Research shows that teams in which members openly share and discuss different viewpoints often exhibit better performance (e.g., Li et al., 2022). Debate is a team-level construct that entails the open, controversial discussion of different perspectives that group members exhibit with regard to tasks and decisions at work (Simons et al., 1999). Debate is a constructive way of handling dissent (Robijn et al., 2020), allows individuals to share their opinion and to participate in group-decision making (Brykman & King, 2021). The effects of debate on important team-level outcomes, such as creativity and performance, have been well established (e.g., Seyr & Vollmer, 2014). However, debate has not yet been linked to individual group member well-being, defined as the extent to which group members experience happiness in their life in general (Baumeister et al., 2013). This lack of research is surprising, given that how groups manage differences in viewpoints and opinions may have significant implications for individuals’ affect and satisfaction. Research on related constructs (e.g., group voice climate, open conflict norms) suggests that debate should be predictive of team member well-being because it presents an important social-contextual factor (e.g., Brykman & King, 2021; Morrison et al., 2011; Robijn et al., 2020). Yet, the role of debate and the potential moderating conditions resulting from group dynamics have been neglected. This currently limits our understanding of how individual well-being may be affected by debate in work groups.

Work is increasingly carried out in groups or teams that are self-managed, which entails that teams have less pronounced hierarchies and high autonomy over decisions (Markova & Perry, 2014). While much scholarship has focused on team-level outcomes, there is also evidence that indicates that the structure of, and processes within, self-managed work groups may have specific implications for individual outcomes, especially for group member well-being (Markova & Perry, 2014). Well-being is an essential outcome affecting not only work-related attitudes and behavior but also health-related outcomes (e.g., Diener, 2009). It can be conceptualized in terms of occupational well-being (e.g., job satisfaction) or in more global life outcomes such as affect and happiness (Baumeister et al., 2013). The latter outcome is the focus of our study as it has implications beyond work-related aspects (Baumeister et al., 2013; Diener, 2009). On the basis of the established beneficial outcomes of debate (e.g., team innovation, Seyr & Vollmer, 2014; decision comprehensiveness and performance, Simons et al., 1999; interpersonal processes, Vollmer & Seyr, 2013), we expect group-level debate to be generally beneficial for individual-level well-being.

However, discretionary behavior such as debate can be strenuous (e.g., Röllmann et al., 2021). We expect that under certain conditions, high levels of debate can entail stressful experiences (e.g., conflict), which negatively affect group member well-being. Specifically, drawing on status conflict theories and the diversity literature (e.g., Bendersky & Hays, 2012; Harrison & Klein, 2007), we argue that the group-level construct divergence of status perceptions may moderate the relationship between debate and group member well-being. Social status is an essential feature in interpersonal contexts referring to the amount of respect, prestige, and admiration that people have in the eyes of others (e.g., Ridgeway & Walker, 1995). In groups, there is often disagreement about the dispersion of social status (Kilduff et al., 2016). As such, social status can be conceptualized as a deep-level diversity dimension in groups; members may differ in their perceptions of how status is distributed in the group (Harrison et al., 2002). We define divergence of status perceptions as the extent to which group members agree in their perception of social status being evenly or unevenly distributed throughout the group.

Disagreement regarding absolute levels and rankings of social status can impair group outcomes and well-being by fueling competitive behaviors and conflict within the group (e.g., Bendersky & Hays, 2012). At the same time, divergent status perceptions, that is, divergent views on the extent to which status differences exist in the group, imply that perspectives on the group differ and status rankings are in flux. In this case, debate might be helpful and necessary in solving divergent views on status within the group. We theorize that divergence of status perceptions might act as a moderator between debate and well-being: When status perceptions strongly diverge in the group, the positive relationship between group-level debate and group member well-being will be stronger. It will be weaker when status perceptions diverge little as this implies that there is a consensus on the level of existing status hierarchies. In such instances, extensive debate may be less positive for individual well-being as it would not be as important to discuss structural dissent and differing expectations that stem from divergent views on the extent of status hierarchies (see Fig. 1). We find support for these predictions based on multi-level modeling of data from 163 individuals in 29 activist groups.

Fig. 1
figure 1

Hypothesized Model

Full size image

The main contributions of our study are threefold. First, we advance the literature on debate, a group level construct, by analyzing its relationship with group member well-being, an individual-level outcome. Debate can be modified and fostered via group interventions or psycho-social education if it shows to be beneficial for well-being. We further respond to calls by researchers to analyze the effects of debate in a field setting (Vollmer & Seyr, 2012). The relevance of a field setting lies in the heightened external validity of study results.

Second, our study contributes to research on social status conflict and dynamics (e.g., Bendersky & Pai, 2018). We focus on divergence of status perceptions at the group-level instead of focusing on an unequal distribution of social status per se or on disagreement about the ranking of individuals (status disagreement; e.g., Kilduff et al., 2016). Adopting a multilevel perspective, we show that within-group divergence regarding the perceived extent of hierarchization in social status interacts with debate in predicting well-being. In contrast to most research on social status that examines ad-hoc student groups in the laboratory (e.g., Anderson et al., 2006), we study groups in a field setting, ensuring the external validity of our findings.

