Abstract
Power, defined as an individual’s perceived capability to influence others, is typically studied in work or romantic relationships, leaving it unclear whether findings can be generalized to other types of relationships, such as friendship. In this study, we applied a dyadic perspective to examine how power shapes relationship functioning between friends. We used an actor-partner interdependence framework with 248 friendship dyads to analyze four aspects of power (i.e., actor power, partner power, perceived partner power, desired power) and their links to global friendship evaluations (i.e., friendship satisfaction and friendship quality) and pro-relationship characteristics (i.e., compassion, trust, and forgiveness). We found that greater power in actors, greater power in partners, and perceptions of high power in friends were associated with more positive global friendship evaluations, higher compassion, and greater trust. By contrast, greater desired power was associated with more negative global friendship evaluations and lower pro-relationship characteristics. Control analyses showed that the findings were largely robust against potential effects of commitment and social desirability. The findings advance existing theories of power by (a) illustrating its dyadic nature, (b) showing that theories of power should incorporate friendship relationships, and (c) highlighting that some power-related findings may be context-specific and are not likely to generalize across different types of relationships.
Introduction
Friendships are a central part of people’s social lives (Hojjat & Moyer, 2017). Most people report a large number of friends, with an average of around 16 friends (Altmann, 2020). Supportive friendships provide sources of emotional support, act as a platform for developing social skills, can positively impact psychological and physical health (Bagwell et al., 2016; Demir & Weitekamp, 2007), and thereby have a strong impact on the self (Baumeister & Leary, 1995; Neyer & Lehnart, 2007). Friendships are close social relationships that differ from romantic relationships, kinship, and work relationships. Compared with romantic relationships, friendships are nonexclusive; compared with kindship, they are voluntary; and compared with work relationships, they are characterized by intimacy and based on common interests (Harris & Vazire, 2016; Hojjat & Moyer, 2017). Friendships are also characterized by equality (Adams et al., 2000; Hartup & Stevens, 2016). In fact, being equal and the absence of power (i.e., the perceived ability to influence others) have been listed as key features of friendship (Laursen & Bukowski, 1997; cf. Veniegas & Peplau, 1997).
However, it seems highly unlikely that friendships are completely absent of power dynamics, as having the power to fulfill one’s needs and goals has been shown to be relevant in a large variety of social relationships (Agnew & Harman, 2019; Keltner et al., 2003). Some studies have even challenged the assumption that friendships are free from processes of influence and control, that is, social power (Stump et al., 2015; Veniegas & Peplau, 1997). In the present research, we began with the premise that power pervades all human relationships, including friendships, and is thereby relevant for understanding how people experience and evaluate their friendships. Building on the rich literature on power in other types of social relationships, we analyzed different aspects of power (actor power, partner power, perceived partner power, and actor desired power) in friendships to provide a comprehensive picture of how power is related to several relationship outcomes. For this reason, we computed Actor-Partner Interdependence Models (APIMs; Kenny et al., 2006) to examine both intrapersonal and interpersonal associations while accounting for nonindependence in friendship dyad data. Demonstrating links between power and multiple aspects of friendship functioning (including satisfaction and quality) and pro-relationship characteristics (including compassion, trust, and forgiveness) would support the relevance of power in friendship contexts.
Key concepts of power in close relationships
Power is typically defined as an individual’s perceived capability to influence others with the goal of attaining core needs and goals (Anderson et al., 2012; Körner, Overbeck et al., 2025; Simpson et al., 2015). Previous research on power in close social relationships has provided insights into the mechanisms through which power shapes the experience and evaluation of these relationships (as described in the following sections). One key feature of power is that it does not exist in a vacuum; instead, as a sociorelational construct, it emerges in social interactions (Keltner et al., 2003). It is therefore important to study power in both interaction partners to better understand the different ways in which one’s own and one’s partner’s power shape relationship outcomes (Overall et al., 2023). However, power has seldom been studied while accounting for its fundamental dyadic nature (i.e., by asking both members of a dyad), making it difficult to separate effects of actor power from those of partner power (Körner, Overall et al., 2025; Körner & Schütz, 2024; Overall et al., 2023). Partner power can be assessed by asking a relationship partner directly (e.g., a friend, romantic partner, or coworker) or by asking the actor to rate their partner’s power. Such perceptions of partners go beyond self-reports and provide unique information (Connelly & Ones, 2010) that can be predictive in relationship contexts (Joel et al., 2020).
