that [lead the clinic team meetings]. Nurses and medical assistants, it’s not like we don’t want to do the work, but we need direction. Doctors need to tell us what to do. We are otherwise directionless.” In another example, we observed attending doctors telling resident doctors not to trust what patients say about themselves (e.g., what their own temperature is) even if the patient worked in healthcare or held advanced degrees. These combined characteristics of differentiated expertise in a deeply institutionalized power hierarchy by role make it likely that lower-power members may have valuable ideas and yet those ideas might be rejected, making this an ideal setting to study upward voice.
The team we studied also provided an ideal case to examine upward voice because it explicitly valued leader inclusivity and psychological safety, which prior scholarship suggests are conducive to fostering voice (e.g., Edmondson and Nembhard, 2006). The team was formed with the intent of improving the clinic; the team generated its own purpose statement: “To improve patient experiences and outcomes by building an efficient, creative, and joyful workplace.” Its multidisciplinary design was publicly acknowledged as critical to generating effective solutions to clinic-wide problems because of the unique expertise each team member brought by representing each role in the clinic; even patients were members of the team. As evidence of the team’s commitment to incorporating the views of multiple stakeholder groups, at its first meeting, the team discussed how every person could share their feedback and what might “hinder your being honest.” Leaders espoused the importance of psychological safety in interviews and in meetings, and all members participated in a training on the importance of psychological safety on teams. Finally, the team was not numerically dominated by either lower- or higher-power members, with equal numbers of higher-power members (doctors, manager, and faculty) and lower-power members (nurse, medical assistant, receptionist, and patient). It was thus striking that this team, like many others studied before, was predominately characterized by moments of failed upward voice by lower-power members of the team.
Data Collection
The first author observed team meetings and typed near-verbatim transcripts in real time, providing a detailed record of meeting interactions. By near-verbatim, we mean the first author sat with a laptop and attempted to type in real time everything that was said, correcting minor typos and small word omissions soon after the meeting. These transcripts were later qualitatively and quantitatively analyzed by mapping the trajectory of each voiced idea longitudinally. Observation began during the team’s first post-launch meeting in September 2012 and continued through June 2014, when the patterns related to upward voicing on the team became recognizably consistent and sufficient to support theory development. Throughout the study period, the team held weekly 120-minute meetings, of which 78 of 89 (88 percent) were observed. In addition, team e-mails and meeting minutes were collected for 87 of 89 meetings (98 percent). The first author also spent approximately 25 hours job shadowing at least one person from each role on the team during the workday in addition to attending conferences, all-staff meetings, and external visits with the team.
The first author also conducted frequent unstructured and semi-structured interviews with each team member before, during, and after the direct observation period to understand their perceptions of the team’s work and interactions as they unfolded, including regular unstructured interviews directly before and after meetings to elicit perspectives on specific meeting interactions. The researcher developed rapport with team members by joining them for meals and clinic events as well as accommodating their schedules for interviews by sometimes meeting outside the clinic. These efforts were particularly important for capturing the experience of patient team members because they were often not present at the clinic outside of team meetings. To understand the team’s organizational context, the first author acted as a participant observer of the organization-wide leadership team that developed and promoted the use of multi-disciplinary quality improvement teams. This leadership team met from May 2012 until June 2013, during which the researcher spent approximately 65 hours attending 42 meetings.
Analytic Approach
Our inductive analytical approach proceeded in three general stages. In the first stage, during the observation period, the first author wrote descriptive memos to capture and explore puzzling team interactions. It was particularly striking that there were discrepancies between higher-power team members’ descriptions of their intent to promote inclusivity and engagement among the lower-power team members and what occurred in practice. In meetings, higher-power members dominated the discussion, speaking more frequently and often outright rejecting or ignoring suggestions that lower-power team members made. As observations continued, the researcher wrote analytical memos and noted a pattern whereby most ideas from lower-power team members were rejected or ignored, and yet some would live on, sometimes even after the team member who raised them stopped advocating for them or had left the team.
To examine how and why some ideas were implemented while others failed, in the second stage of analysis, we engaged in systematic line-by-line coding of the meeting transcripts, tracing every voiced idea discussed over the 21 months of meetings, as well as semi-structured interview data related to these voiced ideas (Charmaz, 2006). We first identified and coded all 208 instances of voiced ideas in the data. In line with prior literature, we coded instances of upward voice using three criteria: (1) the individual stated or took an action demonstrating an idea that benefitted the team or organization’s functioning; (2) it was discretionary; and (3) it was directed to the team or an individual on the team who had the authority to make the change (Morrison, 2011). We then traced and coded what happened to each voiced idea: whether it was accepted, avoided, or rejected in the moment, and for those that endured, what happened after. Most instances of voiced ideas ended immediately after initial rejection, but 49 endured, living on in the team’s subsequent discussions to eventually reach implementation. As we analyzed and compared the trajectories of the ideas that were ultimately implemented, we began to notice the importance of others’ responses to these ideas, some of which occurred at later meetings. The first author went back and coded these responses, resulting in 18 sub-practices that were bucketed into seven higher-order categories that included two forms of resistance (avoiding and opposing) and five cultivation practices (amplifying, developing, issue-raising, exemplifying, and legitimizing) that helped a voiced idea to live on. We traced these ideas through to their ultimate conclusion through the observation period—either failing to reach implementation or reaching implementation, which included materializing the idea into a changed process or physical artifact, piloting the idea, or formalizing the idea by implementing it more broadly in the clinic or larger PeopleHealth parent organization.
In a third stage of analysis, we compared and analyzed the patterns of team members’ actions in response to voiced ideas and uncovered five different pathways exhibiting equifinality (e.g., Misangyi et al., 2017; Furnari et al., 2020) through which voiced ideas reached the same outcome, implementation. The pathways varied by the reason for and amount of resistance ideas faced, the number and type of dominant cultivation practices they underwent, and the time it took for an idea to reach implementation on average. Given the large number of voiced ideas and associated codes—we had 2,424 uniquely coded text fragments in our final analysis—we iterated between qualitative assessments of the patterns of practices and forms of resistance associated with individual upwardly voiced ideas and quantified depictions of these patterns for all voiced ideas (e.g., Creswell, 1998). We used our qualitative understanding of each voiced idea to catalogue the primary reason for resistance to it, often drawing on our interview data of privately disclosed concerns about voiced ideas to determine this. In quantifying the data, we counted instances of each practice, specifying the date it occurred and the role and power level of the person engaging in the practice. This produced counts of the total number and type of practices associated with resisting, cultivating, and implementing an idea, as well as how many weeks a voiced idea took to reach implementation. In addition, we visually charted each voiced idea over time and quantitatively characterized the pattern of practices employed by the voicer and allies of an idea to help it reach implementation over time or not. For an example visualization of a pathway, please see Online Appendix A (http://journals.sagepub.com/doi/suppl/10.1177/0001839220962795). We also leveraged data on a subset of failed ideas that faced similar amounts of resistance for similar reasons as successfully implemented ideas (e.g., whether the idea implied adding to versus removing ownership of core work from higher-power roles) and that were objectively verifiable as valuable ideas (e.g., because they were implemented successfully in other clinics). This comparative analysis enabled us to deepen our understanding of which practices mattered most for different types of ideas to reach fruition (see Online Appendix B for detailed descriptions of examples of failed ideas). Studying these failed ideas also allowed us to rule out alternative explanations—i.e., we did not find that failed ideas were systematically worse, were more difficult to implement, or happened earlier in the observation period—which we discuss in greater detail below. In assessing the emerging patterns, we noted the differing frequencies and combinations of practices over time through which voiced ideas were implemented; this led us to identify five pathways through which voiced ideas reached implementation, which we labeled the allyship, persistence, problematizing, co-crafting, and catalyzing pathways.
Findings
Consistent with prior literature, most ideas voiced by lower-power team members were rejected in the moment, being either ignored or quickly opposed and shot down. Of the 208 ideas upwardly voiced by lower-power team members over 21 months of team meetings, only 34 percent made it past the initial rejection, and even fewer (24 percent) were implemented. By contrast, though not the focus of our analysis, 78 percent of ideas suggested by higher-power team members were accepted in the moment they were first proposed.
An example of upward voice occurred in September 2012 as the clinic began co-locating groups: seating staff members by work group rather than by role. Receptionist1 voiced the idea to the team that receptionists should be included in the reorganization. Her idea was met with silence, and the discussion moved on. Another example occurred in November 2012, when Medical Assistant1 (MA1) voiced an idea to standardize the process for working in the clinic’s laboratory: “I would love to have a meeting with staff, medical assistants, and doctors to talk about their expectations and our expectations. . . . In the lab, people are doing things differently and need to standardize things.” Doctor1 and Doctor3 immediately opposed her idea by saying it was “less important,” ending discussion of it during that meeting. Such instances of rejection in the moment of initial voicing occurred for all lower-power members on the team and often appeared at the time to be failed attempts at voice, which is how existing scholarship on voice would have categorized them.