Third, by investigating our research question in an activist group context, we advance theorizing on activist group processes and provide practical implications for the design of group interactions for activist groups, self-managed teams, or teams with flat hierarchies. Our findings may also explain previous contradictory findings on activist well-being with some studies elaborating on activist burnout (e.g., Gorski et al., 2019; Tavarez, 2022) and others on activism as a supportive factor for activist health (e.g. Binder & Blankenberg, 2016; Watson-Singleton et al., 2021). The relatively loose structure of activist groups requires that group members engage in exchange in order to negotiate their shared goals, roles, and actions (e.g., Luthmann, 2019). In contrast to most research on activism, which is mainly qualitative or does not consider the multilevel structure of individual activists nested in groups, we employ a quantitative multilevel approach and illuminate group-level status dynamics as an important boundary condition.

Theoretical background and hypothesis development

Debate refers to the open discussion of differences that group members exhibit respective to tasks and decisions (Simons et al., 1999). Group members challenge each other’s viewpoints and discuss task-related differences, in other words, they manage existing dissent (Jetten & Hornsey, 2014). Debate is a form of direct participation that involves the disclosure of dissent and can lead to problem-solving (van Knippenberg et al., 2004). Constructs related to, but distinct from, debate are constructive controversy (Tjosvold & Dreu, 1997), open conflict norms (Robijn et al., 2020), intragroup conflict (Jehn, 1995), and proactive behavior in general, but more specifically voice and group voice climate (e.g., Morrison et al., 2011).

Debate is congruent with the first two of four strategies that have been proposed as management strategies for constructive controversy (e.g., Tjosvold et al., 2014): (1) Developing and articulating one’s own position and (2) questioning and analyzing the views of others. In contrast to constructive controversy (e.g., Tjosvold & Dreu, 1997), team members’ contributions in debate are not necessarily constructive (Seyr & Vollmer, 2014). Open conflict norms describe the perception that conflicts can be discussed openly in the team and that it is desirable to share the own view, even if it is controversial as Robijn et al. (2020) show in their study. Open conflict norms encourage team members to express disagreement and foster a positive perspective on dissent (e.g., Jetten & Hornsey, 2014; Tjosvold et al., 2014). In contrast to open conflict norms, debate entails behavior instead of norms and describes a way to manage dissent (but not necessarily conflict), in an open and constructive way. Debate is also conceptually related to direct, informal voice at the group-level as it can be an opportunity for individuals to express their own position (e.g., Wilkinson et al., 2018). Even though debate is not always anticipatory and therefore not always a form of proactive behavior (Tornau & Frese, 2013), participation in debate is self-initiated and impacts individuals or their environments (e.g., Grant & Ashford, 2008). Debate necessitates the existence of differences. However, debate does not measure the quantity or quality of interindividual dissent or intragroup conflict as established by Jehn (1995), but the handling of dissent during group discussions and decision-making. It is different from task or relationship conflict as elaborating on divergent viewpoints does not necessarily equal conflict.

Debate describes an environment in which diverse opinions and stances are openly expressed and discussed. As Simons et al. (1999, p. 665) state: “… diversity must be effectively expressed and integrated into decisions if a group is to derive benefit from it. If diversity is not integrated in this manner, then the coordination costs or other social costs that accompany diversity will drive performance down.” Considering the link between well-being and performance (e.g., Kundi et al., 2021) and the aforementioned results on supportive environments, we examine Simons et al.’s (1999) claim with regard to debate and well-being. Conflict management is a crucial factor to determine which consequences intra-group conflict may have (e.g., Robijn et al., 2020). Via constructive conflict management, existing conflicts can even prove beneficial (e.g., Elgoibar et al., 2017). In consequence, individuals may experience positive feelings.

Debate and group member well-being

We propose that, generally, an environment in which opinions and positions can be shared openly, is beneficial for group member well-being (Brykman & King, 2021; Morrison et al., 2011). Participation in decision-making processes can increase occupational well-being outcomes, such as job satisfaction, via a cognitive and an affective pathway (e.g., Park, 2012). This should also apply to group member well-being and debate as an opportunity for participation and as a way to manage dissent. The cognitive pathway entails a transfer of knowledge, the possibility for group members to develop a deeper understanding of the problem, to acquire more knowledge, and to contribute their own viewpoints and concerns (Locke & Schweiger, 1979). Being able to contribute one’s perspective strengthens self-efficacy, increases participation, and allows for voice. This, in turn, has been shown to promote individual well-being (e.g., Bandura et al., 1999; Röllmann et al., 2021; Weiss & Zacher, 2022). The cognitive pathway further allows for the revelation and the resolution of disagreement or misunderstandings.

