Abstract
The power-status theory of emotions posits that emotions arise out of a process in which people compare the levels of status and power that they experience against some standard of status and power adequacy. Power-status theory is effective at predicting emotions, but it does not formally specify a structural mechanism that explains how actors determine what constitutes adequate or inadequate power and status. The authors argue that legitimacy, or the notion that actors have a right to expect compliance and deference, serves as this mechanism. Focusing on shame, the authors illustrate that perceived losses in legitimacy mediate a substantial portion of the links among status loss, power loss, and shame. The authors also find that legitimacy mediates the relationship between a person’s position within an organization and the shame they experience when their status and power are challenged. These findings advance power-status theory by providing a structural standard for status and power adequacy that explains how losses in status and power generate emotional outcomes.
Theory and research within the sociology of emotions explains how structural and cultural processes affect people’s emotional experiences and expressions, as well as how emotions interact with group structures to shape group outcomes like patterns of deference or group solidarity (Collins 2004). Focusing on the social structural bases of emotional experience, Kemper’s (1991, 2006) power-status theory of emotions provides a useful framework for predicting what emotions will arise in a given social context and for explaining the interactional mechanisms that produce specific emotions. Kemper and Collins (1990) argue that every interaction entails both status and power outcomes for all actors (e.g., status gain, loss, or maintenance), and Kemper (2006) articulates how these structural dimensions map onto actors’ emotional experiences. Several studies have shown support for key predictions from the power-status theory of emotions (see, e.g., Hattke, Hensel, and Kalucza 2020; Kemper 1991; Lovaglia and Houser 1996).
Despite strong empirical support for power-status theory, research on the relationships among status, power, and emotions nonetheless points to a number of opportunities to advance scientific understanding of the structural mechanisms that produce emotions. One such mechanism, legitimacy, plays an implicit, though largely unarticulated role in power-status theory’s account of the ways that social structure affects emotional outcomes. Although power-status theory specifies that emotions arise through a process in which people compare interactional outcomes against some level of expected status and power (Kemper 2006), the theory does not specify where these expectations come from or how people determine what levels of status and power they deserve in a given situation.
On the cognitive and affective bases of status and power expectations, Kemper (2006) assumes that such standards are “deeply held structural parts of personality and identity” (p. 106) but does not provide a clear mechanism for how people acquire an understanding of these standards or how they assess their relative standing during interaction. Rather than treating expectations for status and power as idiosyncratic features of interactants’ personalities or as unspecified aspects of their identities, we argue that legitimacy (Walker, Thomas, and Zelditch 1986; Zelditch 2006), or the collective notion that an actor has a “right” to expect compliance or deference, provides a structural explanation for expected interactional outcomes that incorporates existing power and status dynamics between actors.
In this research report, we develop and test an argument that perceptions of legitimacy mediate the effects of status and power on a key emotional outcome: shame. Shame is a broadly studied social emotion and is a central emotion in Kemper’s articulation of power-status theory. We use a large, online two-stage vignette study to test power-status theory’s claims regarding the structural bases of emotions along with our model incorporating legitimacy as a mediating variable. In the first stage, respondents read a vignette that presented a scenario about an employee who works in a demanding marketing job for a U.S. corporation. The vignette clearly articulates the worker’s structural position as either an executive manager, middle manager, or employee without management responsibilities, and paints a picture of a productive worker who must balance the demands of work and family. At this stage, respondents provided ratings of the worker regarding their perceptions of the worker’s power, status, and legitimacy within the organization.
In the second stage of the study, participants read about a scenario where the worker’s power, status, and legitimacy are potentially threatened by a complaint from a large client about a recent job overseen by the worker. Following the vignette, we asked participants to reassess their perceptions of the worker’s power, status, and legitimacy in light of the complaint, and asked them to predict the worker’s expected level of several possible emotional responses. This strategy allows us to assess respondents’ perceptions that the worker experienced loss of status, power, and legitimacy and relate those perceptions to emotional outcomes. Although power-status theory offers predictions regarding a wide range of emotional outcomes, our analyses focus on a single emotion, shame, as an initial test of our model and leave the examination of other possible emotional outcomes for future research. We estimate models that account for an actor’s perceived legitimacy and assess whether gains and losses in legitimacy mediate the relationship between status loss, power loss, and shame. We also assess whether worker’s hierarchical position, as identified by manager status, predicts levels of shame.
As noted earlier, this research provides an initial test of our theoretical argument incorporating legitimacy into power-status theory. Vignette studies such as those we discuss provide insight into how people perceive actors in a hypothetical situation and how they predict or expect those actors will behave or what emotions they will experience. Because we measure perceptions and not actual behaviors, we view the present study as a first step toward demonstrating the value of incorporating legitimacy into power-status theory. Acknowledging this limitation, the present study contributes to the sociological understanding of emotional processes by (1) testing power-status theory’s claims regarding the structural bases of emotions, (2) identifying legitimacy as a previously unarticulated mechanism within the theory, and (3) assessing our model in a formal organization where status and power are tied to hierarchical positions.