But by tracking voiced ideas longitudinally, we discovered that some ideas that had appeared to fail in the moment in fact lived on to be implemented and improve the team or organization’s functioning. Receptionist1’s idea to co-locate receptionists with clinic teams in “pods,” which had initially been met with silence, was piloted in 2013 and formalized in the clinic in 2014. This change led to significant improvements from the perspective of the team members; Receptionist2 described how it improved her clinical team’s communication and workflow in 2014, exclaiming, “Oh the pods, oh my God that’s like awesome. I love the pods.” MA1’s idea to fix the lab workflow became a formal protocol at the end of 2013 and was implemented clinic-wide in 2014, which resulted in major reductions in the time it took for MAs to receive doctor signatures and helped them complete related procedures and paperwork for patients by closing time.
We found that the voice literature’s prior explanations about what enables voice to be heard—i.e., how a voicer uses a promotive rather than prohibitive frame to improve how their ideas are initially responded to—did not explain whether an idea endured to create change. Of the ideas in our data that reached implementation, 60 percent had a promotive frame at the moment of initial voicing; see Online Appendix D. These data suggest that framing alone did not fully explain ideas’ trajectories on the team. Rather, by granularly tracing voiced ideas in our data over time, we uncovered a collective, social process through which lower-power team members’ voiced ideas lived on to reach implementation and have an impact, sometimes even after the voicer was no longer on the team or was no longer owning or advocating for the idea. We call this the voice cultivation process and describe a set of cultivation practices that members engaged in to resurrect ideas that had initially been shot down, breathing new life into them and helping them endure over time to reach implementation.
Sample List of Voiced Ideas That Reached Implementation
The Voice Cultivation Process
The voice cultivation process began with public voicing of an idea to the team that was initially rejected and resisted, followed by cultivation practices exhibited by team members that helped the idea live on to implementation; see Table 3. Below, we provide definitions and examples for each practice in this process.
Descriptions of Voice Cultivation Practices
Voicing
Consistent with the literature’s definition of upward voice (Morrison, 2011), voicing occurred when a lower-power team member, in this context a receptionist, medical assistant, nurse, or patient, voiced an idea or took a discretionary action in an attempt to improve the team or clinic’s functioning, which required the approval of those in more powerful roles (i.e., doctors, managers, and faculty) to be implemented.
Resistance
Most voiced ideas were rejected in the moment, which we call resistance. This took two forms. First, and most commonly, resistance occurred when team members avoided the voiced idea by ignoring it altogether, changing the subject, tabling the idea by saying it was off-topic or implying it was not sufficiently important to engage with presently, or false promising by appearing to agree but not following up. For example, when team member Patient2 raised the idea of reaching out to patients outside of typical clinic appointments to engage them in wellness interventions, such as in the “community, at churches, or in groups—I’d be happy to help with that,” the doctors, Nurse1, and Faculty1 ignored her comment and continued to discuss whether medical assistants should send e-mail or paper surveys to patients to reach out to them. A second way in which voiced ideas were resisted was through opposing, such as by directly or indirectly explaining why the voiced idea was not worthwhile, would not work, or failed before, or by defending a different option as better. For instance, in response to Receptionist1’s idea to co-locate receptionists with the clinic teams, Doctor3 directly opposed the idea by suggesting it was not worthwhile or important by saying, “We don’t need the front desk sitting with us,” after which Doctor2 added further opposition by saying the move would not improve their work.
Resistance to a voiced idea came from both higher- and lower-power team members, and it typically occurred immediately after an idea was initially voiced, although resistance could occur and recur at any point as voiced ideas were discussed at later meetings. Different types of ideas provoked different amounts of instances of opposing and avoiding based on the underlying reason a voiced idea was resisted. Voiced ideas that implied disruption to the core work of higher-power roles incited the most instances of avoiding or opposing, specifically ideas that implied that doctors, managers, or faculty would either need to take on additional work or relinquish ownership of work they considered core or important to their role. On average, there were five instances of avoiding or opposing in response to voiced ideas in which the underlying reason for resistance was concern about a disruption to the core work of those in higher-power roles, compared with an average of only 1.5 instances of opposing or avoiding in response to voiced ideas that did not disrupt the core work for those in higher-power roles.
Cultivation
Voiced ideas that endured to ultimately reach implementation after initial rejection underwent a cultivation process in which team members engaged in practices that helped one another’s ideas be heard and stay alive despite initially being avoided or opposed. Cultivation consisted of five common practices that supported a voiced idea—amplifying, developing, issue-raising, exemplifying, and legitimizing—which were engaged in by team members or in some cases the original voicer in support of their own idea.
Amplifying encompassed comments that revived a voiced idea by recalling it publicly at a later time, often using emotional or urgent language. For example, in September 2012, Nurse1 voiced her idea to fix and standardize how receptionists transfer information to nurses, soon after which she took maternity leave for three months. While she was out, Receptionist1 amplified Nurse1’s idea by publicly recalling it, which prompted Nurse2 and Doctor1 to talk about Nurse1’s idea the following week. Receptionist1 repeatedly amplified the idea by reviving it through November and December, often using emotional language, such as by saying that receptionist–nurse phone communication “is frustrating. It takes time. There’s lots of repetition and wasted time.” The team had begun implementing Nurse1’s idea by the time she returned from maternity leave in January.
Developing involved team members publicly acknowledging or asking clarifying questions about a voiced idea, which helped others hear and make sense of it. For instance, Doctor2 kept alive the discussion of Medical Assistant1’s idea about fixing ineffective laboratory test processes in the clinic by asking a clarifying question about why medical assistants were working in the lab after hours: “Is there a reason you can’t say no lab results after 30 minutes from when the medical assistant is supposed to leave?” Patient1 further developed the idea by asking MA1, “Are you getting lab requests after hours? At what time?” These questions helped the team realize that the problem was related to doctors and that medical assistants lacked power to fix the inefficiencies with the lab process on their own. Developing was distinctive in that it was a prominent practice in the cultivation of nearly every voiced idea that went on to reach implementation.
Issue-raising occurred when a team member publicly called out a voiced idea’s specific weakness, which created an opening for others to address the weakness or propose alternatives. For example, when Receptionist2 proposed addressing poor conditions in the waiting room for kids, Manager2 began to grapple with related weaknesses associated with the idea and explained them in detail: “No stuffed animals since they carry germs . . . any toys would have to be cleaned every single night, and you can’t guarantee that will happen.” Knowing the specific issues with this idea helped Receptionist2 and the team later generate more viable solutions that addressed Manager2’s concerns: they proposed and decided to install an easily cleaned aquarium and blackboard in a designated pediatric area.
Exemplifying involved taking personal initiative to embody or demonstrate a voiced idea by developing the skills or doing the work to provide evidence that the idea was tractable. Exemplifying occurred when Receptionist1 embodied her own voiced idea to increase receptionists’ training, access, and authority by allowing receptionists to book sick visits (which only nurses had been doing) by showing the team she had the capability to think through the decision-making process involved in executing this idea. She told them, “A patient called and said, ‘I have two children, and they are both sick.’ I asked about fever, and she said, ‘99.5 and 100.5.’ I would have just given them the appointments, but I couldn’t put them in there. . . . If instead you called in and said ‘I have leg pain,’ I’d ask right or left and how long it’s lasted. If it’s been two weeks, I don’t think it’s urgent like a fever, so I can book you for the next day or the following or else I’d send it to the nurse to triage.” By exemplifying the decision-making capabilities necessary to decide how to book patient appointments, Receptionist1 helped convince the team that, with guidelines and training, receptionists would be capable of appropriately booking some sick visits instead of only nurses doing so.
Legitimizing included publicly vouching that a voiced idea made sense, solved an important problem, could be achieved, or had been successfully implemented elsewhere, often by using specific examples or making it one’s own without citing the original source. For example, in January 2013, Patient1 voiced a concern that the team did not communicate effectively with the clinic and suggested ways to use different types of media to communicate, such as visuals, by saying, “We need to be doing things in multiple media and on a consistent basis.” Her comment was ignored in the moment but was legitimized the following month by Doctor1 who described her visit to another clinic: “The room we walked into, it had a grid; it looks like the White House war room, it was incredible. There were fish bone diagrams; there were big things they were working on.” Doctor1’s public legitimizing of Patient1’s idea helped to support the use of big, easy-to-read visual displays during future projects to improve communication between the team and clinic.
Implementation
Implementation occurred when a voiced idea was realized in practice in the team, clinic, or entire organization. We observed three types of implementation, with increasing levels of impact on the organization. The first, which we refer to as materializing, occurred when a voiced idea “became real” through the creation of a physical document (e.g., a new checklist) or action in which the idea was implemented (e.g., MA1 leading a huddle for the first time), most commonly by team members or within the bounds of the team. The second type of implementation was piloting, where an idea was socialized, refined, and tested beyond the team in the broader clinic. The third type was formalizing, where an idea became routinized as the standard way that the team, clinic, or organization operated and came to be taken for granted as the default way of operating. Often one voiced idea achieved multiple types of implementation, such as first materializing and then piloting.