In this regard, the literature on biased group decision-making suggests that without proper debate, a false consensus can develop and disguise existing problems, ultimately leading to erroneous group decisions (e.g., Akhmad et al., 2021; Henriques, 2020; Janis, 1983; Reese, 2020). A climate in which dissent can be actively addressed instead of being neglected is beneficial for both work groups and their members (Senge et al., 1994). By engaging in controversial discussion, people are confronted with differing viewpoints, knowledge, and theories (Johnson & Johnson, 2009). This often results in curiosity, search for new information, learning, increased creativity, innovativeness (Vollmer & Seyr, 2012), and working toward shared goals (Markova & Perry, 2014). It is likely that differing expectations and perceptions of the group can be discovered and addressed.

Accordingly, as Tjosvold et al. (2014) point out in their work on constructive conflict management, open conflict norms contribute to learning and problem-solving. Through debate, employees can acquire new knowledge or skills that help to fix problems and, accordingly, work procedures can be improved (Zacher, 2021). This is positive for employee-well-being (e.g., Watson & Tregaskis, 2022). If knowledge remains unshared, group members are likely to experience cognitive disagreement, affective conflict, and disengagement in group communication and task-related behavior (e.g., Kirchmeyer & Cohen, 1992; Stasser & Titus, 1985). We propose that debate is particularly relevant in self-managed work groups: Self-managed work groups do not have a formalized hierarchy so that debate should be especially important to discuss expectations on the group, allocate roles and responsibilities as well as to organize and coordinate tasks (Markova & Perry, 2014).

The affective pathway between debate and group member well-being involves the facilitation of basic needs, namely competence, autonomy, and relatedness (Deci & Ryan, 2000; Milyavskaya & Koestner, 2011). The impact from group context variables on group member well-being is mainly mediated by basic needs fulfilment, also in the case of open conflict norms (Robijn et al., 2020). Robijn et al.’s (2020) reasoning is applicable to debate: The possibility to share and discuss even differing views fosters feelings of autonomy and, in line with self-determination theory (Yu et al., 2018), enhances basic needs fulfilment (Deci & Ryan, 2000; Robijn et al., 2020).

Further, similar to voice, debate is, in a positive sense, challenging (Fay & Sonnentag, 2012) and, thus, can promote experiences of mastery and competence (Deci & Ryan, 2000). Individuals can share their position more easily in a setting of high debate, thus directly influencing their working conditions and reducing stressors which can increase well-being due to the inherent feelings of competence and autonomy (e.g., Ganster & Fusilier, 1989; Uribetxebarria et al., 2021). Open conflict norms and debate contribute to mutual understanding (Elgoibar et al., 2017; Tjosvold et al., 2014), which is crucial for the fulfillment of relatedness. A meta-analysis on democratic enterprises further finds that direct participation is related to employee satisfaction (Weber et al., 2020), presumably by augmenting feelings of control, competence, and relatedness. We thus hypothesize:

  • Hypothesis 1: Debate is positively related to group member well-being.

Divergence of status perceptions as a moderator

Even though we expect a positive main effect from debate on group member well-being, debate may involve costs, such as investing considerable resources like time and effort as Bolino et al. (2010) indicate in their conceptual paper that critically assesses appeals for proactive behavior at work. A high level of collective decision-making can increase work load and intensify intragroup dissent (e.g., Carter, 2006; Jetten & Hornsey, 2014). Participation in debate may also increase the risk of depreciation by significant others, or publicly exhibiting flaws in one’s reasoning (Gebert et al., 2006), and making oneself vulnerable (Bolino et al., 2010). Further, debate can feel futile when there is no perceived need for ongoing discussions and arguments. On the basis of theories on social status and diversity literature (e.g., Bendersky & Hays, 2012; Harrison & Klein, 2007), we propose that divergence of status perceptions moderates the relationship between debate and well-being.

In group settings, status hierarchies usually develop quickly and informally, and status beliefs influence behavioral patterns and group interactions (Ridgeway & Erickson, 2000). The amount of status that people have in the eyes of others is a perception and mainly subjective (e.g., Anderson et al., 2012). It has been assumed that group members develop a high level of consensus concerning the distribution of status hierarchies within groups and the concrete ranking of individuals (e.g., Anderson et al., 2006). However, according to more recent empirical research, this is often not true (e.g., Kilduff et al., 2016).

The construct divergence of status perceptions relates to ranking-based conceptions of status disagreements (Kilduff et al., 2016; Ridgeway & Walker, 1995) in assessing differences in status perceptions. Our conceptualization builds on and, at the same time, expands this conceptualization. It assesses not the level of hierarchization or concrete rankings, but whether group members have the same perception of the extent of how social status is distributed within a group. Divergence of status perceptions does not involve a judgment about how social status is distributed hierarchically. Rather, it assesses the amount of within-group divergence in that judgment. As such, it reflects disagreement with respect to an important attribute in the group and according to Harrison and Klein (2007) indicates separation (i.e., differences in position or opinion among members). A high divergence of status perceptions could mean that two members think that social status is very evenly distributed throughout the group, whilst two others think that social status is very unevenly distributed throughout the group. Three others might judge that social status is neither very evenly nor very unevenly distributed. In case of a low divergence of status perceptions, the group members would have a high level of agreement regarding the question how evenly social status is distributed, regardless whether they all perceive a high degree of hierarchization or whether they all perceive a low degree of hierarchization of social status. It is likely that a high divergence of status perceptions goes along with diverging expectations regarding intra-group equality, transparency, and decision-making. Regardless of the level of hierarchization of social status, the divergence of the perception of this hierarchization is relevant for the prediction of well-being, because it is a dissent that likely interacts with debate, which is a way of handling dissent.