Theoretical Background
One of the most long-standing areas of interest within the sociology of emotions focuses on explaining how emotions emerge in interactions on the basis of factors such as actors’ structural positions and cultural norms regarding emotional experience and expression (Hochschild 1979; Kemper 1981). In treating emotions as structural outcomes, sociologists have devoted considerable attention to examining how differences in status and power between interactants shape their affective experiences (Lovaglia and Houser 1996; Simon and Nath 2004). Another line of research has called attention to the role of legitimacy—the collective belief that certain actors possess the legitimate right to expect compliance and deference (Zelditch and Walker 1984)—in determining emotional responses to injustice (Clay-Warner 2006; Johnson et al. 2016). We integrate these lines of inquiry by exploring how legitimacy affects how people respond emotionally to the ways their status and power shift during interaction.
Power-Status Theory of Emotions
Power-status theory posits that every interaction produces structural outcomes along two dimensions, power and status, and that these structural outcomes trigger specific emotional responses (Kemper 1991, 2006). Here, power refers to an actor’s relative ability to compel compliance from an interaction partner (Weber [1921] 2019), while status encompasses relations of voluntary deference. Focusing on dyadic relations, Kemper (2006) states that in any interaction, each actor’s power and status can rise, decline, or remain the same. Overall, power-status theory assumes that interaction partners desire to maintain expected levels of status and power for self and other (Kemper 2006).
Our research is concerned entirely with consequent emotions, which are the emotions that result from immediate interactions. When people experience their power and status as adequate, they tend to experience positive emotions, while deviations from the expected levels in either direction tend to produce negative emotions. For example, the theory predicts that when one’s power is adequate, they will feel secure, but tend to feel negative emotions when they perceive that their power is either excessive (guilt) or inadequate (fear). Specific consequent emotions correspond with each possible combination of power and status gains, losses, and maintenance between actors.
Shame, Power, and Status
Power-status theory offers predictions for a range of positive and negative emotions. As an initial test of our model, we focus on structural mechanisms that produce shame. Shame reflects a negative evaluation of the self as individuals assess their position and identity relative to others (Leeming and Boyle 2013). Many scholars consider shame to be the quintessential social emotion that promotes adherence to norms and rules and has the potential to repair or destroy social bonds (Braithwaite 1989; Scheff 1988). Scheff (1988) argues that shame comes about as individuals constantly evaluate themselves within social interactions, and most individuals carry at least low levels of shame. Shame is also key in Kemper’s (1978) initial articulation of power-status theory. Power-status theory predicts that shame is uniquely associated with experiences of status insufficiency (Kemper 1991). When individuals believe that they have received more status than they deserved, they experience shame (Kemper 1978, 2006). Power-status theory also posits that if an unworthy action tarnishes a status claim, then there is a perception that one has wronged the group or behaved incompetently, and this also results in shame. On the basis of power-status theory, we formulate the following hypothesis:
Kemper does not directly identify shame as a consequent emotion related to power loss. However, Kemper and Collins (1990) contend that power loss destroys confidence. Although we acknowledge that Kemper and Collins argue that fear and anxiety are the primary responses to power loss, we also acknowledge that researchers have identified shame as an underlying emotion undergirding low confidence and esteem (Jacoby 2016; Owens 1993). Thus, our argument provides exploratory prediction to expand our understanding of the underlying mechanisms in power-status theory. We draw on scholarship identifying self-confidence, or lack thereof, as a root cause of shame. The self-confidence that is lost when power is lost, as theorized by Kemper and Collins, may be a consequence of elevated shame. To examine this possibility, we test the following exploratory prediction:
Legitimacy, Power, and Status
Within power-status theory, the structural comparisons that produce emotions (e.g., adequate vs. insufficient status) assume that individuals weigh their experience of status and power against some baseline expectation for what they believe they are due in the situation. Although a person’s expectations for power and status might arise from any number of sources or deeply held personality structures, we argue that baseline expectations are not so idiosyncratic, but are shaped by structural factors related to legitimacy. Because legitimacy represents a collective understanding of what constitutes appropriate social relations (Zelditch 2006), incorporating legitimacy into the power-status model can offer insights into the structural determinants of emotion and allow more precise predictions regarding what emotions will arise under different conditions.