Five Pathways for Cultivating Voice
In examining how different voiced ideas reach implementation over time, we identified several sequences and frequencies of specific cultivation practices that varied based on the different reasons for resistance to a voiced idea. In analyzing these sequences and frequencies, we uncovered five pathways through which voice reached implementation after facing initial rejection: allyship, persistence, problematizing, co-crafting, and catalyzing pathways. Table 4 describes the five pathways, and Online Appendix C illustrates them visually. All of the voiced ideas in our data that ultimately reached implementation were characterized by following one of these five equifinal pathways; the voiced ideas that failed to reach implementation did not follow any of the five pathways. The specific pathway that characterized a voiced idea’s trajectory was linked to the underlying reason it provoked resistance, particularly deriving from whether and how the idea implied changes to higher-power team members’ work. The allyship and persistence pathways were associated with voiced ideas that did not imply disruption to higher-power team members’ core work, whereas the problematizing and co-crafting pathways were common for voiced ideas that did imply such disruption. The catalyzing pathway differed in that it occurred when a shock acted as a catalyst for resurrecting a voiced idea that had been forgotten and stalled, providing an opportunity for that idea to be revived, legitimized, and developed anew. Ideas followed clear trajectories: once on a pathway, they continued that way rather than moving across pathways. Below we provide a detailed example of one voiced idea that reached implementation along each pathway, and Online Appendix D describes each of the 49 voiced ideas we identified, organized by pathway.
Characteristics of the Five Voice Cultivation Pathways
Average number of each cultivation practice listed in parenthesis for voiced ideas within each pathway.
We coded cultivation practices by team when three or more individuals were part of a single instance of a practice; almost every instance was during developing (asking follow-up questions) and implementation (multiple people helped materialize or pilot). These team-level practices accounted for 6% to 11% of practices across the pathways.
Allyship pathway
In the allyship pathway, a voiced idea was kept alive through the support and engagement of other team members. Although resistance was relatively minimal on this pathway as compared with other pathways, averaging only two instances of opposing or avoiding per voiced idea, the reasons for resistance on this pathway were distinctive in that they tended to be rooted in other team members’ assumptions that the voiced idea was not feasible or important. Examples include Receptionist2’s idea to give clinic receptionists a role in team huddles, which was seen as not important, and Medical Assistant2’s idea to offer training for all roles in the clinic for a colon screening project, which was seen as infeasible. Given the reasons for resistance to these ideas, legitimizing was a particularly central practice to overcome such resistance because allies stepping forward to vouch for an idea’s importance and feasibility was vital to keeping the fledgling idea alive. The dominant practice of legitimizing was complemented by the cultivation practices of developing and issue-raising, which further helped voiced ideas gain traction despite others’ initial feelings that the ideas were infeasible or unimportant. This was one of the fastest pathways, averaging 41 weeks for a voiced idea to reach implementation.
An example of this pathway is how MA2’s voiced idea for medical assistants to lead huddling overcame feasibility concerns to be fully implemented and formalized across the entire clinic. Soon after MA2 joined the team in September 2013, the team was lamenting that huddling—short morning and afternoon check-in meetings for doctors, nurses, and medical assistants working together—was not routinely occurring in the clinic despite huddling being a nationally recognized best practice for improving care delivery. The team attributed the root cause to doctors refusing to do the work required to prepare for and follow up after these meetings and often just not showing up. During a meeting, MA2 voiced a novel idea to make medical assistants responsible for huddling, noting that she already did much of the prep and follow-up work of huddling anyway. She explained, “It’s helpful to have my own notes on something like this . . . it’s helpful to know how to room [patients] without going into [the medical record].” However, her idea was avoided in the moment, and the team spent the rest of that meeting and the next two meetings brainstorming how to get doctors to start leading huddles.
MA2’s idea was revived three weeks later when Doctor1 referred back to it, saying, “Should a medical assistant run it?” to which the team defaulted back to discussing how doing so would fail because doctors would not feasibly buy in. Later in the meeting, Manager2 legitimized the information MA2 shared in the first meeting about already doing the work of huddling by saying, “I’m sorry, I don’t mean to talk, but we need to empower medical assistants if we want this to work. Poor MA2 is doing all the work on the huddle list. If we want MA2 or the other medical assistants to be calling the shots, they need to be able to say: ‘We are going to huddle now, let’s go.’” Doctor1 then developed the idea further, asking, “What do you think about who should be running the huddle?” and MA2 responded that “a medical assistant would work as well (as a doctor) because they are more hands on . . . it’s doable for the medical assistants.” Doctor3 then engaged in issue-raising, sharing his concern about whether it is feasible for medical assistants to serve in a leadership role: “Not everyone is comfortable stepping into a leadership role, even for 5 minutes. That’s a lot to ask for some. . . . I wish we had a [doctor] huddle champion in the current system.” Doctor1 then shared a solution to this problem by suggesting adding formal structures to help medical assistants: “We need to set expectations that medical assistants run [the huddles] . . . [by having] a structure where there is a checklist for medical assistants,” to which Doctor3 responded positively.
The idea reached implementation in the form of materializing when the team spent the remainder of the meeting drafting a new written checklist and procedure for how medical assistants should lead huddles in the clinic. At subsequent meetings (weeks 4–7), Nurse1, Doctor1, Doctor3, MA2, and others helped further legitimize and develop the idea by raising and solving issues about the change and by providing feedback on how the new procedure was working. They then further implemented the idea by piloting it in the broader clinic, during which medical assistants were granted authority to page doctors to remind them to attend. Nine weeks after this initially voiced idea, medical assistant–led huddling had been formalized; it diffused throughout the clinic and was institutionalized as the default way the clinic operated. For instance, a permanent box for huddling was added to clinic whiteboards to increase public accountability of the practice, and six months later huddling became part of the formal job expectation that was shared with all new medical assistants during onboarding through training on the standardized checklist for leading huddles.
Although MA2’s idea was initially ignored for three weeks, it was recalled by Doctor1, legitimized by Manager2, and developed by the team. This example illustrates how voiced ideas that may appear to have failed in the moment can be accessed and revived at a later point and shepherded to implementation by the allyship of others. We uncovered 13 other voiced ideas that followed the allyship pathway in our data.
Persistence pathway
In the persistence pathway, a voiced idea was kept alive primarily through the persistence of the idea’s original voicer but with at least some cultivation by other allies on the team. As with the allyship pathway, ideas on this pathway faced relatively minimal resistance; they faced an average of one instance of opposing or avoiding per voiced idea, and they did not threaten to change the work of higher-power doctors or managers. Unlike the allyship pathway, the common reasons for resistance to these ideas were other team members’ lack of awareness of the problem or lack of understanding of other roles’ work and skills. For example, ideas on this pathway included recognizing and using lower-power team members’ computer skills, which other roles did not understand or appreciate, and creating signs to improve the flow of interactions with receptionists, which other team members were not aware was a major problem.
Because there was a lack of awareness and understanding of what these ideas entailed, they reached implementation through the repetition and actions of the ideas’ original voicers and their dogged personal initiative to take action outside team meetings to raise awareness of the problems or demonstrate the types of work or skills needed to address them. Exemplifying practices enacted by the original voicers were thus a central feature of this pathway: voicers personally demonstrated the work or skills needed to solve problems about which other team members were unaware, aligning with the saying “it’s better to ask for forgiveness than permission.” But voicers could not fully pursue their ideas alone; it was vital that at least some other team members also cultivate them. This occurred primarily through amplifying an idea (i.e., bringing it up later) and developing it (i.e., asking questions to understand it in greater detail). For example, after Receptionist1’s idea to improve patient flow to receptionists and auxiliary services by creating signs was met with silence, she independently created and posted sample signs with her own manager’s permission; once others became aware of how well the signs were working to solve a problem they previously did not know was even an issue, the team was willing to engage more with her idea. Voiced ideas that reached implementation on this pathway took somewhat longer than those on the allyship pathway, averaging 52 weeks.
The persistence pathway is illustrated in how Receptionist1’s voiced idea to give a lower-power staff member a place at Clinic Leadership team meetings reached implementation over 38 weeks. Receptionist1 shared her idea at a team meeting after lamenting that Clinic Leadership had rejected a form she created for patients, noting that their decision negatively impacted her daily work: “Clinic Leadership . . . shot it down. To this day, I am still writing it down each time because they said no. . . . I would like to see one of us [staff] be able to go to a Clinic Leadership meeting and have an opinion. A lot of decisions are made there, and we don’t have a say.” This idea was quickly opposed by Faculty1, who said that staff had been invited to vote on this idea but “not enough people said yes.” At this moment, MA1 stepped forward to help cultivate Receptionist1’s idea by amplifying it, sharing that “it should have been mandatory” to attend a session where the idea was discussed and suggesting that it was not strongly communicated that staff were welcome. Receiving the support of at least one ally was a critical aspect of this pathway, but even so, Faculty1 avoided further conversation by tabling the idea, claiming it was irrelevant to the purpose of that particular team meeting to decide on projects: “This is a specific conversation we want to come back to, but how is it related to picking a project?” Faculty1, who trained doctors but did not work with patients in the clinic, participated in Clinic Leadership meetings and believed she spoke on behalf of the entire team, including staff, when she participated in those meetings. But Receptionist1 said in an interview that she believed Faculty1 was not aware of the day-to-day problems staff faced, did not understand the complexity of the staff’s work, and could not effectively represent staff concerns and interests at the Clinic Leadership meetings.