We focus on social status as an essential element in human societies and specifically within work groups (Magee & Galinsky, 2008). Longitudinal and experimental research has demonstrated the significance of social status, social status conflicts, and complementarity of status behavior for group member well-being and group processes, also due to its evolutionary basis (e.g., Anderson et al., 2012; Bendersky & Hays, 2012; Bendersky & Pai, 2018; Kilduff et al., 2016; Lee et al., 2018). Even though status can be a source of conflict and negative feelings within groups (e.g., Magee & Galinsky, 2008), status characteristics theory (e.g., Berger et al., 1977), as well as theories on distributive justice (e.g., Deutsch, 1985) suggest that it is not the existence of status differences per se, but disagreement regarding the social status hierarchy within a group that often leads to negative consequences. For example, status conflict can lead to deteriorated group performance because group members engage in competitive behaviors which consume task-relevant group processes (Bendersky & Hays, 2012).

In summary, on the basis of theories on social status and diversity literature (e.g., Bendersky & Hays, 2012; Harrison & Klein, 2007), we expect that divergence of status perceptions moderates the relationship between debate and well-being. We assume that specific levels of debate and divergence of status perceptions are more or less compatible, so that the combination of these factors should exert a more or less positive effect on group member well-being. Debate involves an open approach to dissent that enhances information sharing (Seyr & Vollmer, 2014), and informs members about the current status situation in the group (Weiss & Morrison, 2019). We expect debate to be more helpful when status perceptions strongly diverge and the differing individual status construals have to be addressed and resolved. In this case, intense debate would be necessary to settle divergent expectations about the group or to have a better understanding of the positions of other group members (Bendersky & Pai, 2018). In case of low divergence of status perceptions, debate would be less necessary to discuss structural dissent and differing expectations. Therefore, the relationship between debate and well-being should be weaker as potential negative effects of debate could be more pronounced.

  • Hypothesis 2: Within-group divergence of status perceptions moderates the positive relationship between debate and group member well-being, such that the relationship is stronger when status perceptions strongly diverge and weaker when status perceptions diverge little.

Method

Participants and procedure

We test our hypotheses by investigating activist groups. Activism can be defined as a form of work (e.g., Banks, 2020) and activist groups are comparable to teams in the work context but without receiving payment or having a formal contract. Activist groups, defined as groups of individuals engaging in continuous or recurring behavior of advocating for a political cause (Klar & Kasser, 2009), are one example of self-managed work groups. In addition to the original definition of Klar and Kasser (2009), the continuity or repetition of activist behavior is important for the labeling of behavior as activism in this study, as we expect long lasting behavior to result in more ambivalent consequences than occasional activism (Chen & Gorski, 2015). This further underlines its applicability to paid work as paid work is also a regular activity. Unpaid political work can also fulfill the so-called “latent functions” of work (time structure, social contact, collective purpose, a social status, and activity; Jahoda, 1982; Ulich & Wiese, 2011).

As activist work is not predefined, activists need more explicit exchange than other work groups to coordinate their teamwork (Luthmann, 2019). Group exchange is therefore a very frequent activity in activist groups which leaves space for a variety of expressions of debate. Further, activist groups do not have formalized hierarchies so that there is a lot of room for divergent perceptions of social status within the group (Soteri-Proctor et al., 2016). Also, the factors that influence how activism impacts activist well-being are still unknown with contradictory evidence on activism as a stressor (e.g., Gorski et al., 2019; Tavarez, 2022) or as a resource (e.g. Binder & Blankenberg, 2016; Watson-Singleton et al., 2021). Taken together, activist groups are a well-suited sample to analyze the predicted relationships between debate, divergence of status perceptions, and well-being.

Data were collected from 163 members of 29 activist groups (M = 5.59, SD = 2.68). The recruitment took place between August 2018 and September 2019. The sample description is displayed in Table 1. The inclusion criteria entailed that every participating activist group pursued social and/or ecological goals and worked toward profound changes regarding the topic. The topics the groups worked on are depicted in Table 2. Further, it was required that the groups worked with a basic democratic claim and were self-organized (Soteri-Proctor et al., 2016). We only included groups in which group members were generally not paid. However, three individuals received a small reward for their engagement (1.84% of the sample, a maximum of 400€ per month).