A final distinction exists between the state of experiencing power and desired power or the power motive (Murphy et al., 2022). Desiring power does not imply that an individual actually feels powerful (or powerless) but simply indicates that the person aspires to be and enjoys being powerful. The distinction between feeling powerful and desiring power is supported by the differentiation of the two aspects in measurement instruments as well as the opposing correlates and consequences that felt and desired power can have (Kim et al., 2019; Körner & Schütz, 2021; Murphy et al., 2022). In this research, we distinguish between the aforementioned aspects of power, namely, actor power, partner power, perceived partner power, and actor’s desired power.
Power theories (e.g., Guinote, 2017; Keltner et al., 2003; Magee & Smith, 2013) mostly focus on the consequences of actor power, and only a limited number of theories postulate partner effects (e.g., Overall et al., 2023; Simpson et al., 2015). Nonetheless, all these theories assume that power is predictive of relationship outcomes. For example, actor and partner power have been found to be related to the investment model of commitment (Rusbult et al., 1998) by showing relevant links to satisfaction, investment, quality of alternatives, and commitment (Lennon et al., 2013). Power is also relevant to relationships from the perspective of equity theory (Hatfield & Rapson, 2012). People are concerned with the extent to which relationships are rewarding, and, as relationships are more rewarding for high-power actors (because they can fulfill their wishes and goals through the counterpart), this dynamic can explain how power is linked to satisfaction. By contrast, people also strive for fairness; thus, people with low power or a strong desire for power might be less happy because they feel they are not getting enough back.
Previous studies on power in friendships
There is an abundance of research on power in organizational (Anderson & Brion, 2014) and romantic relationship contexts (Agnew & Harman, 2019). For instance, power has been shown to increase the subjective experience of social distance (Magee & Smith, 2013), which could be detrimental to friendships but not necessarily to work relationships. This contrasts with the finding that power and social distance could even facilitate goal pursuit (Anderson & Brion, 2014) and consequently increase the effectiveness of work relationships.
In comparison, power and its outcomes could use further study in friendships despite the fact that, several years ago, researchers called for analyses of the role of power in friendships (Adams & Blieszner, 1994; Rubin et al., 2008). A few studies showed that power is indeed a relevant component in friendship relationships. One study found that when participants were asked to remember a situation when someone else had power over them or they had power over someone else, more than one third of the remembered situations involved friendship relationships (Dunbar & Johnson, 2015). Other studies have found power imbalances in approximately 50%–60% of friends (Davidson & Duberman, 1982; Veniegas & Peplau, 1997), supporting the call to study power in friendships. Except for a study reporting a positive link between an actor’s power and their friendship satisfaction via authenticity (Kifer et al., 2013), most studies have failed to provide clear links between aspects of power and friendship functioning. For example, participants were asked who feels more powerful in a same-sex friendship, and such ratings of relative power were related to friendship quality, showing that perceptions of equal power were positively linked to relationship quality compared with perceiving oneself or the friend as being mostly in charge (Davidson & Duberman, 1982; Veniegas & Peplau, 1997). The use of such relative power measures, however, made it impossible to understand whether high actor power or high partner power was relevant. Nevertheless, Veniegas and Peplau (1997) found that power was more predictive of friendship outcomes than gender differences, further highlighting the central role of power dynamics for friends. Other researchers have studied friendship dyads among adolescents and found that similar levels of power among friends were unrelated to intimacy, perspective-taking, and amount of time together (Updegraff et al., 2004). They asked for an actor’s power but calculated relative-power variables that were based on the responses of both friends (Updegraff et al., 2004), thus contributing to the same aforementioned issue. In another dyadic friendship study, participants were asked to rate their same-sex friend’s influence over the other one (similar to perceived partner power) and found it to be positively related to negative friendship features, such as criticism and conflict. However, only zero-order correlations with aspects of relationship functioning were reported for both friends individually (Stump et al., 2015). Dyadic data analysis techniques (Kenny et al., 2006) that could account for mutual influences between dyads were not employed. Finally, a recent study on adolescents found that perceived power in a specific friend (the partner) was positively linked to depressive and anxiety symptoms in the actor (Schacter et al., 2023), but the dyadic nature of friendships was not considered, as the study assessed only individuals.
To sum up, initial research has shown that power is a topic that is relevant to relationship functioning in friendships, although absolute power measures seem more predictive than relative power measures (Stump et al., 2015; Updegraff et al., 2004). Furthermore, past studies have seldom employed dyadic samples or appropriate analysis techniques and have suffered from methodological shortcomings (e.g., conflating actor and partner power) and theoretical narrowness (e.g., employing only a single power measure). The present research was aimed at overcoming these shortcomings by distinguishing between several aspects of power, using well-validated power measures, and employing dyadic data analysis techniques. We applied a comprehensive approach to assess actors’ and partners’ experiences and evaluations of the relationship, including global relationship evaluations and pro-relationship characteristics regarding how the relationships have been shaped by the various aspects of power. We argue that relationship functioning encompasses both evaluative and behavioral-motivational components, reflecting how individuals perceive the quality of their relationship and how they act and think to maintain or enhance it. Evaluative components, such as satisfaction and relationship quality, provide insight into the relational correlates of power, whereas behavioral-motivational components, such as forgiveness, trust, and compassion, reflect the practical implications of power for relationship maintenance behaviors and attitudes (see Lennon et al., 2013; Wieselquist, 2009).