Receptionist1 seized an opportunity to exemplify her idea at a meeting two weeks later by volunteering to liaise with a member of Clinic Leadership for scheduling. When Faculty1 asked for someone with scheduling skills, Receptionist1 responded, “I could meet with the operations manager to figure out who can do this.” After some back and forth, Faculty1 agreed to allow Receptionist1 to talk with the operations manager but asked her “not [to] talk to her before I talk to [another member of Clinic Leadership].” At the next team meeting a week later, it happened that Nurse2, who had recently joined the team, suggested that a nurse and medical assistant join the Clinic Leadership meetings, and Receptionist1 took the opportunity to further amplify her idea, adding “And a receptionist.” The idea was again opposed by Faculty1, who noted that the idea had failed in the past: “We tried that, and anything substantial needs to be approved by Clinic Leadership.” In the next meeting, Receptionist1 pressed on, describing how she had exemplified successfully liaising with a Clinic Leadership member about a minor question that was raised at the last team meeting. In the meeting following that, MA1 amplified Receptionist1’s idea by repeating it in response to disappointment over another Clinic Leadership decision that had adversely affected staff, saying, “What about being in the Clinic Leadership meeting?” This time, instead of closing off the discussion, Faculty1 replied, “That is not a closed conversation.”
In the weeks that followed, Receptionist1 continued exemplifying her idea by directly connecting with members of Clinic Leadership on minor questions that came up on the team. For instance, at a meeting 12 weeks after she first voiced her idea, Receptionist1 took it upon herself to represent the front desk to the team and Clinic Leadership to get permission for decorating their workspaces before the December holidays. Eventually Faculty1 started to act as an ally by publicly acknowledging the intermediary work Receptionist1 had been doing informally in connecting with the operations manager to represent the team. As noted above, having at least minimal support from allies was necessary on this pathway, even though the original voicer played the dominant role in cultivating her own idea. Over the coming weeks, as Receptionist1 continued exemplifying her idea by liaising with Clinic Leadership, the team began to take her word on the operations of the Clinic Leadership team, as noted when Patient1 asked, “Do we need to have Clinic Leadership bless this?” and Receptionist1 replied, “Nope . . . they told me just to do it.”
Nine months after voicing her idea, the exemplifying practices of Receptionist1 as an effective staff intermediary with leadership resulted in its implementation. A new formal role of “sub-manager” was created, which granted her a regular seat at the Clinic Leadership team meetings to represent staff concerns. (After the observation period, the Clinic Leadership team also offered MA2 a similar position to attend the meetings.) In addition to this example, we observed nine other cases of voiced ideas reaching implementation through the persistence pathway.
Co-crafting pathway
The co-crafting pathway differed markedly from the allyship and persistence pathways. Co-crafting arose when voiced ideas were disruptive to higher-power team members’ core work, specifically because the ideas required doctors or managers to relinquish work over which they felt ownership. These ideas faced much greater initial resistance—four instances of opposing or avoiding per voiced idea on average, which is twice as much as the allyship and persistence pathways. Examples of voiced ideas that reached implementation through this pathway include an idea to stop doctors from driving every project, including staff projects, so that those in roles most affected by the project could sometimes drive, and an idea to allow non-doctors to reach out to patients outside of typical clinical appointments instead of only doctors owning this work. The co-crafting pathway was unique in that it kept voiced ideas alive by engaging the team in collectively crafting how to remove or delegate work away from higher-power roles to lower-power roles such that higher-power team members—who might feel threatened by the change—were able to get on board with giving up ownership of a task. This process took time to reach implementation—58 weeks on average—and relied on a mix of amplifying and developing practices, with some legitimizing, to move the idea forward.
The co-crafting pathway is illustrated in how MA1’s idea to allow non-doctors to facilitate the clinic’s monthly Population Health sub-team meetings resulted in the implementation of a new structure across the clinic in which receptionists were assigned to lead these meetings. For seven months, the team had struggled to effectively manage clinic Population Health meetings, which the team helped create and was overseeing. At these monthly meetings, 12 groups composed of doctors, residents (doctors in training), nurses, medical assistants, and receptionists who worked with the same groups of patients would come together to determine which of their patients should be contacted for screenings (e.g., mammogram, colonoscopy) or appointments (e.g., physical, vaccine). Clinic leadership issued a survey that found these meetings were nearly unanimously considered inefficient and ineffective, despite the importance of the task to patient care. A central problem was that residents were assigned to lead the meetings as a leadership-building opportunity, but they did not know the patients well or were rotating at another hospital some weeks, so most struggled to lead the meetings. Nurses, medical assistants, and receptionists said they felt the meetings added no value above what they already knew. The voiced idea to have non-doctors lead Population Health meetings was similar to the earlier example of MAs leading quick daily huddles in lieu of doctors, but it was considered disruptive to doctors’ work because doctors said they felt ownership over these longer and more complex decision-making meetings and believed it was their responsibility to lead them.
During a team meeting in May 2013, Nurse1 explained how boring she found the Population Health meetings. In response, MA1 voiced the idea that perhaps medical assistants could run the meetings, since they often had to keep track of individual patient needs. She exclaimed, “I could do it. I could lead.” Her idea was acknowledged by Doctor3 but not acted upon, and it continued to be swept aside and ignored in the weeks to come. A few weeks later, MA1’s idea was legitimized externally when Doctor2 shared how clinic leadership would like nurses or MAs to rotate leading the Population Health meetings, given continued complaints about the meetings. However, for the next two weeks the conversation on this topic continued to ignore the idea of non-doctors leading the meetings, and instead Faculty1 and Doctor1 tried to improve doctors’ performance at leading the meetings.
For the next 27 weeks, the Population Health meetings continued with doctors at the helm, and the problems persisted. At a meeting in December, the team continued to ignore the possibility that other roles could lead these meetings, and doctors on the team discussed why certain clinic doctors were doing a poor and inconsistent job leading, but the pattern of repeated use of three dominant cultivation practices—developing, amplifying, and legitimizing—continued. Faculty1 used developing by clarifying a statement about why doctors didn’t show up and then amplified the idea by stating that sometimes doctors “can’t be at the meeting. Do we need someone leading the meeting who’s there?” But Doctor1 and Doctor3 returned to strategizing about how to get doctors to serve as “champions” and run Population Health meetings.
Eighteen weeks later, the idea of non-doctors leading the meetings resurfaced when the team was putting a skit together to demonstrate to the entire clinic best practices for running Population Health meetings. Doctor1 asked who should be playing the part of leading the meeting, to which MA2 responded dejectedly, “A doctor.” Doctor1 then legitimized MA1’s idea from long ago by responding, “You know, it doesn’t have to be a doctor.” Nurse1 amplified the idea the following week in response to a question about who should be running the computer during the Population Health meetings, saying, “It could be rotated so everyone learns and takes the chance . . . so it could be a team member [non-doctor] leading each one.” In response, Faculty1 developed Nurse1’s comment by asking a clarifying question about whether she meant team members would lead the whole meeting or just run the computer, to which the nurse replied that it would be everyone rotating for both, which was then legitimized by others as “a really good idea.” Three weeks later when the idea was brought up again, it was attributed to Nurse1 (and not MA1) when Receptionist1 said to Nurse1, “Your idea was to have everyone [lead],” to which the nurse responded, “Yeah, I forgot about that, that was my idea.” Doctor3 still opposed it, suggesting that sub-managers be brought in to lead instead. But Doctor1 interjected that there weren’t enough sub-managers to run all the meetings simultaneously and suggested Faculty1 socialize the idea of non-doctors running the meetings at the next faculty/doctor clinic meeting.
The idea reached implementation first through being materialized when the team decided to have a staff member lead in the clinic-wide skit and the team designed a process for how staff would run the meetings and wrote talking points for Faculty1. At the next meeting after the skit, the team continued to materialize the idea by developing tools, like agendas, for staff members to run the meetings more effectively. The team further implemented the idea by extensively piloting staff-led meetings, focusing on fine-tuning the clinic-wide process in which receptionists became the default leads, with the exception of two groups that already had effective full-time senior doctors. This occurred 59 weeks after the initial idea was voiced.
The co-crafting pathway typically involved a long process like the one above, in which numerous higher- and lower-power team members used the predominant cultivation practices of developing, amplifying, and legitimizing. The original voicer’s idea often became disassociated with the person who voiced it, as this example shows: Nurse1 rather than MA1 was credited with this idea in the end.