At a basic level, legitimacy is the collective sense that something is “right” (Zelditch 2006). When applied to interactional power and status hierarchies, legitimacy refers to the collective belief that actors’ positions within that hierarchy are appropriate. When a group’s status order is legitimated, higher status actors can expect deference from those lower in the status order (Ridgeway, Johnson, and Diekema 1994). Similarly, when a group’s power order is legitimate, power relations stabilize and certain actors are collectively granted authority rights and can expect compliance from others (Walker et al. 1986).
Groups with legitimated power and status orders have members that generally understand their place within the hierarchy. In such groups, there are two sources of legitimacy: (1) authorization, in which an actor higher up in an organizational hierarchy designates or supports an authority structure, and (2) endorsement, in which an actor’s peers or those below them in the hierarchy support their position (Johnson, Ford, and Kaufman 2000). Authorization and endorsement are not characteristics of individual actors, but reflect a collective understanding of group structure. It does not matter whether an actor’s peers or subordinates actually support them; endorsement requires only that people perceive that others support the higher status or more powerful actor (Walker et al. 1986). In legitimated structures, perceptions of legitimacy are key to understanding what constitutes appropriate levels of status/deference and power/compliance between actors.
Research, mostly tied to distributive justice, links legitimacy to emotional responses. These studies have found that legitimacy affects people’s emotional reactions to injustice (Johnson et al. 2000, 2016). Indeed, authorization and endorsement reduce the likelihood that group members will express negative emotions in response to an unjust distribution of resources (Hegtvedt and Johnson 2000). Accordingly, legitimacy is a structural “opposing force” to the negative emotions that people report in response to injustice (Johnson et al. 2016:96). More broadly, legitimacy may serve as a protective factor in situations where status and power are threatened.
Bringing Legitimacy into Power-Status Theory
Existing work on emotions and legitimacy dovetails with the overall model of power-status theory. In discussing how people perceive their status and power in a given context, Kemper (2006) suggests that we have a sense of the status and power that we are due. That is, we perceive that each actor has a legitimate status and power position. Legitimacy theory tells us that this shared understanding of appropriate levels of power and status is rooted in processes of authorization and endorsement. Stated differently, legitimacy defines the levels of deference and obedience that actors can expect. A failure to receive expected support from an actor higher up in the organizational chain can signal a loss of authorization. A failure to receive expected support from a peer or subordinate group member can signal loss of endorsement. As such, in legitimated structures, authorization and endorsement provide the standard against which a person can assess whether their current experiences of power and status reflect adequacies, surpluses, or deficits.
When assessing whether one has gained or lost status or power in an interaction, it is not necessary to have a standard beyond the immediate situation. In an organization, in which legitimacy structures are tied to positions, it is possible for an actor’s behavior in a situation to result in a shift in the actor’s legitimacy without a shift in the legitimacy of the position occupied (Evan and Zelditch 1961; Walker et al. 1986). Thus, although individuals hold legitimated positions, local situations can alter the legitimacy of the actor.
To the extent that losses or gains in status or power can signal an inappropriate surplus or insufficiency, then this sense of loss threatens the legitimacy of the actor, generating an emotional response. When an individual loses status and power, then their legitimate right to deference is threatened. To the extent that status and power losses trigger losses in perceived legitimacy, these losses should trigger negative emotions because the individual has lost legitimacy tied to a structural position that provides them with deference and obedience. For our example, we expect that having a client complain to a superior about the quality of work should delegitimatize our hypothetical worker and directly undermine their power and status. Accordingly, we formulate the following hypothesis:
And following from the foregoing logic, alongside our argument related to the relationship between power loss and shame, we offer the following exploratory prediction:
Organizational Roles and Legitimacy
As previously noted, in this research we use a very specific structured organizational context, a corporate workplace in the United States, where hierarchical relationships between individuals are formally defined, and there is differentiation between executive managers, midlevel managers, and workers without management responsibilities. These roles activate perceptions of status and power among the members of the group. Higher status individuals are often promoted into positions that have greater power, and the status and power associated with these higher level positions are relatively stable and self-sustaining (Chen et al. 2012, Magee and Galinsky 2008). Thus, managers are expected to have higher status and power, and this expectation is self-reinforcing (Magee and Galinsky 2008). Studies show that even if individuals were randomly assigned to status and power positions, the higher ranked individuals tend to be evaluated more favorably (Humphrey 1985).
Executive-level managers have typically experienced multiple levels of promotion and are accorded the status and power of these elite positions. Middle-level managers, on the other hand, work in more precarious positions. In fact, many middle level managers identify more with their subordinates than with their leaders (Gjerde and Alvesson 2020), and in the modern era of globalization where internal labor markets have diminished, the career prospects for middle managers are dimmer, leading to a greater divide between executive- and middle-level management (Hassard, Morris, and McCann 2012).