Power and global relationship evaluations
Relationship quality is considered a central aspect of intimate relationships (Joel et al., 2020). It reflects a multidimensional construct, comprising the evaluation of various domains, such as intimacy or conflict resolution (Fletcher et al., 2000). Relationship satisfaction is a specific construct that refers to how happy or content individuals are with their relationships (Funk & Rogge, 2007). Satisfaction describes an individual’s subjective evaluation of their personal needs, whereas quality describes a more objective evaluation and is thereby more oriented on actual behavior (Cepukiene, 2019).
In friendship contexts, three dimensions of friendship satisfaction have been proposed (Jones, 1991), namely, emotional support, assistance, and companionship. Emotional support reflects the confidence that a friend will provide comfort and will not intentionally harm the relationship. It is based on intimacy and self-disclosure. Assistance describes instrumental support that is based on the balance of giving and receiving to ensure mutual benefits (reciprocity). Companionship captures socializing initiatives, shared activities, and the enjoyment thereof. Friendship quality has been conceptualized using similar but not identical dimensions compared with friendship satisfaction. Bukowski et al. (1994) proposed the five dimensions companionship, help, security, conflict, and closeness. Companionship describes the amount of quality time spent together, help describes the actual support provided by the friend, and security describes the emotional availability and trustworthiness of the friend. Conflict captures the extent of rivalry and disagreement within the dyad, and lastly, closeness describes the emotional connectedness mirrored in actual emotional reactions in emotional situations.
The specific characteristics of power can be expected to shape how people evaluate their relationships. Having the power to fulfill one’s needs and goals should contribute to a person’s satisfaction with their relationship. People who can satisfy their desires and express themselves and approach others with confidence are likely to experience their relationships as more rewarding and fulfilling (Galinsky et al., 2003; Keltner et al., 2003; Kifer et al., 2013). Indeed, actor power is associated with positive emotions and life satisfaction (Keltner et al., 2003; Kifer et al., 2013). In work, romantic, and friendship contexts, actor power was positively linked to relationship satisfaction (Kifer et al., 2013), and in romantic couples, actor power was positively linked to actors’ relationship quality (Körner & Schütz, 2021, 2024). Thus, actors experiencing high power in their friendship are likely to report higher friendship satisfaction and quality.
Power can also be expected to shape partners’ relationship evaluations. Power is a fundamentally dyadic concept that has been theorized to shape not only intrapersonal but also interpersonal outcomes (Simpson et al., 2015). In close relationships, power usually activates social responsibility instead of selfish goals (Chen et al., 2001). Friendships are typically characterized by common interests, communal orientations, and strong interdependence (Laursen & Bukowski, 1997). As such, a partner’s power can be expected to activate responsibility because friends usually care about each other and strive for equal interactions (Hatfield & Rapson, 2012). Perceiving a friend as someone who actively forms the relationship, addresses conflicts, and expresses desires may be appreciated by the actor. Studies on romantic couples have identified positive links between power and partner’s relationship quality (Körner & Schütz, 2021, 2024; see also Murphy et al., 2022). Thus, friends of actors high in power (perceived by friends or self-reported by actors) are likely to experience higher friendship satisfaction and quality.
Finally, a person’s desire for power is just as important in relationship contexts as their experience of power (Murphy et al., 2022). Desiring power over one’s partner may reflect a lack of power, that is, one’s inability to express and pursue one’s goals and needs with one’s partner, but such a desire for power could also reflect antagonistic behaviors aimed at dominating others, driven by competitiveness, a need for superiority, or a disregard for the friend’s autonomy. For example, a friend who desires power might wish to regularly decide what they do or may belittle the other to establish superiority. Desiring power has been linked to lower levels of well-being and life satisfaction and higher levels of negative affect (Emmons, 1991; Veroff, 1982). In romantic contexts, desired power was either unrelated or negatively related to relationship quality (Körner & Schütz, 2021; Traeder & Zeigler-Hill, 2020; Zeigler-Hill & LaCross, 2023). Transferring the findings from general and romantic contexts to friendships, one may expect friends high in desired power to experience lower friendship satisfaction and quality.