Problematizing pathway
Voiced ideas on the problematizing pathway also implied change to the work of higher-power team members and thus provoked substantial resistance. But the ideas reaching implementation on this pathway incited even more resistance because they required doctors and managers to add to their core work rather than only relinquish or delegate work as on the co-crafting pathway. Ideas on the problematizing pathway faced the most resistance of all pathways, averaging six instances of opposing or avoiding per voiced idea, with high levels of opposing. Given such major resistance, ideas on the problematizing pathway were kept alive with the key practice of issue-raising: when team members raised specific issues with a voiced idea, they created an opening for other team members to develop relevant collective solutions that enrolled the support of higher-power members and generated public pressure to accept those solutions when initial concerns had been publicly addressed. We also observed three other prevalent cultivation practices on this pathway—developing, amplifying, and legitimizing—for a total of four core cultivation practices, one more than any other pathway. Example voiced ideas on this pathway included creating greater transparency around patient waiting time by having doctors track how late they were running to publicly post in the patient waiting room, and allowing patient satisfaction ratings of doctors to impact team projects in ways that created more work for doctors. Voiced ideas on the problematizing pathway on average took 57 weeks to implement and involved nearly double the amount of cultivation practices as the next-highest pathway, requiring an average of 49 cultivation practices per voiced idea to reach implementation.
This pathway is illustrated by Nurse2’s idea to stop nurses from taking patient calls that could instead go directly to doctors. At a meeting in December 2012, Nurse2 shared that nurses were being inundated with calls from patients wanting or needing to speak directly to their doctors. She described feeling forced to call patients back on a doctor’s behalf: “How frustrating it is for patients to get phone calls they do not want to get. They don’t want to get a call from a nurse, but from a doctor.” In response, Receptionist1 immediately legitimized the idea by vouching for it: “We should put ‘doctor only’ [notes in the computer system] so that you [nurses] know not to touch them.” Nurse2 responded by issue-raising, sharing how Receptionist1’s suggestion would not work in practice because “we can forward them [the call-back requests] to doctors, but they [doctors] send them back [to nurses].” Doctor3 then opposed the idea by saying it “can’t be fixed” without explaining why.
In the weeks following Nurse2’s initial suggestion, the idea continued to be further developed and amplified, and issues continued to be raised with it. In the next meeting, it was revealed in discussion of a separate topic that of the 400 daily calls coming into the front desk, about 100 go to a nurse and 10 go to a doctor. In response to this new information, Doctor3 became more open to engaging with Nurse2’s idea that he had opposed at the previous meeting, asking if some of the calls could go to doctors, thus further developing the idea. The week after, Nurse2 left the team when Nurse1 returned from maternity leave. But Nurse2’s idea persisted when Faculty1 amplified it by saying, “If it’s not a nurse question, then it [the call] shouldn’t go to the nurse.” Nurse1 then developed the idea further, after which Faculty1 amplified the idea again. In response, Patient1 engaged in issue-raising by pointing out a specific concern with the idea: “How would doctors feel about having five calls waiting?” In response, the team developed a solution to track how many calls doctors would get in a day. At the subsequent meeting, Doctor2 developed the idea further by asking clarifying questions along with Nurse1 and Faculty1. For instance, Doctor2 asked, “In terms of things to change, who was going to handle what kind of call? Normally all the calls go to the nurse and nothing to the doctor.” Nurse1 then developed the idea further by saying, “For me it’s more beneficial to have a doctor, but I’m confused when to ask. When does the front desk go to the doctor directly?” The idea continued to be amplified by multiple team members at a meeting three weeks later, including by Receptionist1 who said, “If I were the patient, and I said I wanted to talk to the doctor, and a nurse calls me back, it would piss me off.” Another three weeks later, Faculty1 amplified and further developed the idea by asking whether doctors call patients back on the same day. Doctor2 and Receptionist1 then opposed calls going directly to doctors by advocating for nurses continuing to screen all the calls for doctors, after which Doctor2 raised another issue: “They don’t have a system in place for receptionists to know which calls are urgent and require doctors.” Receptionist1 noted, “We at the front desk need a buffer. Why don’t you talk to the nurse first?” Amplifying the original voiced idea, Nurse1 said, “Nothing has to come to the nurse.” Receptionist1 further engaged in issue-raising by sharing her concern that “If you open that floodgate, they [front desk] will be sending everything [to doctors]”; she then helped develop the idea by asking for clarification about which doctors should be receiving these patient calls. At the meeting a week later, Nurse1 legitimized the idea by reiterating that “all [calls] go to nurses; some could go to doctors,” which was supported afterward through amplification. Doctor2 then engaged in more issue-raising by noting that only nurses and not doctors have a system in place to respond to patients within 24 hours. Doctor3 addressed this specific concern by suggesting they consider a new “nurse receptionist protocol” for handling calls, legitimizing the idea as doable.
At meetings over the next ten weeks (weeks 12–22), a similar cycle occurred in which many allied team members developed, amplified, and legitimized the idea, while Doctor2 and Receptionist1 alternated between resisting the idea by saying it would not work without giving a reason and issue-raising by pointing to specific issues with it. While their resistance in the form of opposing and avoiding may seem similar to issue-raising, by pointing out specific problems with the idea, issue-raising differed in that it kept the idea alive as other team members publicly devised solutions to the raised concerns. For example, when Nurse1 legitimized the idea again in week 14, Receptionist1 engaged in issue-raising by saying receptionists lack skills to know which calls to route to doctors, to which Nurse1 and Doctor1 suggested, “We can do training [of receptionists], if we think it’s important.” In week 16, Doctor2 engaged in issue-raising by saying that nurses could “work at the top of their license” dealing with phone calls that would have been sent to doctors, prompting the team to discuss whether receptionists should collect information before sending it to doctors. Doctor2 continued issue-raising into week 17, and in response, Doctor1 amplified the idea of doctors directly taking calls in an emotional statement, saying, “We as a clinic often put the front line as a defense for blocking us. . . . To put the front as a block is not customer friendly. We should not be getting so entrenched in a standpoint. I’d rather call back [the patient] and problem-solve with them. That’s where we as a culture can move.”
After this last round of issue-raising in May 2013, the team made significant progress on getting everyone collectively onboard with implementing the idea. Doctor3, Faculty1, and Doctor1 were regularly acting as allies legitimizing the idea, with Doctor3 saying, for instance, “The decision of whether a nurse calls when the patient says no nurse, it’s a waste of time for them, for the nurse, for me, and so I wish they’d just sent it to me.” Doctor2’s opposition began to wane during this meeting; he said, “I’m in agreement. One of the stodgier doctors said the nurse should print it and bring it to the doctor. A not stodgy doctor said whoever last saw the patient [should call the patient back].” Faculty1 and Doctor1 seized on this opportunity to talk through how this might work in practice and address related logistical issues with the team. The meeting ended with Faculty1 agreeing to formally inform Clinic Leadership about the team’s decision to start routing some patient calls to doctors rather than first going through nurses; her explicitly informing leadership about this decision marked the first instance of the idea’s implementation in the form of materializing, which occurred at 22 weeks. The idea was further materialized when the team spent the next meeting developing a formal process and checklist for receptionists to route certain calls to doctors, as well as a plan for communicating this new process to receptionists. For the next few meetings, the team fine-tuned the process through piloting, and by August 2013, the idea initially voiced by Nurse2 the previous December was formalized as the official callback process used throughout the entire clinic.
Ideas on this pathway required the highest number of cultivation practices of any pathway, with distinctively high rates of issue-raising. In pointing out an idea’s specific weaknesses, issue-raising created openings for collective solutions to be generated and eventually attenuated resistance. Six other voiced ideas that faced similar levels of resistance due to adding work to higher-power roles reached implementation through this problematizing pathway.
Catalyzing pathway
In the final pathway we observed, voiced ideas reached implementation after a shock acted as a catalyst to revive and breathe new life into stalled or rejected ideas by spurring new instances of the cultivation practices of legitimizing, developing, and amplifying. Such shocks occurred when a team member quit the team (i.e., Patient1 or MA1) or joined it (i.e., Patient2, Patient3, or MA2) in a way that other team members connected to the previously stalled voiced idea, with the exception of one shock that occurred when a guest speaker shared a moving success story with the team. Ideas following a trajectory consistent with the catalyzing pathway faced resistance for a variety of reasons. Some altered higher-power members’ work, such as stopping an ineffective lab test process that forced doctors to sign off on tests they agreed were necessary but did not want to be personally responsible for. Others faced resistance due to concerns about feasibility, importance, awareness, and understanding, such as addressing patient preferences for outreach calls and notifications. The catalyzing pathway was the longest of the five pathways, averaging 74 weeks, given that ideas on this pathway were stalled before being resurrected after a catalyzing shock to reach implementation.