Table 1 Sociodemographic characteristics of the participants
Full size table
Table 2 Primary goals of the activist groups
Full size table

We contacted 102 activist groups based in seven different German cities via email. Many groups were reluctant to participate due to time constraints and work overload, other groups were concerned with privacy issues, skeptic to be the subject of research and its results. The first author visited all of the groups that expressed some interest in the study and a total of 29 groups agreed to participate. All members of each participating group were kindly asked to participate, but there was no obligation to do so due to the voluntary nature of data collection. Some group members refrained due to personal time constraints, privacy concerns, or carelessness. It was however a requirement that two thirds of the group members took part so that a certain representativeness could be ensured. The first author or a student research assistant conducted a semi-structured interview with two to three members of each group to collect context information on the group. The questionnaire data were assessed via individually filled-in pen and paper questionnaires, preceded by information on the voluntary nature and privacy of the study participation.

Measures

Unless otherwise indicated, responses were provided on 5-point scales ranging from 1 (strongly disagree) to 5 (strongly agree). Following recommendations by Konradt et al. (2015), the team level constructs were operationalized by using items that reference the team-level (i.e., referent-shift consensus model, Woehr et al., 2015). The full measurement instrument is included in the Appendix (Table 6).

Debate

We measured debate with a translation of the adapted four-item scale by Simons et al. (1999). As debate cannot be assessed as an objective measure at the team level, we computed individual values that we aggregated to the group-level using the mean, following Simons et al. (1999). Two example items are “Group members openly challenge each other’s opinions” and “In the discussion of issues, group members state clear disagreement with each other.” We aggregated the group member judgment via an elemental composition model (Chan, 1998). In our study, Cronbach’s alpha was .79, Omega total .83. The average variance extracted was .50. With regard to aggregation indices, ICC1 was .11, ICC2 was .41 (calculated following Bliese, 2000), rwg = .62 (see LeBreton & Senter, 2008). Although the extent of within-group agreement (ICC1) was relatively high, reliability of the group means (ICC2) and rwg were below conventional cutoffs (Bliese, 2000). However, the estimates correspond to a moderate agreement following LeBreton and Senter (2008). LeBreton and Senter (2008) argue that strong ICCs are above all necessary when decisions with high importance have to be taken such as firing or hiring of individuals. For many other research questions, such as ours, low to moderate agreements suffice (LeBreton & Senter, 2008). Further, the items are referencing the team level so that team members directly address the team level even if their perceptions do not fully agree. To exclude groups with a low rwg would lower the number of groups that can be included in the analysis, thereby weakening the statistical strength of the analyses (Lindell & Brandt, 2000). Further, it would entail range restriction, whereby the size of effects should be systematically underestimated (Lindell & Brandt, 2000). Thus, we regard aggregation to the group level as justified.

Divergence of status perceptions

We measured the extent to which status perceptions diverged with the five-item scale by Anderson et al. (2012). We adapted the scale so that it measured perception of social status distribution within the group. In our study, Cronbach’s alpha was .90, Omega total .93. The average variance extracted was .64. Two example items are: “Some group members have a higher level of respect in others’ eyes than others” and “Some group members have higher social standing than others.” As our research question focuses on within-group divergence of status perceptions, we calculated the standard deviation for each group (Harrison & Klein, 2007). Similar approaches exist in the literature (e.g., Lindell & Brandt, 2000). As the level of status hierarchies in the group and divergence of status perceptions intercorrelate, we control for the influence of the level of status hierarchies in the group but do not include a separate moderating term as multicollinearity is bound to arise (Lindell & Brandt, 2000). The level of status hierarchies in the group is the aggregated mean of perception of social status distribution.

Well-being

We assessed group member hedonic well-being with a three-item scale by Baumeister et al. (2013). In our study, Cronbach’s alpha was .90, Omega total .90. The average variance extracted was .75. Example items are “In general I consider myself happy” and “Taking all things together, I feel I am happy.”

Covariates

Besides the level of status hierarchies in the group, we controlled for other relevant variables. The gender diversity of the group can influence individual well-being (e.g., Haile, 2012). Further, it can impact the perception of the workplace and its opportunities (Mousa, 2021), including debate. We calculated gender diversity with the R-package {diversity} (Solanas et al., 2012) that computes Blau’s Index (1 - ∑\limits_{i = 1}^k {p_i^2}; Blau, 1977). When there is no variety (all individuals pertain to the same category), the index equals zero. Options given were male, female, and other. As debate presupposes some degree of dissent, we controlled for intragroup conflict to disentangle the influence of conflict and debate on well-being, using the scale of intragroup conflict (Jehn, 1995). Intragroup conflict is differentiated into task and relationship conflict, and both correlate with well-being (e.g., De Dreu et al., 2004; Jehn, 1995). One relationship conflict item was omitted as it could not be plausibly adapted to the activist context (‘How much are personality conflicts evident in your work unit?’). Psychological safety depicts the shared belief by team members that it is safe to take interpersonal risks in the team without being afraid of negative consequences (Edmondson, 1999b). Psychological safety is an antecedent for processes in teams such as communication and information sharing as well as for individual outcomes like empowerment and work engagement (Newman et al., 2017). We used the German version of the 7-item scale of psychological safety (Edmondson, 1999a; Goller & Laufer, 2018).