Power and pro-relationship characteristics
Pro-relationship characteristics are prosocial behaviors, motivations, and attitudes aimed at maintaining healthy relationships with aspects such as compassion, trust, and forgiveness subsumed under these characteristics. Compassion is an emotional response to the suffering of others, characterized by care and a genuine desire to alleviate the distress (Pommier et al., 2020). Trust is defined as the belief that another individual can be relied upon to keep their word or promises and the belief that other people are benign (Rotter, 1980). Forgiveness is a response that enables a person to replace negative thoughts, feelings, and behaviors toward an offender with prosocial reactions (McCullough et al., 1997). Importantly, there are different types of forgiveness: offense-specific, dyadic (across offenses in a specific relationship), or dispositional (the tendency to forgive multiple others across situations; Worthington, 2019). We assessed offense-specific forgiveness (i.e., forgiving a specific transgression) because this type is fundamentally dyadic and, as it refers to a single incident rather than an enduring relationship pattern, can be considered a consequence of power (Karremans & Smith, 2010; Körner, Heydasch et al., 2022). Compassion, trust, and forgiveness are positively linked to relationship stability and commitment (e.g., Wieselquist, 2009).
Power is a strong predictor of pro-relationship behaviors in work and romantic contexts. Studies have reported primarily negative effects of power on variables such as compassion, sacrificing, or accommodation behavior (Kim et al., 2019; Righetti et al., 2015; Schilke et al., 2015; van Kleef et al., 2008). These findings make sense, as power holders are less dependent on others and do not need to attend to others’ needs (Thibaut & Kelley, 1959). Yet, some of the effects of power on pro-relationship behavior depend on the type of relationship. For example, power has been positively linked to forgiveness and trust in close committed relationships (Karremans & Smith, 2010; Körner, Schütz, et al., 2022; Körner & Schütz, 2021). Power holders’ elevated confidence and reduced tendency to ruminate were found to explain these findings. Overall, an actor’s experienced power can be expected to shape pro-relationship behavior, but both positive and negative effects on compassion, trust, and forgiveness are possible.
Power can also be expected to shape the pro-relationship behavior of partners. Perceiving a relationship partner as powerful may lead individuals to engage in behaviors such as compromising and offering support in order to maintain the relationship (as powerful people are less invested; Sprecher et al., 2006). People might also strategically adopt pro-relationship behaviors to avoid potential negative outcomes (e.g., conflict) with the powerful friend. In fact, partners’ power was positively linked to actors’ other-focused behavior, that is, the tendency to support and promote partners’ needs and goals (Overall et al., 2023). Partners’ power was also positively linked to actors’ forgiveness in romantic relationships (Körner et al., 2022). Thus, friends of high-power actors (perceived by friends or self-reported by the actors themselves) are likely to self-report higher pro-relationship characteristics (i.e., higher compassion, trust, and forgiveness).
Finally, desired power may predict pro-relationship characteristics as well, as desired power involves a strong self-focus (Zeigler-Hill et al., 2024) while ignoring the needs of partners and prioritizing personal goals over mutual interests. Desired power often goes along with dominance and controlling behaviors (Murphy et al., 2022) and is correspondingly positively linked to negative relationship outcomes, such as domestic violence (see Kim et al., 2019). By contrast, pro-relationship characteristics may require collaboration and reciprocity. Thus, high desired power is likely to be linked to lower pro-relationship characteristics (i.e., lower compassion, trust, and forgiveness).
The present study
With the present study, we aimed to investigate whether and how different aspects of power—experienced, perceived, and desired—are associated with various indicators of relationship functioning in friendships. We studied friendship dyads, which allowed us to differentiate between actor effects, partner effects, and perceiver effects and to account for the interdependence in friendships by employing Actor-Partner Interdependence Models (APIMs; Kenny et al., 2006). Specifically, we expected that actor power, partner power, and perceptions of the partner’s power would primarily be positively linked with indicators of relationship functioning and that desired power would be negatively linked with indicators of relationship functioning.
We conducted additional analyses to control for some variables on conceptual and methodological levels. To show that the associations between power and relationship functioning were specific to power, we controlled for both actors’ and partners’ commitment. We chose commitment because it is central to relationship power (Rusbult & Buunk, 1993; Thibaut & Kelley, 1959) and is associated with relationship variables such as satisfaction, forgiveness, and trust (e.g., Lennon et al., 2013; Wieselquist, 2009) in various types of relationships, including friendships (Lin & Rusbult, 1995). We assessed commitment with an established scale and by asking friends to rate the closeness of their friendships. The use of this rating was based on the assumption that participants would understand closeness in terms of commitment (Berscheid et al., 1989), an assumption that was further supported by the high correlations between closeness and commitment in this study and in the literature (Rusbult et al., 1998). On a methodological level, we controlled for social desirability to ensure that the associations between power and relationship functioning were not influenced by participants’ tendency to present themselves in a positive light.