Although any actor within an organization or group can lose status on the basis of adverse workplace interactions, higher ranking individuals, such as executive managers, are more protected than lower ranking individuals as they are allowed more latitude and variability in behavior (Guinote et al. 2002), and individuals in lower positions in the social hierarchy are at greatest risk of losing their status and power (Magee and Galinsky 2008). For example, a client’s complaint about an executive manager’s work quality may not result in the same degree of status loss as it would for a lower level worker because higher level executives may be afforded greater benefits of doubt regarding their job performance. Burrill (2021) demonstrates further that individuals who occupy higher status positions often maintain a positive view of themselves, and that expectations for competence are “sticky,” in that higher status actors retain their sense of competence, even when moving to lower status positions. Given research demonstrating that individuals with higher status and power are protected from many negative effects related to adverse events, we expect that respondents will perceive that managers will experience less shame than workers, with executives experiencing the least shame, and we formulate the following hypotheses:
Hierarchical structures are more stable when positions within the structure are legitimated (Magee and Galinsky 2008). Given that legitimacy is contingent on norms associated with roles, along with perceived authorization and endorsement of actors within those roles, and the titles of manager versus nonmanager are widely agreed upon roles, management positions themselves are legitimated power and status structures, and occupying a higher position indicates some level of authorization. Yet higher level leaders are capable of losing legitimacy. Indeed, when others within the structure believe that a higher ranking actor is abusing their position or if the system is deemed inequitable, the organizational structure may lose validity and become illegitimate (Magee and Galinsky 2008). Executive managers work in positions with the greatest legitimacy and thus should be protected from a loss of legitimacy, while middle-level managers work in more precarious positions, making it unclear how perceptions of legitimacy will shift. We hypothesize as follows:
More specifically, executive legitimacy is expected to be robust, and legitimacy should not explain their already low shame. In contrast, legitimacy is fragile for middle-level managers. Thus, although we expect respondents to perceive managers to experience lower shame than their working counterparts, a loss in legitimacy could increase perceptions of shame. In sum, we predict that losses in power and status trigger losses in perceived authorization and endorsement. Reduced legitimacy, then, will predict the emotions that arise in the interaction. In what follows, we describe a test of our predictions and present the results of our analyses.
Method
We analyze data from an online vignette study. The data we analyze for this study were originally collected for a separate project examining status and power dynamics in organizations. We focus on formal organizational settings because they provide a clear framework for assessing participants’ perceptions of an actor’s legitimacy (i.e., support from a superordinate and from peers and subordinate group members). Vignette studies like this one have been shown to be effective at testing predictions regarding both status and power outcomes (Webster and Dippong 2022) and emotional processes (Johnson et al. 2000).
Participants
We collected data using a Qualtrics survey distributed through the Prolific online platform. The original study sampled American Prolific respondents 18 to 59 years of age who were currently working full-time. Although the sample is not necessarily representative of the U.S. working population, our goal here is not to estimate population parameters or to generalize from our sample to a larger population. Our goal is to test the theoretical argument we present and as such we are not drawing inferences about the population, but rather we are drawing inferences about the validity of our model (Dippong and Jillani 2025). On the basis of the design of the study for which these data were collected, Qualtrics randomly assigned individuals to 1 of 12 conditions, with 100 respondents in each condition. The original study included data from 1,213 respondents, and with the exception of excluding participants for failed attention checks (described later), our data includes the entire sample from the original study. Because the manipulations used for the original study are not relevant to the questions we address here, we collapse data across all nonrelevant conditions for our analyses. 1
Over the course of participating, respondents completed a two-phase study, involving two separate but related vignettes and responded to two sets of survey questions. To ensure data quality, we included three attention check questions at the outset of the questionnaire assessing participants’ recollection of the three manipulations (gender, management status, and work location). Respondents were given two opportunities to correctly answer each manipulation check, and for participants who failed at least one of the checks twice (n = 6) the study immediately concluded. To further ensure data quality among those who passed the initial attention checks, our analyses include only individuals who provided accurate answers to two additional attention checks embedded later in the questionnaire. Eight individuals failed both of the subsequent attention checks or declined to answer the attention checks and were removed from the final analytic sample.
Procedures
All participants in the analytic sample completed both phases of a two-phase vignette study. In phase 1, respondents were presented with a version of the following scenario. For ease of presentation, we present the following scenario with manipulated variables represented by one of the manipulated options in brackets. These manipulated options were presented randomly across respondents. [Does not have management responsibilities] could also be [is a midlevel manager] or [is a high-level executive manager]. [Michelle] could also be [John]. [Work from home] could also be [work in-person at the corporate office]. We do not examine these latter two manipulations in the present study. The vignette holds age, number and age of children, marital status, occupation, work experience during the COVID-19 pandemic, and productivity constant.