The catalyzing pathway is illustrated by Patient1’s idea to have doctors play a less central role and patients play a more central role on the team. During a discussion about adding another doctor to replace a doctor who had left the team, Patient1, with her voice trembling, said, “I did a lot of thinking after the last meeting, about the whole [team] and what’s not working for me. . . . You [doctors] come in and overshadow. . . . You don’t need multiple levels [of doctors—residents and attending]. [Doctors] on the team have strong opinions and personalities, and other roles don’t.” Patient1’s comment was opposed in the moment by Doctor2, who provided a long explanation in a measured, neutral voice of why Patient1 was wrong about “voices not being heard, which isn’t necessarily related to doctors.” Patient1 continued to push her idea that doctors should play a less central role on the team and specifically that not replacing the departing doctor with another doctor could help, sharing that she felt unheard: “[This] is pulling teeth every week. . . . Discussions where I have no input, I have no input. . . . I know business processes . . . and I am not being appreciated or heard, so I am not going to do it. It’s not worth it to me to come here. . . . The perspective of everyone is important.” Patient1 was opposed by lower-power and higher-power team members alike; for example, Receptionist1 suggested Patient1 try to respect and work around the status quo, and Doctor2 called for a “time check” to move on from Patient1’s concerns, even as Patient1 began to cry. Faculty1 made an executive decision to replace the doctor on the team with another doctor, and Patient1’s concerns were not revisited.
The idea seemed put to rest until a shock occurred when Patient1 quit the team in frustration the following month. Though Patient1’s voiced idea appeared to fail in the moment, her act of quitting the team served as a catalyzing event that led team members to revive, legitimize, and develop it. For example, Doctor1 shared after Patient1 left: The patient piece, I feel like it is definitely a clinic [problem]. Even when Patient1 was there . . . the balance of patient voice and everyone else’s voice was very skewed. I didn’t think of patient-oriented projects much. That was a big one for me. I still feel hierarchy among our roles—heavy with doctors and doctor voice. We tend to speak more confidently . . . it’s not a level playing field in terms of empowerment, and I would include the patient too. [The team] is a mini clinic, so I think Patient1’s experience and leaving reflects what we are like as a clinic.
As the doctors began to retrospectively revive and legitimize the content of Patient1’s voiced idea after she left, they also came to see that they did not know how to prevent the issues Patient1 had raised from happening again when recruiting new patients to the team. They began to develop the idea by looking for ways to make changes and began asking more questions. For example, Doctor1, Doctor3, and Faculty1 sought advice from PeopleHealth’s patient advocate on specific actions the team could take to not repeat the issues that Patient1 had brought up.
Patient1 leaving was referenced as legitimizing the oversized influence of doctors on the team, and the team developed concrete plans to improve the experience and better support contributions from patients on the team. The doctors and Faculty1 decided to have two patients on the team instead of one. Doctor3 said at that meeting, “We didn’t consciously feel [the team] included patient experience.” And Doctor1 shared, “I could feel in myself and see in other folks that [project work] was super heavy only for [doctors], and MA1 was checked out. The balance of projects should include different roles, multiple projects, and projects of different sizes. . . . I love the idea of two patients [on the team instead of one].” In this quote, Doctor1 also revived another of Patient1’s previously voiced and dismissed ideas to pursue a variety of projects of differing sizes and beneficiaries, including patients. The idea first reached implementation when the team spent time during the subsequent 15 meetings creating an interactive group interview to recruit and onboard two new patients as team members. By 2014, the team had formal processes to give patients’ input additional weight—selecting projects that only patients found important “to honor the patients” and to acknowledge that patients on the team might see or want different things than doctors and staff. Patient2 later benefited from the changes in how the team treated patients’ opinions, saying, “They value your opinion; they ask for it, and I just give input. That’s the reason I do it from my perspective. . . . It doesn’t mean that [patients] are more important than anything else. It’s just new. And the fact that they’re doing it, I think it’s great. I think [patients] should be included exactly the way they are. . . . I think they’re doing it right.”
This example shows how through the catalyzing pathway a voiced idea that had been rejected and dismissed was later resurrected. Team members reengaged with the idea after a related shock, which served as a catalyst for restarting the cultivation process, with the practices of legitimizing, developing, and amplifying helping the original voiced idea move from being stalled to reaching implementation. The finding that a team member change shocked this team into reevaluating ideas and processes aligns with past findings in the team literature that member replacement can disrupt team processes and knowledge structures (Zellmer-Bruhn, 2003; Summers, Humphrey, and Ferris, 2012) and can make teams more adaptive (Gorman, Amazeen, and Cooke, 2010). But we also found that the shock alone was not sufficient to generate implementation of the initial voiced idea; instead it spurred a cultivation process in which that idea was also legitimized, developed, and amplified by allies, even when the departed member was no longer on the team to advocate for the idea. It took 61 weeks, but Patient1’s voiced idea resulted in concrete changes being implemented: the number of patients on the team increased, and non-doctors were given more influence in running the team and selecting projects. By the end of the study period, 44 percent of projects were being led by a patient or staff member, as compared with 20 percent of projects at the beginning of the study period. Ten other voiced ideas were implemented through the catalyzing pathway.
Alternative Explanations
One may wonder whether the cultivation process is really responsible for ideas living on to implementation, rather than there being underlying differences in the ideas themselves that drove whether they lived on or not. First, were ideas that failed in the moment simply worse ideas, deserving of the winnowing that the other members were conducting by rejecting or ignoring them outright? Our data do not suggest this was the case. Many rejected ideas were verified as objectively viable because they were either successfully used in similar settings (e.g., other outpatient clinics in the local area) or were mandated by the PeopleHealth organization at a later date, directly implying their potential appropriateness. Consider Nurse2’s failed voiced idea to have a non-doctor clinician dedicated to serving the increasing number of walk-in patients who came in with urgent needs but did not have appointments. This idea never reached implementation, but we know it was not simply a bad idea because other clinics in their network successfully implemented it and the larger PeopleHealth organization eventually mandated that all clinics create walk-in clinician appointments. Similarly, Receptionist1 voiced the rejected idea to improve clinic efficiency by having medical assistants schedule specialist referral appointments directly with patients while they were in the examination rooms together rather than having patients track down the referral receptionist afterward. Although this idea did not make it to implementation, it was used successfully at other clinics. These and many other examples of failed ideas being successfully implemented elsewhere suggest we cannot simply write off the rejected upwardly voiced ideas as being without merit. See Online Appendix B for descriptions of a subset of failed ideas otherwise similar to successful ones, including evidence of their potential value.
Second, were failed ideas harder or more controversial, failing because of the challenges they posed and resistance they incited rather than a lack of the voice cultivation process? Indeed, our findings indicate that the extent of and underlying reason for resistance to different voiced ideas played a material role in determining the pathway describing the trajectory through which voiced ideas overcame resistance to reach implementation. But our findings do not indicate that the extent of and reason for resistance determine whether an idea stays alive to reach implementation or not. To explore this further, we matched the most promising failed ideas with otherwise similar successfully implemented ideas based on the amount of and reason for resistance, such that we had at least two failed ideas with similar amounts of and reasons for resistance for each pathway for the purpose of comparison, e.g., comparing ideas that provoked similar levels of resistance because they added to doctors’ core work. Despite these similarities in the ideas and the level of and reason for the resistance they faced, the process through which they were supported or not by allies differed in notable ways. We found that the predominant reason failed voiced ideas were not successfully revived to reach implementation was that team members did not use commensurate and appropriate combinations of cultivation practices related to the underlying reasons the ideas faced resistance, in contrast to the similar ideas that reached implementation. For instance, voiced ideas that faced resistance because they added work for doctors failed when the key cultivation practice of issue-raising was absent. When other ideas that similarly incited resistance because they added work for doctors were able to live on to reach implementation (i.e., on the problematizing pathway), issue-raising was a predominant practice and was supported by developing, amplifying, and legitimizing—the dominant practices on the problematizing pathway.
Third, were the successfully implemented voiced ideas primarily raised at a later stage in the team’s development, after the team had evolved to develop trust and work well together, so that this trust and teamwork capacity were responsible for implementation of these ideas rather than the voice cultivation process? Our data do not support this interpretation, as 39 of the 49 voiced ideas that eventually reached implementation were initially raised in the first six months of the team’s life working together. While more voiced ideas were certain to come in the future and to live beyond the period of study, the large number of ideas voiced early in the team’s life suggests that getting to implementation did not rely on evolving trust or team familiarity alone—the cultivation process was essential even before trust and familiarity developed on the team.
Discussion
Understanding whether and how upwardly voiced ideas can live on beyond their initial utterance to reach implementation is vital for organizations and teams to benefit from the potentially valuable expertise, experience, and wisdom of all employees. By conducting one of the first longitudinal studies of voice, tracking voiced ideas over two years to understand how they can stay alive to reach implementation despite initial resistance, we developed an empirically grounded theoretical model of the “voice cultivation process,” which can occur in teams to help voiced ideas achieve their initial intent of improving team or organizational functioning. In doing so, we have advanced a new perspective on voice that looks beyond the initial moment of voice and the dyadic relationship between voicers and authority holders, both of which have been the focus of past research. By examining the life of the idea over time on a team, rather than the reception and framing of a voiced idea in the moment, we discovered that even ideas that are rejected and appear to fail when first voiced can be revived and kept alive by other team members to ultimately reach implementation.