Transparency and openness
The study was not preregistered. Data, syntax, and material are available online (https://osf.io/pzm2c/?view_only=c73026cfa0ca4eaab4becc20bfead174). We report all measures and exclusions.
Method
Participants and procedure
Friends were recruited via email lists and social networking sites across Southern and Western Germany. They were instructed to recruit a friend who could fill out the same survey. Friendship dyads were linked in the data using dyad-specific codes. Overall, 567 participants, mostly students and White, took part in the online survey. Of these, 496 participants formed complete dyads (248 friendship dyads; 70.6% women, 28.6% men, 0.8% nonbinary; M age = 26.23, Mdn age = 22.00, SD age = 10.35, Range: 18–72). Most individuals were heterosexual (79.5%); others were bisexual (10.9%), gay or lesbian (4.0%), pansexual (2.7%), or asexual (1.0%). Participants had been friends for an average of 9 years (Mdn = 7.46, SD = 9.02). For some participants, friendships included additional components (with the option to tick multiple components), namely, kinship (6.5%), work relationship (8.1%), former romantic relationship (1.4%), current romantic relationship (5.7%), former sexual relationship (0.6%), and current sexual relationship (7.5%). Yet, most friends indicated that their friendship did not include any of the aforementioned components (74.4%). Furthermore, friends perceived their relationship as very close on a scale ranging from 1 to 10 (M = 8.93, SD = 1.38).
The survey began with questions about demographics, followed by measures of power and relationship functioning. Survey completion time was approximately 20 min. Each friend responded independently. A dyad code was generated to match friends. We assessed achieved power for typical small-medium effect sizes with this sample size when variables were correlated across partners as they were in the present study (α = .05; mean correlation between errors = .40; mean correlation between actor and partner variables = .28; Ackerman et al., 2020). We had high statistical power (.997) to detect small-medium actor effects of β = .20 and high power (.94) to detect small partner effects of β = .15.
Instruments
Power measures
Experienced power
Descriptive statistics, Cronbach’s alphas, and intraclass correlations across measures.
Note. The values above the diagonal are within-friend correlations. The values below and on the diagonal are between-friend correlations. N = 248 dyads.
*p < .05. **p < .01. ***p < .001 (two-tailed).
Perceived partner power
Analogous to experienced power, the Personal Sense of Power Scale (Anderson et al., 2012) was adapted to assess perceptions of the friendship partner’s power (“In my friendship…”) with six items (e.g., “My friend can get me to listen to what they say,” “Even when my friend tries, they are not able to get their way,” reverse-scored). Reponses were given on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).
Desired power
We assessed participants’ desire for power using the corresponding subscale of the Feeling Power and Desiring Power Scales (Murphy et al., 2022) adapted to match the friendship context. Participants responded to six items (e.g., “I would enjoy having authority over my friend,” “I like to tell my friend what they should do”). Reponses were given on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).
Relationship functioning
Friendship satisfaction
Friendship satisfaction was assessed with three items (Jones, 1991): emotional support (“How satisfied are you with the emotional support you receive from your friend, e.g., feeling cared about, discussing personal problems?”), assistance (“How satisfied are you with the assistance you get from your friend in daily activities, e.g., helping you with chores, giving you information?”), and socializing (“How satisfied are you with the socializing you do with your friend?”). In addition, we administered an item on overall friendship satisfaction, “How satisfied are you overall with your friendship?” A total score was computed based on the four items. Responses were given on a scale ranging from 1 (not at all happy) to 7 (very happy).
Friendship quality
We used the Friendship Qualities Scale (Bukowski et al., 1994) to assess how friends evaluated their overall relationship. To adjust the scale for adults, we dropped the companionship subscale from the original scale and rephrased some items (see the Online Supplement). The final scale comprises four subscales: help (5 items, e.g., “My friend helps me when I have problems with something”), security (5 items, e.g., “If something bothers me, I can tell my friend, even if I can’t tell other people”), emotional connectedness (5 items, e.g., “I feel happy when I’m with my friend”), and conflict (4 items, e.g., “My friend and I don’t agree on many things”). The conflict subscale was reverse-scored to obtain an overall score of friendship quality. Responses were given on a scale ranging from 1 (not at all happy) to 7 (very happy).
Compassion
We used the Compassion Scale (Pommier et al., 2020) to assess experiences of kindness, mindfulness, and sensitivity to the suffering of others. As we adapted the scale to the friendship context, we removed four items focusing on common humanity (see the Online Supplement). The resulting scale had 12 items (e.g., “I empathize with my friend when they are unhappy,” “I don’t think much about my friend’s worries” [reverse-scored]). Responses were given on a scale ranging from 1 (almost never) to 5 (almost always).