[Michelle] is married, 35 years old, and has two young children, a toddler and an infant. [Michelle] works full-time in a demanding marketing job for a major U.S. corporation. [Her] role requires continuous interaction with her team. [Michelle] [does not have management responsibilities]. Typically, while [Michelle] is at work, their children go to daycare. During the recent pandemic, [Michelle’s] company required individuals to work from home. This was helpful because [Michelle] and [her] spouse were unable to send their children to daycare during the pandemic, and [Michelle] was able to work at home alongside the young children. Nevertheless, [Michelle’s] productivity remained similar to her peers. Because many of the pandemic-related restrictions have been lifted, Michelle’s two young children recently returned to full-time daycare. However, [Michelle] still has to deal with unexpected childcare responsibilities that occasionally arise during the workday. Michelle continues to work in the same demanding marketing position at the same company where [she] [does not have management responsibilities]. [Michelle’s] company is preparing to re-open the offices. Some employees are going to be required to return to the office, while others will be allowed to continue working from home. Each person will present their case on whether or not they should work at home or return to the office. The company will consider people’s preferences, alongside considerations for what is best for the success of the business. [Michelle’s] preference is to [work from home].
After reading this vignette, participants responded to a series of questions in which we assessed their baseline perceptions of the vignette worker’s power, status, and legitimacy.
Following the first questionnaire, the respondents completed a second phase of the study. In phase 2, the respondents read a follow-up vignette designed to create changes in participants’ perceptions of the worker’s status and power. To create this change, we used an adverse workplace event in which the worker’s boss received a criticism about the worker: We now want you to think about [Michelle]’s job experiences in the upcoming year. Please read the following paragraph carefully. Six months have passed since [Michelle]’s company re-opened the office. A large client that has been with the firm for five years is dissatisfied with the results from a recent job overseen by [Michelle]. The client complained to [Michelle]’s boss and threatened to stop working with [Michelle]’s company.
After this second vignette, participants responded to questions in which we asked them to assess the worker in terms of power, status, and legitimacy in light of the adverse event. We also prompted them to “put themselves in the worker’s position” and assess the extent that the workers would experience a range of negative emotions, such as shame. This allows us to assess the extent to which changes in these variables predict perceived emotional responses to the adverse event. Figure 1 presents an overview of the study design.
Study design.
Variables
Our analytical models analyze participants’ perceptions of the likelihood that the employee will experience shame in response to the adverse workplace event. We test our predictions using models that include two focal predictors, namely changes in the focal actor’s status and power. Additionally, we estimate models that include reductions in perceived legitimacy as a mediating variable. Last, to examine the differing effects of status, power, and legitimacy change across organizational roles, we estimate a series of models that predict shame from the vignette worker’s management status. Our models also control a variety of respondent characteristics.
Dependent Variables
We measure perceptions of the employee’s levels of shame using the shame subscale of the State Shame and Guilt Scale (Tangney and Dearing 2002). This subscale is designed to measure immediate experiences of shame, rather than long-term emotional patterns or traits and is well suited to assess shame in response to an event and/or within a specific context. For the shame scale, we informed respondents that they should imagine that they were in the employee’s position and to report the likelihood that the employee would experience a range of shame-related feelings while discussing the complaint with their boss. The shame subscale includes items assessing 10 different shame-related feelings: wants to sink into the floor and disappear; feels small; feels like a bad person; feels humiliated, disgraced; feels worthless, powerless. We assessed internal consistency using categorical principal components analysis, which relaxes the assumptions of linear relationships between variables, does not assume that variables are multivariate normal, and permits analyses of ordinal data (Linting et al. 2007). The shame subscale demonstrates strong internal consistency (α = .91).
Independent Variables
As noted earlier, we measured perceptions of status and power both before and after the adverse workplace event. On the basis of power-status theory’s emphasis on interactional outcomes, we estimate models in which we predict emotions from changes in perceived status and power. At both time points, we measured status using a single item, which asked participants the extent to which they agreed or disagreed with the statement, “[Michelle/John] seems like the kind of person who has high status in her/his organization.” As such, status is framed in terms of standing or esteem within the organization. Scores ranged from (1 = “strongly disagree”) to (5 = “strongly agree”). We calculated changes in status by subtracting the phase two measurement from the phase one measurement (status change = T1 − T2). Accordingly, positive values reflect losses in status, while negative values reflect status gains. In a separate measure, participants reported whether the employee probably has a lot of power over other people at work.
Perceptions of power were measured on the same five-point scale as status perceptions, using the item “[Michelle/John] probably has a lot of power over other people at work,” which again anchors the definition of power as relative to other employees within the organization. Again we calculated changes in power by subtracting phase two measurements from phase one measurements (power change = T1 − T2). Given that power and status are tied to roles in formalized organizations, we also estimate separate models that predict shame from the organizational role assigned to the worker. These roles include executive manager, middle manager, and nonmanager.