We identified five dominant cultivation practices enacted by allies and voicers that helped voiced ideas stay alive: amplifying, developing, issue-raising, exemplifying, and legitimizing. We also documented five distinct pathways, which relied on different combinations of practices and took varying lengths of time, through which voice could be cultivated to reach implementation, and we detailed the specific reasons for resistance and combinations of cultivation practices associated with each pathway. Our findings suggest the need for reconceptualizing how we understand and study voice in organizations to account for the social process that may ensue when ideas are voiced publicly in teams so that even initially rejected voiced ideas can be revived to live on through the support of allies.
Contributions to Our Understanding of Upward Voice in Organizations
This study makes multiple contributions to scholarship on upward voice in organizations. First, while the voice literature has focused on the antecedents and outcomes of upward voice (e.g., Liang, Farh, and Farh, 2012; McClean, Burris, and Detert. 2013), we find that the process through which voice proceeds is vital to whether it lives on to reach implementation. Our longitudinal data reveal that initial responses to voice are often inadequate for understanding the true trajectory of voice when it occurs in public settings on groups or teams. We find that one cannot predict whether voice will be successful by examining only its initial utterance. In our data, many ideas that seemed to fail initially in fact lived on at later points in time. Because prior work has relied on cross-sectional approaches (see reviews by Morrison, 2011; Bashshur and Oc, 2015), extant voice scholarship has not been able to differentiate voiced ideas that fail altogether from those that fail only in the initial moment and are later revived. This suggests prior scholarship may be missing crucial instances in which initially rejected voice lives on to make a difference, and it highlights the importance of examining the process through which voice reaches implementation rather than examining only the initial voicing moment and self-reports of voice frequency. We visually depict our theory of the voice cultivation process in teams in relation to prior scholarship in Figure 1.
A Theory of Voice in Teams in Relation to Prior Voice Scholarship: The Voice Cultivation Process*
The longitudinal perspective in our research and implied in our theory depicted in Figure 1 also calls for reconsidering the meaning of past findings about the antecedents, moderators, and outcomes of voice (depicted with dotted lines in Figure 1). Research on voice antecedents may need to explore whether the same factors that lead to privately voiced ideas in dyads are applicable to publicly voiced ideas in teams—for example, exploring how team climates, such as the level of psychological safety, or leader behaviors, such as the level of commitment to inclusivity, affect whether and how voiced ideas reach implementation in teams and with what consequence. Regarding moderators of voice, while past research has shown that the framing of voiced ideas may enhance their likelihood of positive reception (e.g., Chamberlain, Newton, and LePine, 2017), we found that voice’s initial framing was insufficient to explain whether a voiced idea reached implementation; voiced ideas that were initially rejected were sometimes taken up by other team members at a later time. It may be that framing is particularly important for initial reactions or during dyadic voice exchanges. When ideas are voiced in public, team settings in which members are given additional time and space to reflect, team members may exhibit greater openness to reengage with an idea later on. The initial framing may also be more salient for authority holders who feel they must assert themselves, e.g., because they feel threatened (see Burris, 2012); lower-power members may be better able to focus on the idea itself as it pertains to their own experiences and needs. By analyzing ideas instead of focusing on the person-level, we found that an individual could voice ideas that reached implementation while also voicing ideas that failed to reach implementation. In this view, the initial framing of voice in group or team settings and a person’s voicing frequency may not matter as much as whether others take on and collectively engage with voiced ideas.
Moreover, this longitudinal view suggests that rejection in the moment does not mean a voiced idea fails to reach implementation. Outcomes of voice may be more nuanced than has been depicted in prior studies on upward voice. Like earlier work, we find that voice can have negative consequences for the voicer (e.g., Milliken, Morrison, and Hewlin, 2003; Burris, Detert, and Romney, 2013). But our findings suggest that while employees whose ideas are initially rejected may suffer negative consequences and even quit the team in frustration, their ideas may still reach implementation, sometimes resulting in positive change for their role, their team, or their organization over time (depicted with plus and minus signs in Figure 1). For example, Patient1, the first patient on the team we studied, voiced several ideas that were rejected or ignored in the moment, but these ideas lived on after she quit the team. They were referenced months later, particularly by the doctors, as the team reflected on and engaged in structural changes that subsequent patients on the team found beneficial for their role (e.g., having two patients on the team and prioritizing patient-initiated projects).
As we followed ideas over time, we also observed that the original idea holder did not always receive credit for their idea, suggesting that the outcomes for voicers and the outcomes for teams or organizations may differ: one may suffer while the other benefits. This raises challenging ethical implications for advising lower-power members to take the chance to upwardly voice ideas, especially when they imply change for higher-power groups. On the one hand, if these team members care about the idea itself enough to risk negative career consequences, voicing may be worth it for the chance of the idea being taken up later by others. On the other hand, if negative career risks are felt to be too great or the idea not sufficiently compelling, voicers may benefit from focusing their voiced ideas on those that face less resistance; voicing may be safer when resistance to an idea is due to concerns about feasibility, importance, or raising awareness, rather than adding to or removing the core work of those in higher-power groups. The question of who gets credit, why, and whether team members should allow others to take credit for and implement another’s ideas may also be informed by ongoing work on gender in organizations, which has found that women who speak up promotively do not receive the same positive credit that is afforded to men who do so (McClean et al., 2018). To promote equity in the workplace, public displays of allyship and efforts to grant credit to initial voicers may be uniquely important when voicers are women (Smith and Johnson, 2017) and people of color.
An additional contribution to the voice literature is to highlight the importance of the public, social, and collective aspects of voice in organizations. While existing literature conceptualizes voice as dyadic (e.g., Detert and Burris, 2007)—often imagined as the private conversation between a subordinate voicer and superordinate supervisor accepting or rejecting a voiced idea—we study voice in a naturalistic setting to uncover the role that team members can play in supporting an idea through implementation. Accounting for voice in collective settings is especially important for building theory about voice that reflects organizational reality (e.g., Barley, 1996; Bechky, 2011) given that the majority of workers today spend most of their time at work in groups and teams (Cross, Rebele, and Grant, 2016). While prior work may be predictive of dyadic voice attempts between a subordinate and supervisor (Burris, Detert, and Romney, 2013), it is imperative to understand how others who witness a voiced idea may act as allies or opponents to affect whether it reaches implementation.
We find that ideas voiced in front of others may be rejected or ignored, but they cannot be unheard and so have the potential to resurface at a later time. Our study suggests that support for these ideas comes in many forms (amplifying, legitimizing, developing, exemplifying, and issue-raising) and may not always look supportive. The practice of issue-raising, in which specific concerns about an idea were raised, often had a critical tone but also kept the idea alive and prompted the team to engage in pointed problem solving to publicly address the issues raised, moving the idea closer to being realized through implementation. We also find that the responses of both lower- and higher-power team members matter. While leaders may greatly influence the climate for inviting or excluding contributions in a group (Edmondson and Nembhard, 2006), the capacity to bring voiced ideas to implementation does not reside with them alone. Our data reveal numerous instances of lower-power members amplifying, developing, issue-raising, and legitimizing others’ ideas, even when higher-power members resisted them. But to be clear, lower-power members are not always allies: we also document how lower-power members can resist voiced ideas. These findings indicate the importance of the mediating role of allyship of others beyond the voicer–supervisor dyad in affecting the trajectory of a voiced idea.
A final novel contribution to the voice literature is the discovery of specific pathways and a toolkit of practices through which different types of voiced ideas may reach implementation. Our study suggests it is critical to match cultivation practices to address the underlying reason a voiced idea faces resistance. This perspective separates the voiced idea from the voicer, which prior work has conflated (Chamberlain, Newton, and LePine, 2017), to examine how a voiced idea can still reach implementation and change the direction of a team or group discussion at a later date, even when the person who originally voiced the idea is no longer on the team. Our inductive analysis led us to discover that the extent to which a voiced idea threatened to change the core work of the most powerful groups in an organization appears critical for understanding the pathway that leads to implementation. This suggests that team members might be able to strategically deploy certain cultivation practices based on the source of resistance to an idea. For instance, trying to use cultivation practices common to the persistence pathway, such as exemplifying one’s own idea and driving it forward—which is similar to the guidance to use repetition and consistency when one holds a minority opinion in a group of equal-status members (e.g., Moscovici, Lage, and Naffrechoux, 1969; Nemeth and Kwan, 1985; De Dreu and Beersma, 2001)—may be a low-yield strategy if the idea implies change to the core work of higher-power groups but a viable strategy if not. By examining not only the framing of an idea but also its content, we identified specific underlying reasons for resistance to different ideas that different patterns of voice cultivation practices can help overcome for voice to be realized through implementation.