Trust
We used the Interpersonal Trust Scale (Buck & Bierhoff, 2012) to assess reliability and trustworthiness vis-à-vis one’s friend. The scale has 10 items (e.g., “I can expect my friend to tell me the truth,” “If my friend promised to do me a favor, they would keep it”). Reponses were given on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).
Forgiveness
We used the Transgression-Related Interpersonal Motivations Inventory (McCullough et al., 1998) to assess forgiveness. We presented a specific scenario to ensure that the responses to the scale were comparable across participants (with X referring to their specific friend): “You are in the park with other friends and X. During the conversation, X suddenly tells everyone about a very confidential secret of yours. In previous conversations with X, you have often emphasized that this secret is very personal and that nobody should know about it. You are very hurt, and you feel a breach of trust toward X. When you approach X about it later, X plays down the situation.”
Then, participants completed items from two subscales: revenge (5 items; e.g., “I’ll make X pay”) and avoidance (7 items, e.g., “I will avoid X”). Responses were summed across subscales and recoded so that higher scores reflected greater forgiveness. Reponses were given on a scale ranging from 1 (strongly disagree) to 5 (strongly agree).
Control variables
Commitment
Participants rated six items from the Investment Model Scale (Rusbult et al., 1998) to assess their commitment. One item from the original scale was removed, as it was romance-specific, whereas the other items could be adapted to the friendship context (e.g., “I want our friendship to last for a very long time,” “I am committed to maintaining our friendship”). Reponses were given on a scale ranging from 1 (strongly disagree) to 7 (strongly agree).
Friendship closeness
A single item, “How close is your friendship?” was used to assess friendship closeness. Reponses were given on a scale ranging from 1 (not very close) to 10 (very close).
Social desirability
Social desirability was assessed with a validated short scale comprising four items (e.g., “I always say what I think,” Lück & Timaeus, 1997). The scale is based on the Marlowe-Crown Social Desirability Scale (Crowne & Marlowe, 1960) and had true/false response options.
Data analysis strategy
First, intraclass correlations (ICCs) were computed using the pairwise correlational method (Kenny et al., 2006). Within-friend correlations are associations between power ratings and relationship functioning ratings made by the same person (above the diagonal, Table 1). Between-friend correlations are associations between the ratings made by the two friendship members (on and below the diagonal).
Then, we computed a series of APIMs for indistinguishable dyads (Olsen & Kenny, 2006) estimated with structural equation modeling and maximum likelihood estimation in Mplus 8. Note that, in line with APIM terminology, we use the term effect, but it does not imply causality. To account for the indistinguishability of members within a specific friendship dyad, we set all corresponding paths, means, intercepts, and (residual) variances equal across partners (Olsen & Kenny, 2006). We modeled both friends’ power scores as simultaneous predictors of their relationship functioning. The APIM assesses actor (e.g., link between Friend A’s power and Friend A’s compassion) and partner effects (e.g., link between Friend A’s power and Friend B’s compassion) to account for the interdependence in measures across friends (Figure 1). When controlling for commitment and social desirability, we simultaneously modeled both friends’ (a) power and (b) commitment or social desirability scores as predictors of both friends’ relationship functioning. We controlled for one variable at a time to isolate the specific effect of each control variable and to be able to clearly interpret the contribution of each variable (e.g., social desirability might obscure a link not only between power and friendship satisfaction but also between commitment and friendship satisfaction). This approach also helped mitigate the risk of multicollinearity between the two highly correlated commitment measures. Because we had five criteria for each power measure, we adjusted the statistical significance of the b coefficients for multiple tests (p < .01, two-tailed). We computed bootstrapped 95% Confidence Intervals (k = 5,000 samples) and effect sizes Δ indexing the change in relationship functioning in SDs when power changed by 1 point (Δ = b/SD; see Körner et al., 2022a). Specification of the APIM Estimating the Effects of Power on Relationship Functioning in Friendships. Note. Continuous arrows = actor effects. Dashed arrows = partner effects.
Results
Table 1 presents descriptive statistics, reliabilities, and correlations. All scales showed acceptable Cronbach’s alpha values, but the perceived partner power scale was slightly below the value of .70, and the social desirability scale had a somewhat low Cronbach’s alpha.
Primary analyses
Results of APIM analyses on the associations between experienced power and relationship functioning in actors and partners.
Note. b = Unstandardized regression coefficient, CI = Bootstrapped 95% confidence interval, Δ = Effect size. N = 248 dyads. The bold values indicate significant b coefficients (p < .01).