Mediating Variable
We use a two-item measure of legitimacy, again measured at two time points. One item assessed perceived authorization of the employee by superordinate actors within the organization (“People at the top of the organization probably expect [Michelle/John] to act as an authority on her/his team”), and a second item assesses perceptions of endorsement from peers and below (“People on Michelle/John’s team likely see her/him as a legitimate leader”). As with our measures of status and power, our goal was to anchor perceptions of legitimacy within the context of the organization. The two items are highly correlated (r = .82) and therefore warrant being combined into a single item. Because we argue that changes in perceived legitimacy mediate the relationship between status, power, and shame, we measured perceptions of authorization and endorsement at two time points, before and after the adverse workplace event, and calculated change using the difference between the two measurements (legitimacy change = T1 − T2). Positive values reflect loss of legitimacy and negative numbers reflect gains.
Control Variables
Our models control for key demographic characteristics of respondents, including age, respondent gender (1 = cisgender female/woman), race (1 = white) and education level (1 = less than a bachelor’s degree).
Analysis
We examine our theoretical questions using generalized structural equation modeling (GSEM) with robust standard errors to estimate path models connecting perceived changes in status, power, and legitimacy to levels of perceived shame. Figure 2 presents our path model. We begin by modeling the direct effects of power and status change on our dependent variables. Next, we examine the extent that legitimacy mediates the effects of power and status on emotions, and where appropriate, estimate the proportion of the total effect that the mediator explains. Finally, we examine the effects of organizational role on perceived experiences of shame.
Path diagram for generalized structural equation model.
As can be seen in Figure 2, although we use GSEM to estimate our path models, the models include only observed variables and do not include any latent variables. Additionally, it is worth noting that GSEM does not produce most of the typical indicators of model fit associated with structural equation modeling (e.g., confirmatory fit index). Because we estimate saturated models and use robust standard errors, most measures of comparative fit are inappropriate for the models we report.
Results
Our overall sample size was 1,202, though 12 respondents provided incomplete data on at least one key variable in our model. After listwise exclusion of participants with incomplete data, our final analytical sample size is 1,190. Table 1 presents our sample characteristics. Our respondents ranged between 25 and 59 years old, with a mean of 37 years. Our sample is, on average, more educated than the general public, with approximately 70 percent reporting an education of a bachelor’s degree or higher (compared with 38 percent as reported by the U.S. Census Bureau in 2022). Our sample is roughly evenly split between participants who report their gender identity as man and those who report as woman, with an additional two percent reporting other gender.
Participant Characteristics (n = 1,198).
Descriptive Statistics
Table 2 presents descriptive statistics for our key theoretical measures. Looking at our shame measure, we see in Table 2 that average shame ratings hover around the midpoint to the index (15.77 on a scale ranging from 5 to 25). For our independent variables, looking first at status, on average participants initially rated the target just above the midpoint of the 5-point scale (3.69), and this score was reduced by an average of 0.831 points following the adverse workplace event. Ratings of power mirror those of status, with an initial mean around the scale midpoint (3.43) and an average loss of less than 1 point on the scale (0.590). Initial legitimacy perceptions were slightly above the midpoint of the nine-point scale (7.62), with an average loss of 1.74 points following the adverse event. Paired-sample t tests demonstrate differences between initial and postincident perceptions of the employee’s status (t = 31.32, p < .001), power (t = 23.03, p < .001), and legitimacy (t = 31.96, p < .001), with all three variables reflecting a significant reduction following the negative event. These results provide strong evidence that the workplace incident scenario produced the intended effects on respondents’ perceptions.
Descriptive Statistics (n = 1,189).
Significant difference compared with executive manager (p < .001).
Significant difference compared with midlevel manager (p < .001).
Significant difference compared with midlevel manager (p < .05).
Independent-samples t tests show that the gender of the focal actor had no significant effect on perceptions of shame (p > .90, two-tailed). Similarly, t tests show that the gender of the target had no significant effect on initial status, power, and legitimacy ratings, though respondents perceived that men employees lost more power than women employees (t = 2.84, p < .01). In terms of managerial roles, we expected to see differences in power, status, and legitimacy across organizational positions, as different position labels connote differing levels of status, power, and organizational support. One-way analyses of variance reveal significant differences across managerial levels in terms of status loss (F = 3.15, p = .04) and legitimacy loss (F = 6.50, p = .002), but not power loss. There were also significant differences in perceived emotional responses for shame (F = 15.52, p < .001). These results are driven primarily by perceived differences between managers and nonmanagers (analyses available upon request).