Analyses

We conducted all analyses with the statistical software environment R (R Core Team, 2020). We ran multilevel confirmatory factor analyses (CFA) with the scale items of debate, divergence of status perceptions, and well-being to examine our measurement model using the package {lavaan} (Rosseel, 2012). In the CFA, we used robust estimators to account for missing data and the small level-2 sample size (Yuan-Bentler correction). We assumed that a reasonable model fit was indicated by comparative fit index (CFI) and Tucker-Lewis index (TLI) values greater than .95 and root mean square error of approximation (RMSEA) values smaller than or close to .06 (Hu & Bentler, 1999). We did not use SRMR values as the approximation of the multilevel SRMR between-level is heavily dependent on level-2 sample size and is not reliable for small sample sizes (Ene, 2020).

We tested our hypotheses regarding both the main effect and the moderator effect using multilevel modeling (see R syntax in the online supplementary material), using the package {lmerTest} (Kuznetsova et al., 2017). We used REML estimation as this accounts for small level-2 sample sizes (Hox & McNeish, 2020; McNeish & Stapleton, 2016). To analyze the moderator effect more in detail, we examined the respective simple slopes and tested the effect of different expressions of divergence of status perceptions (high (+0.50 SD above the mean), mean, and low (−0.50 SD below the mean)) using the package {reghelper} (Hughes & Beiner, 2022). Snijders and Bosker (2011) recommend using multilevel modeling when the number of level-2 units is 20. Other researchers advocate for the use of multilevel modeling when data is hierarchically structured as disregarding hierarchically structured data sets can lead to biases, regardless of the number of level-2 units (e.g., González-Romá & Hernández, 2017). It has been shown that multilevel modeling can perform well with small level-2 sample sizes when robust estimators and simple models are employed (Hox & McNeish, 2020; McNeish & Stapleton, 2016).

Results

Prior to testing our hypotheses, we investigated bivariate correlations among variables and the fit of our measurement model. Pearson correlation results showed that debate and group member well-being were positively correlated, providing initial support for Hypothesis 1 (see Table 3). However, it should be noted that correlations do not account for the nested structure of our data.

Table 3 Descriptive statistics and correlations
Full size table

The model fit was adequate (CFI = .98, TLI = .96, RMSEA = .05, see Table 4). We compared our measurement model with the other possible model in which both level-2 factors (debate and divergence of status perceptions) load on the same factor. The fit statistics of the alternative model indicated worse fit than our measurement model (CFI = .91, TLI = .87, RMSEA = .09, see Table 4).

Table 4 CFA results
Full size table

Hypothesis tests

The multilevel regression results are displayed in Table 5. Supporting Hypothesis 1, results showed that debate was positively related to group member well-being (γ = .22, p = .047, see Table 5). The relative fit indices (Akaike Information Criterion AIC, Bayesian Information Criterion BIC; Hamaker et al., 2011) suggested that the random-intercept model displayed the best explanation of variance (random-intercept model with debate as predictor: AIC = 389.20, BIC = 416.47, p = .002 compared to the intercept-only model). The marginal R2 that describes the variance explained by the fixed factors alone was .14 (Nakagawa & Schielzeth, 2013). The conditional R2, describing the variance explained by both fixed and random factors was .23 (Barton, 2022). The parameters of the hypothesized model were superior to the null model (the parameters of the intercept-only model were σ2Group = .12 (.344), σ2Residual = .68 (.824), AIC = 398.31, BIC = 407.40).

Table 5 Results of null-model, unmoderated, and moderated multilevel regression models predicting group member well-being
Full size table

Hypothesis 2 states that divergence of status perceptions moderates the relationship between debate and group member well-being, such that the positive relationship is stronger when divergence of status perceptions is high and weaker when divergence of status perceptions is low. We found a significant interaction effect between debate and divergence of status perceptions on group member well-being (γ = .99, p = .006, see Table 5). To further probe this interaction, we conducted simple slope analyses (the moderator variable was centered at the grand mean). Findings revealed that the higher the divergence of status perceptions, the more positive the association between debate and group member well-being. Specifically, for high divergence of status perceptions (moderator value of +0.50), the relationship was γ = .76, SE = .22, p = .001. For low divergence of status perceptions (moderator value of −0.50), the relationship between debate and group member well-being was not significant (γ = −.23, SE = .19, p = .238). The corresponding plot is displayed in Fig. 2. The moderated random-intercept model explained significantly more variance (see Table 5) than the intercept-only model and the unmoderated model (AIC = 384.72, BIC = 418.05, p = .01). The marginal R2 was .17 (Nakagawa & Schielzeth, 2013). The conditional R2 was .28 (Barton, 2022). Overall, we found support for Hypothesis 2 with the limitation that the negative relationship between debate and group member well-being is not significant in the case of low divergence of status perceptions.