Parallel results were found for partners’ power. As expected, greater partners’ power was associated with more positive global friendship evaluations and higher trust and compassion in actors (see right half of Table 2). Effect sizes were medium to large (but approximately half as large as the actor effects). Again, the link between partners’ power and forgiveness in actors was not significant.
Results of APIM analyses on the associations between perceived power and relationship functioning in actors and partners.
Note. b = Unstandardized regression coefficient, CI = Bootstrapped 95% confidence interval, Δ = Effect size. N = 248 dyads. The bold values indicate significant b coefficients (p < .01).
Results of APIM analyses on the associations between desired power and relationship functioning in actors and partners.
Note. b = Unstandardized regression coefficient, CI = Bootstrapped 95% confidence interval, Δ = Effect size. N = 248 dyads. The bold values indicate significant b coefficients (p < .01).
Control analyses
Results of the control analyses that were aimed at clarifying whether the links between power and relationship functioning were robust to potential confounding effects of commitment and positive self-presentation are presented in the Online Supplement. The associations between actors’ power and relationship functioning remained similar when we controlled for commitment, friendship closeness, or social desirability. Also, the associations between partners’ power and friendship quality, trust, compassion, and forgiveness remained similar when we controlled for commitment, friendship closeness, or social desirability. However, with all three robustness checks, the link between partners’ power and friendship satisfaction was no longer significant.
The associations between perceived partner power and relationship functioning remained similar after we controlled for commitment, friendship closeness, or social desirability. Only the link between perceived partner power and forgiveness was no longer significant after we controlled for commitment or friendship closeness.
The associations between desired power and friendship quality, trust, and compassion remained similar after we controlled for commitment, friendship closeness, or social desirability. However, the negative link between desired power and friendship satisfaction remained significant only after we controlled for friendship closeness, but it became nonsignificant after we controlled for commitment or social desirability. The negative link between desired power and forgiveness remained significant after we controlled for friendship closeness or social desirability, but it became nonsignificant after we controlled for commitment.
Overall, 20 associations were tested with 60 robustness checks; 52 out of these 60 tests remained significant (or nonsignificant) as in the primary analyses. Thus, the results of the primary analyses were quite robust, but they also showed that commitment was strongly linked to relationship functioning, reducing the amount of variance that could be explained in power, and that partners’ power was not linked to actors’ friendship satisfaction.
Discussion
This research provides the first systematic analysis of the interplay between social power and various indices of relationship functioning in friendships. We differentiated between actor and partner power, perceptions of one’s partner’s power (i.e., perceived partner power), and the desire for power (power motive), thereby answering past calls for research to separate these aspects, as they reflect properties of different individuals and can have different links to relationship outcomes (Körner, Overall et al., 2025; Körner & Schütz, 2024; Overall et al., 2023).
We found that friends who experienced high power reported higher friendship satisfaction and quality. Power provides the ability to fulfill core needs and goals (Anderson et al., 2012; Overall et al., 2023). Friends who experience power are better able to get their way and are consequently happier (see also the approach/inhibition theory of power; Keltner et al., 2003), as their needs and goals are met. This finding is in line with research on friendships and romantic couples that has demonstrated a positive link between power and friendship satisfaction, relationship satisfaction and quality (Kifer et al., 2013; Körner & Schütz, 2021, 2024). However, our research is correlational and does not allow for conclusions regarding causality. Also, we found that power was positively linked to compassion and trust, which is in line with research on power in romantic relationships (Körner & Schütz, 2021) but opposite to power findings from work contexts (e.g., Schilke et al., 2015; van Kleef et al., 2008). This may be because the often communal nature of friendships leads power to be linked to these pro-relationship characteristics (see Chen et al., 2001) and does not result in power being used and construed for selfish actions (see ten Brinke & Keltner, 2022).
Actors who perceived their friends as powerful or whose friends reported having high power showed better relationship functioning (e.g., higher friendship quality, trust, and compassion). This finding is in line with research on romantic couples showing positive interpersonal associations between power and relationship quality and other-focused behavior (Körner & Schütz, 2021; Overall et al., 2023; see also Körner, Overall, et al., 2025) as well as positive intrapersonal associations between perceived partner power and relationship quality (Murphy et al., 2022). Friendships are typically characterized by common interests and less focus on exchange (Clark & Mills, 2012). In this context, power may be seen as a responsibility (Scholl, 2020), where taking charge serves not only one’s own well-being but also that of the friend. Yet, our findings are in contrast with a friendship study that reported that greater perceived partner power was associated with higher conflict, criticism, and pressure (Stump et al., 2015). The authors used a relative power measure that conflated actor and perceived partner power, which may explain this surprising finding: Participants might have assumed that when they perceived their partner as powerful, they were not in charge, explaining the link with negative relationship evaluations. In our study, the effect sizes for perceived partner power (actor’s ratings of partner’s power) were stronger than those for partner power (partner’s self-ratings). This finding suggests that an actor’s perception of their friend’s characteristics better predict relationship functioning than the friend’s self-perception.