Path Models
Turning next to our predictive models, we examine the direct effects of status and power loss on experiences of shame. In models 1 and 2 in Table 3, we see that the effects of both status loss and power loss are significant and positive. Greater perceived losses in status and power are associated with greater perceptions of shame, though the effects of status are somewhat stronger. Thus, both status and power loss predict greater levels of perceived shame, and these relationships are robust to controls for respondent characteristics. This provides support for hypothesis 1 and exploratory prediction 1. This is somewhat contradictory to what power-status theory would predict, as shame is seen as uniquely associated with status outcomes. Here, we find that status and power loss are associated with greater shame. This lends support to our argument that losses in self-confidence associated with power loss are a mechanism through which power loss can lead to shame (though we do not directly test this specific mechanism). It is also worth noting that one of our control variables, age, is significantly related to perceived emotional response as younger respondents perceive significantly more shame.
Effects of Status and Power Loss on Shame.
p < .05. **p < .01. ***p < .001.
Next, we incorporate the effects of legitimacy in Table 3, model 3. Although not shown in Table 3, changes in status and power are both significantly and positively predictive of legitimacy loss (status loss: b = .835, SE = .076, p < .001; power loss: b = .599, SE = .082, p < .001). The results illustrate that changes in legitimacy are significant and positive, with greater loss of legitimacy generating greater shame. Once legitimacy is included in the model, the effects of status loss and power loss on shame become nonsignificant. Specifically, when we include legitimacy loss in the model, the effect of status loss on shame is reduced to nonsignificance, and mediation analyses demonstrate that legitimacy mediates approximately 57.5 percent of the relationship between status loss and shame. We find similar results for power loss. Once legitimacy loss is included in the model, power loss becomes nonsignificant. Legitimacy loss mediates 76.6 percent of the relationship between power loss and shame. These results support hypothesis 2 and exploratory prediction 2.
For our next set of analyses, we replace direct measures of shame and power with our hypothetical employee’s management status given that we test Kemper’s theory within an organization with formalized roles tied to status and power (see Table 4). Model 1 presents the effects of management status without controls for respondent characteristics, and model 2 adds these controls. We find in model 1, as expected, that our respondents expect an employee who is either a middle or executive managers will have lower shame than workers with no management responsibilities. The effect of middle management is not robust to controls, however, and as such, support for hypothesis 3b is somewhat tentative. We also contrasted the effects of our hypothetical worker being an executive or middle manager from model 2 (with controls) and found that executive managers were expected to have significantly lower shame than middle managers (χ2 = 13.03, df = 1, p = .0003) and executive and middle managers jointly have lower shame than workers (χ2 = 32.02, df = 2, p = .0000). This provides further support for hypothesis 3a.
Effects of Organizational Role on Shame.
p < .05. **p < .01. ***p < .001.
Table 4, model 3 adds legitimacy loss. The significant effect reiterates the findings from Table 3 as legitimacy loss is associated with greater shame. Looking first at executive managers, being an executive manager was not significantly related to perceptions of legitimacy loss, with nonmanagers as the reference category (b = −.195, SE = .128, p = .126). The lack of a significant relationship between executive manager and perceived legitimacy loss is partially attributable to the smaller percentage of executive-level managers who experienced a perceived decline in legitimacy. Only 62 percent of executive-level managers had a decline in ratings of legitimacy postcomplaint. Meanwhile, 73 percent of middle-level managers experienced a decline in legitimacy ratings, and 68 percent of workers experienced a decline. Given the nonsignificant effect of executive manager on our mediating variable, we find that legitimacy does not significantly mediate the relationship between executive-level managers and shame.
The effects for midlevel managers, who experienced the greatest decline in legitimacy, painted a different picture. The midlevel manager role was significantly related to perceived loss of legitimacy (b = .294, SE = .140, p < .05). Once we control for legitimacy, the effect of midlevel manager becomes significant. This suggests a suppression effect for middle managers as perceptions of legitimacy loss masked an underlying negative association between middle management status and shame. Without accounting for legitimacy loss, middle managers did not significantly differ from nonmanagers in shame. However, once legitimacy loss was included, middle managers exhibited significantly lower shame, suggesting that their heightened experience of legitimacy loss counteracts what would otherwise be lower shame. Thus, a loss in legitimacy is associated with shame. Middle managers experience more legitimacy loss than nonmanagers because they have structurally precarious authorization and endorsement (e.g., limited power) but also greater accountability to executive management than the average worker. Their distinct managerial experience helps explain why reduced legitimacy offsets their lower shame.
Discussion
Our research contributes to theory on emotions by formalizing a mechanism through which power and status structures generate emotions. Power-status theory implies, but does not fully explicate, a role for legitimacy as actors compare realized power and status against baseline expectations of what they should experience. We illustrate that legitimacy is key to understanding why the structure of social relations produces specific emotions.