Fig. 2
figure 2

Relationship between Debate and Group Member Well-Being moderated by Divergence of Status Perceptions

Full size image

Discussion

We aimed to investigate how debate in activist groups is related to group member well-being. Further, we examined the moderating role of divergence of status perceptions in the relationship of debate and group member well-being. The results mainly support our predictions: In general, group-level debate seems beneficial for group member well-being. However, the specific group context is important – debate is more beneficial when there is more within-group divergence concerning the extent to which group members perceive social status to be evenly or unevenly distributed throughout the group. Low divergence of status perceptions does not moderate the relationship between debate and well-being.

Our findings are mainly consistent with the expectations derived from literatures on debate (e.g., Simons et al., 1999) and status conflict (e.g., Bendersky & Hays, 2012). Our results add to the body of evidence showing that workers who engage in discussions and actively tackle disagreement in group settings experience higher well-being (Park, 2012; Weber et al., 2020). A potential explanation for the previous and our current findings is the fulfilment of basic cognitive and affiliative needs through debate (Park, 2012; Weber et al., 2020).

The result that the positive relationship between debate and well-being is stronger when divergence of status perceptions is high supports existing expectations from theories on status conflict (Bendersky & Hays, 2012; Kilduff et al., 2016) and expands the existing evidence with insights on debate as a way of managing dissent. When divergence of status perceptions is high, members do not behold the same idea of social status distribution, which positively interacts with a setting of high debate. The insignificance of the (negative) relationship between debate and well-being in the case of low divergence of status perceptions is likely attributable to the positive main effect of debate and well-being, which prevents potential side effects of debate from being influential. This is comparable with insights from the proactivity literature that show that a positive effect of basic needs fulfillment is relatively independent of context and not easily hampered (Cangiano et al., 2019; Röllmann et al., 2021).

Theoretical and practical implications

Our study provides relevant insights for theory and practice. Our main theoretical implications are threefold. First, using a multilevel perspective, we establish that debate is positively related to group member well-being. We further demonstrate that the extent of debate interacts with the extent of within-group divergence of status perceptions: The latter pose a boundary condition for the relationship between debate and group member well-being. This adds to the literature on participation and debate by explicitly targeting well-being (Vollmer & Seyr, 2012). It thereby helps to better understand the consequences of debate. It is distinct from existing studies, such as the study by Robijn et al. (2020), as it directly addresses the behavioral measure of debate. Further, we use a general measure for well-being and not work related outcomes such as work engagement. Our research is especially insightful for activist groups, self-managed teams, and teams with flat hierarchies as these have a higher need to negotiate roles and social status is less formalized (e.g., Luthmann, 2019). The relevance of debate is therefore enhanced in this context and consequences of inadequate debate should be more pronounced for these groups due to the absence of formal hierarchies. Our findings also offer new opportunities for refined research on debate as a team emergent state, that is cognitive, affective, and motivational states of teams (Marks et al., 2001; Rapp et al., 2021). Debate, contrary to other team emergent states such task or relationship conflict and psychological safety, has not been given much attention in team research. Our study shows that debate as a construct assessing the sharing of different perspectives in a group appears to be important as it relates to the well-being of group members above and beyond the effects of conflict and psychological safety. Future studies could focus on processes between debate and well-being to allow for casual interpretation, or could consider other contextual moderators of the association of debate and well-being such as perceived social support or leadership.

Second, our study advances the growing literature on status dynamics by showing that it is not only the distribution of essential group characteristics per se that is relevant. Theorizing and evidence suggest that status disagreement as disagreement of concrete ranking of individuals in the status hierarchy can be detrimental to group functioning (e.g., Kilduff et al., 2016). Our results expand the literature as they indicate that within-group divergence regarding the perceived extent of hierarchization of social status is also a relevant factor that should be taken into account when examining status perceptions in groups and teams. According to our study, it is central whether the fit between the group’s level of debate and the level of divergence of status perceptions is adequate. We hope that this insight stimulates new research in the field of group dynamics and status perceptions by offering a new possible lens through which status dynamics but also other group phenomena can be conceptualized. Future studies should consider within-group divergence in the perception of crucial group characteristics and possible interactions of these differences with other features of the group, such as the extent of open discourse.

Third, we advance the literature on activism by showing that individual and group-related factors from work and organizational as well as social psychology research are relevant in the study of activist well-being (see Banks, 2020). In particular, we can regard activist groups as self-managed teams. Studies reporting activist burnout (e.g., Tavarez, 2022) can be contrasted with research showing that individuals who are politically active experience higher well-being than those who are not (e.g., Watson-Singleton et al., 2021). Evidence generally suggests that activism may be a double-edged sword with findings demonstrating positive and negative associations between activism and activist well-being (Harré et al., 2009; Vestergren et al., 2017). Our research expands this literature by adopting a multilevel perspective which, thus far, has often been ignored. By understanding activist groups as self-managed work teams, adopting concepts and theories from the group and team literature may help advance the study of activism. By the same token, activism and related contexts (e.g., unpaid labor, reproductive work, social engagement) seem to present alternative and fruitful contexts to investigate other aspects pertaining to individual well-being and status dynamics.