Limitations and Opportunities for Future Research on Voice in Teams
While ethnographic longitudinal studies conducted within a single context are useful for discovering processes to build new theory, future work is needed to test and further refine the theory and ideas developed in this paper in other contexts and with more teams of different types. For example, voiced ideas on virtual or globally distributed teams or teams with fluid membership may require different sets of practices or pathways to reach implementation (e.g., Neeley, Hinds, and Cramton, 2012; Kerrissey, Mayo, and Edmondson, 2020; Lee, Mazmanian, and Perlow, 2020).
Since prior scholarship has primarily theorized and studied voice as a dyadic exchange between a voicer and authority holder, it remains an open empirical question—and a rich vein for future research—to test whether the antecedents, moderators, mediators, and outcomes associated with dyadic voice exchanges also operate similarly for voice in team settings. As we previously argued, voice in teams may be inherently different from voice that occurs privately because voice in a public, team context may be recalled later by team members who initially heard the voiced idea (e.g., Soderstrom and Weber, 2020). Exploring the moderating effects of individual and team characteristics (e.g., personality composition, team size, and team structure), prior relationships (e.g., team member familiarity or prior experience working together), shared experiences (e.g., team launch), temporality (e.g., short duration teams), and dynamism (e.g., changes in goals, tasks, or membership) (Kerrissey, Satterstrom, and Edmondson, 2020) may uncover how these team attributes affect the trajectory of voiced ideas and could be fruitful avenues for future research. For example, future studies could examine how the distribution of power on the team affects both the process (e.g., the type of cultivation practices the team uses) and outcomes. The extent to which teams are characterized by voice coming from a few dominant team members rather than being equally distributed among members (Sherf et al., 2018) may affect the types of practices and pathways that voice takes to reach implementation and may require studying many teams in a controlled setting over time.
While we used longitudinal ethnographic observation to identify mediators, other researchers could use a variety of innovative approaches to study and test the theory generated here on other teams. For example, future research could use thin slices of team interactions to determine whether observers’ (e.g., managers’) perceptions of the mediator pathways we identified (e.g., behaviors associated with allyship, persistence, co-crafting, or problematizing) predict successful implementation of voiced ideas (Satterstrom et al., 2019). In addition, the cultivation process may unfold differently in teams with varying experiences and motivations; for example, past research has indicated that early successes can generate positive emotions that lead to cooperation (Heerdink et al., 2013) and may lead to more allyship practices on the team. Finally, while we did not observe a different effect of promotive versus prohibitive framing on the voice cultivation process, future research could investigate whether framing approaches interact with cultivation practices to lead to more or less implementation. Although it was outside the scope of our paper, studying silence and failed voice in teams also may have important implications for individual, team, and organizational outcomes.
Regarding outcomes in the study of voice, this study suggests the urgent need for future studies of voice to measure whether voiced ideas get implemented or not. If the purpose of voicing is to improve organizational or team functioning (e.g., Hirschman, 1970; LePine and Van Dyne, 1988), voice can have its intended positive impact only if voiced ideas are acted upon in practice through implementation. In reviewing the scholarship on voice, we were surprised by the lack of studies that traced specific voiced ideas to implementation. The majority of studies instead asked voicers to recall moments of voice (e.g., its framing, who they voiced to) and then used supervisors’ assessments of the voicer (e.g., how useful their ideas generally are, how often they speak up) or the unit (e.g., how well the department performs) or used voicer assessment (e.g., did you feel heard, do you want to quit) as outcome measures. Future work that also separates the voicer from the voiced idea and traces whether or not the idea resulted in positive change through implementation is needed to ensure that the antecedents of voice identified in existing literature are truly predictive of voice that is not only well-received in the moment but is acted upon to make change.
There are also boundary conditions related to the specifics of our case and data that have implications for what other contexts the voice cultivation process may generalize to most easily. We developed this new theory by using observational data from weekly team meetings, but conversations between team members outside meetings may have shaped the observed dynamics during meetings, in line with scholarship on coalition formation processes (e.g., Murnighan, 1978; Van Dyke and McCammon, 2010). While our interviews and job shadowing observations with team members after and outside of team meetings suggest this was not a major issue, especially as these team members did not regularly work together in the clinic, future research that has access to observational data both during and outside of group or team meetings is needed to examine how such private conversations may shape the public voice cultivation process. Allyship behavior may be more prevalent when it is planned in advance of the meeting. For example, a recent popular press article (Eilperin, 2016) suggested women serving in the Obama administration purposefully served as allies to one another to amplify one another’s ignored ideas during meetings. While this example can be read as an instantiation of allyship through cultivating another team member’s voiced ideas in another context, it also suggests that such allyship can be promoted through purposeful strategizing rather than the more organic and less premeditated allyship observed in our case.
Other boundary conditions relate to the characteristics of the team used to develop our new theory. Because our aim was to understand how voice might stay alive in teams to reach implementation, we used purposive sampling to analyze data from a team that offered enough comparable successful cases of upward voice that reached implementation, in line with best practices for qualitative theory-building papers (e.g., Patton, 1990, 2007; Creswell, 1998; Langley, 1999; Locke, 2001; Yin, 2009). This was especially important given that most studies suggest upward voice is risky, rare, and hence hard to observe sufficiently for building theory (e.g., Morrison and Phelps, 1999; Morrison and Milliken, 2000; Premeaux and Bedeian, 2003; Grant and Ashford, 2008). The team we studied was numerically balanced with equal proportions of higher- and lower-power members, was established with the explicit goal of being inclusive to develop interdisciplinary solutions to interdisciplinary problems, and completed team training to encourage lower-power members to feel comfortable speaking up. Yet despite the team’s good intentions, lower-power members’ ideas were disproportionally ignored and rejected relative to higher-power members’ ideas, making it perhaps more representative of typical teams than we initially expected.
While the benefits of leveraging ideas and perspectives voiced by employees across all organizational levels are well-documented (e.g., Bunderson, 2003; Hinds and Pfeffer, 2003; Dahlin, Weingart, and Hinds, 2005), barriers abound that prevent upwardly voiced ideas from being implemented and positively impacting team and organizational functioning. Our study uncovers a novel set of practices that team members can use to help voiced ideas live on over time, demonstrating that it is not only the leader or initial voicer who is solely responsible for helping an idea to endure and be realized. Voiced ideas can reach implementation to create meaningful change in organizations and teams through other team members’ efforts in the voice cultivation process, even in the presence of initial resistance.
Our research sets a new theoretical agenda for the study of voice that goes beyond the voicer–leader dyad to account for the critical role of others to collectively influence whether and how voice reaches implementation. By employing novel methods in the study of voice—using longitudinal and inductive, qualitative methods—our study opens new avenues for future research to uncover additional pathways and practices through which voice can reach implementation to improve team and organizational functioning. And our data suggest that in practice, by daring to speak up, voicers make their ideas common knowledge and give their ideas an opportunity to live on and generate change, even if those ideas are initially rejected or the voicers leave the team before they gain traction. Team members also do not need to be heroic voicers to drive improvements in their organization or team: they can accomplish this by acting as allies to help cultivate others’ valuable ideas so they stay alive and reach implementation. Because valuable ideas are always needed, and speaking truth that is heard can be incredibly hard, the voice cultivation process offers a route toward change.
Supplemental Material
Satterstrom_Kerrissey_DiBenigno_online_supp – Supplemental material for The Voice Cultivation Process: How Team Members Can Help Upward Voice Live on to Implementation
Supplemental material, Satterstrom_Kerrissey_DiBenigno_online_supp for The Voice Cultivation Process: How Team Members Can Help Upward Voice Live on to Implementation by Patricia Satterstrom, Michaela Kerrissey and Julia DiBenigno in Administrative Science Quarterly
Footnotes
Acknowledgements
This paper greatly benefited from the insightful feedback and friendly reviews of Jeff Polzer, Sara Singer, Andrew Knight, Leslie Perlow, Somava Stout, Heidi Gardner, Connie Gersick, Katherine Kellogg, Gerardo Okhuysen, Frances Milliken, Curtis Chan, Elad Sherf, Beth Bechky, Julie Battilana, Luciana Silvestri, Matthew Lee, Ting Zhang, Basima Tewfik, Catarina Fernandes, Jean-Francois Harvey, members of the NYU Qualitative Workshop, GroupsGroup, Craft, HBS OB Lab, HBS NERD Lab, and HBS RCS. We are also grateful for the feedback from audience members and seminar participants at the Richard Hackman INGRoup Dissertation Award lecture, NYU Wagner, UT Austin, and AOM. We are grateful to John Wagner and three incredibly insightful reviewers for developing the potential of this work, as well as Joan Friedman for invaluable copyediting. Finally, this study would not have been possible without the support of the healthcare team members studied. This research was supported in part through the Susan G. Cohen Doctoral Research Award of the Academy of Management Organizational and Change Development Division and the Eric M. Mindich Research Fund for the Foundations of Human Behavior, Harvard University.
Supplemental Material
Authors’ Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