We find that legitimacy explains how status structures shape these emotional outcomes. A change in status shifts perceptions of legitimacy of an actor’s position in the hierarchy, and this invokes greater shame after an adverse event. Actors respond emotionally because their perceptions of appropriate social relations related to authorization and endorsement (i.e., the legitimated structure) are challenged. This directly advances the theoretical understanding of emotions by modeling the mechanism through which people assess appropriate levels of status and power in interaction.
We anticipated that loss of legitimacy would mediate the effects of both power and status, and found support for this claim, at least in formal organizational settings. When we focus on structurally generated positions associated with both status and power, we find that respondents report perceptions that workers in positions with the greatest and most legitimate power and status will have the lowest shame and smallest change in legitimacy. Although respondents perceive that workers with moderately high status and power and precarious legitimacy may have lower shame than lower level workers, this is offset by the precariousness of legitimacy. Thus, the extent that legitimacy serves as a mediator depends on the amount of legitimacy that an actor can lose. When legitimacy can erode (i.e., when status is contested) there is an emotional consequence. When legitimacy is insulated, structural advantage governs emotions directly.
In our research, power and status structures are formalized through occupational rank, and as such, our analyses cannot be directly applied to situations in which power and status structures have not been legitimated. That is, of course, that legitimacy can only function as a mediator in interactions with a legitimated power and/or status order. Kemper (1991) acknowledges that stable power structures, like the one used in this research, have less frequent invocation of devices to maintain power (i.e., confrontation, manipulation, etc.). Furthermore, even with invocation, power structures are formalized. We examined the effects of legitimacy in only one specific context. It is possible that a larger number of power and status threatening events or perhaps more extreme events would generate a scenario where a sufficient level of lost power results in a sufficient level of lost legitimacy to generate an emotional response. Future research should test this.
It is worth noting that status and power are multifaceted constructs that scholars have conceptualized in a variety of different ways. Concerning definitions of status, Chen et al. (2012) note that across various literatures, the term has been used to denote a person’s relative standing in terms of group membership (for example, one’s status as a member of an ingroup or outgroup), as well as one’s place within hierarchical structures of dominance, prestige, and competence. Alternatively, Kemper and Collins (1990) define status in terms of “voluntary compliance to other actors . . . marked by willing deference, acceptance, and liking . . . [involving] the voluntary provision of rewards, benefits, and gratifications without threat or coercion” (p. 34). Similarly, the concept of power has been used to describe relations in which one actor or group is able to achieve their desired ends, even when facing resistance (Kemper and Collins 1990; Weber [1921] 2019), as a product of relations of dependence (Emerson 1962), or as the ability to extract valued resources from others on the basis of one’s position within an exchange network (Thye 2000).
One limitation in the present study is that the measures we use to assess status and power do not differentiate between the various possible social meanings of these words, and as such, it is possible that respondents interpreted these concepts inconsistently. That is, our vignette described a fictitious actor and asked respondents to rate that person in terms of status and power within their organization, but we did not anchor those concepts in specific definitions that respondents could draw on in making their assessments. Given the potential conceptual overlap between status and power (i.e., dominance as a basis for status and dominance as a use of power to achieve one’s intended end), it remains possible that participants were unable to clearly distinguish status from power. This is a particularly noteworthy limitation, given that our analyses explicitly aim to model the unique effects of status and power on shame. Considering potential variability in respondents’ understandings of status and power, it is possible that our results would have turned out differently were we able to more clearly partition these two concepts. Future research is needed to assess how status and power operate to affect emotions when they are clearly defined and distinguished, as well as whether and how each unique facet of status and power contributes to the experience of shame and other emotions.
As a vignette study, the findings we present represent only the first successful test of our model. We have provided evidence that our theoretical arguments hold insofar as we have modeled the mediating effects of legitimacy as they relate to people’s perceptions of what is likely to occur in a fictitious scenario and related to a single emotional outcome. Establishing the validity of our model more fully would require behavioral data that assesses emotional outcomes to actual events and using a wider range of both positive and negative emotions. As a next step, future research should examine how status loss, power loss, and legitimacy loss affect emotional outcomes using a laboratory experiment that replicates our vignette design by creating a legitimated power and status order and assessing how changes in legitimacy affect the level of shame that people experience. As a first step, however, the current study demonstrates substantial promise for our theoretical model.
Conclusion
Power-status theory is generally effective at predicting emotional outcomes, but the theory lacks a more nuanced articulation of the mechanisms that explain why emotions develop out of power and status structures. Our research integrates status-power theory with legitimacy theory to identify a structural mechanism that accounts for emotions. It also produces a clear call for research to enhance understanding of the mechanisms that clarify how power-status structures frame human interaction.
Footnotes
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported, in part, by funds provided by the Faculty Research Grant program at the University of North Carolina at Charlotte.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.