With regard to practical implications, our results suggest that debate, as a means of open and controversial discussion, should be fostered, especially when perceptions of group members diverge concerning the social status distribution within the group. Creating an atmosphere where criticism can be expressed openly without being punished is particularly relevant in this context. Also, explicit group decisions and actively sharing dissent instead of backroom talks are a crucial step to make debate possible. It is further promising to anchor moments of individual reflection and small break-out groups before making important group decisions. This enables all group members to form an opinion in advance of discussing and deciding on important issues. It may also be worthwhile engaging in critical team-based reflection after an issue has been discussed in order to bring to surface how status perceptions might have influenced the group-decision making process (e.g., Weiss et al., 2017). Discussing how much status perceptions diverge throughout the group would further enable groups to assess how important it is for this concrete group to thoroughly strengthen debate.

Limitations and future research

Notwithstanding the strengths of our study, such as the inclusion of real teams (Salas et al., 2006; Vollmer & Seyr, 2012), there are a number of limitations that should be taken into account and addressed in future research. First, all data were collected at a single point in time (i.e., cross-sectional design). We therefore could not measure the assumed underlying processes, so that we cannot be certain as to why and how exactly the established relationships develop. Future studies should include an assessment of the underlying mechanisms with at least three points in time for data collection. Doing so would allow for temporally separating the cognitive and affective pathways that we assume to be the base of the relationship between debate and well-being.

Second, as we relied exclusively on self-report data, common method bias could be an issue (Podsakoff et al., 2012). Future research should address this problem by including objective measurements of debate via observer ratings. Separating the predictor and outcome variables by assessing different points in time and controlling for baseline levels could be another approach to diminish common method bias in future studies on debate.

Third, due to the relatively small number of groups included in the study, it is important to replicate our findings using larger samples of self-managed work teams. This would help to ensure the generalizability of the results and allow for more robust conclusions regarding the multilevel CFA (Ene, 2020; Sagan, 2019). The coefficients and variance components of multilevel regressions remain unbiased also in small level-2 samples, the standard errors can however be underestimated (Maas & Hox, 2005; McNeish & Stapleton, 2016). In the current study, we have addressed the potential difficulties associated with sample size by employing a simple multilevel model and using robust estimators. It has been shown that multilevel modeling with small level-2 sample sizes can perform well when these measures are taken (Hox & McNeish, 2020; McNeish & Stapleton, 2016). As omitting multilevel modeling can lead to serious biases, and multilevel modeling has been recommended when data is hierarchically structured (e.g., González-Romá & Hernández, 2017), or when a level-2 unit size of 20 is reached (Snijders & Bosker, 2011), we regard our approach as justified. Although McNeish and Stapleton (2016) conclude that multilevel results can be informative even with ten clusters or less, our results have to be interpreted more cautiously than results from studies with a larger level-2 sample size as, for example, the statistical power of the results cannot be ascertained (McNeish & Stapleton, 2016).

Regarding the representativeness of the sample, it is important to point out that activists are a sample that is very hard to access. However, the sample consists of groups that are active in different German cities and are diverse in terms of size and goals. Most activist studies consist of few participants and rely on qualitative analysis (e.g., Freeman, 2015; Gorski et al., 2019). A smaller number of quantitative studies (e.g., Dwyer et al., 2019, N = 299 at T1, N = 91 at T1, T2, T3; Schwartz et al., 2022, N = 284) exist, but they do not focus on the group structure of activist behavior that, however, is a very common form of activist organization.

Finally, not all members of each group took part in the survey. Thus, due to the voluntary and not compensated nature of our investigation, we cannot rule out that selection biases have taken place. Stressed, opposed, or reticent individuals not taking part can have an impact on the expression of central variables in our study, especially on the individual-level outcome variable well-being. Also, the relationships of our central study variables could differ for non-participants. As the sample is generally difficult to access for research, confirming the results with a sample of complete activist groups is a desirable but very challenging endeavor. It is again desirable to replicate the study in other self-managed work teams.

In addition to the future research ideas derived from limitations in our study, several research avenues should be pursued. It is promising to analyze if it is truly the level of debate that is essential or rather the debate climate (following research by Brykman & King, 2021, on voice climate). Concerning divergence of status perceptions, future research should take into account the potential (moderating) effect of divergence of status perceptions, also in other contexts such as groups with a more formal hierarchy. An interesting approach would be to examine how divergence of status perceptions interacts with status disagreements as pointed out in the current status conflict literature (e.g., Bendersky & Hays, 2012). Moreover, it is very promising to investigate more systematically whether divergence in the perception of other group characteristics, such as the perception of social support, recognition, time pressure, or psychological safety, is also influential.

Conclusion

Our study shows that the extent of group-level debate has the potential to enhance group member well-being in self-managed teams like activist groups. However, our study points to divergence of group members’ status perceptions as a relevant boundary condition in this context. Our results indicate that the positive relationship between debate and group member well-being is stronger when divergence of status perceptions is perceived to be high rather than low. Thus, not only the distribution of social status is important, but also how divergent group members perceive social status to be distributed throughout the group. Our findings thus advance the study of debate and conflict management as well as status dynamics in groups and point out implications for individual well-being.