Finally, an actor’s desire for power was particularly negatively related to friendship quality, trust, and compassion. These findings are in line with previous research that has shown that individuals with a high desire for power tend to report lower well-being and life satisfaction (Emmons, 1991; Veroff, 1982). Well-being has been found to be strongly positively linked to relationship quality (Love & Holder, 2016; Rowsell & Coplan, 2013). The findings also accord with studies reporting negative associations between desired power and relationship quality among individuals in romantic relationships (Traeder & Zeigler-Hill, 2020; Zeigler-Hill & LaCross, 2023; see also Kim et al., 2019). Whereas most past research has focused on romantic relationships, the present study is the first to show a negative link between desired power and relationship functioning among friends. It seems that those who seek power over their friends may do so at the expense of positive attitudes and behaviors, thereby engaging in hurtful actions (e.g., disregarding others’ suffering) and fostering negative reactions (e.g., mistrust). Additionally, from the perspective of communal versus exchange orientations (Clark & Mills, 2012), individuals with a high desire for power are more likely to approach their relationships with an exchange orientation (Locke & Heller, 2017), where they view interactions in terms of personal gain rather than mutual benefit. This mindset may erode trust, compassion, and the quality of friendships. For example, being compassionate toward a friend might not feel like something that will pay dividends or help one achieve superiority. From an equity perspective (Hatfield & Rapson, 2012), individuals with a high power motive may be unhappy with their friendship because it does not feel as rewarding as they would like (e.g., their friend is working against their attempts to keep the friendship in line with communal norms, and the power motivated person may thereby feel that they are not getting what they deserve, i.e., superiority and influence).
Because friendships can be formed more easily and in larger numbers than romantic relationships (exclusiveness) and work relationships (limited options), power motivated individuals may be less inclined to restrain themselves in their attempts to achieve a higher power position in their friendships compared with their romantic or work relationships. However, friendships are also resolved more easily than romantic (less emotional investment) or work relationships (higher avoidability of future interactions). Therefore, the friends of power motivated individuals may also be less inclined to endure their friend’s execution of power, which increases social distance and voids the premise of equality (Adams et al., 2000; Hartup & Stevens, 2016; Magee & Smith, 2013). Thus, reciprocity is more important for maintaining a fulfilling friendship compared with other less flexible forms of social relationships. Desired and experienced power in friendship dyads are complex but also pivotal for understanding the functioning and dynamics of friendship relationships.
Theoretical advances and implications
The present research provides evidence that power in friendships should be more fully integrated into existing power theories. Power theories usually focus on general contexts or specifically on work or romantic relationships (Guinote, 2017; Keltner et al., 2003; Magee & Smith, 2013; Simpson et al., 2015; Thibaut & Kelley, 1959). For example, neither the approach/inhibition theory of power (Keltner et al., 2003) nor Guinote’s (2017) framework of power as activating, wanting, and goal seeking use the term “friend” in their reviews. More concerning, friendships are usually close relationships, but studies on power in close relationships have tended to focus primarily on romantic couples, ignoring friendship relationships (Kim et al., 2019; Simpson et al., 2015). Our analyses show that power is associated with a host of relationship functioning variables, underscoring the predictive value of power for friendship evaluations and pro-relationship characteristics. Developing and testing theories on how power operates among friends, including proposing antecedents and consequences that may differ from other social relationships will have the potential to guide future research on the role of power in close, voluntary, nonromantic relationships. Similarly, we expect that proposing or expanding theories on how power operates in other understudied relationships (e.g., teacher-student; parent-child) may also enrich the understanding of this core relationship variable.
The findings also highlight that power is related not only to an actor’s relationship functioning but also to the friendship partner’s relationship functioning. As power is a sociorelational construct (Anderson et al., 2012; Körner, Overbeck et al., 2025) that emerges within the social context of two people, it should shape the feelings and evaluations of both interaction partners. The scarce research on power in friendships has not provided a sufficient methodological account of the interpersonal relevance of power, nor has it distinguished between actor and partner power (e.g., Updegraff et al., 2004; Veniegas & Peplau, 1997). Enhancing the current knowledge base, we found several links between partner power and relationship functioning. As such, this research provides one of the first tests showing that and how power shapes the relationship evaluations of both friends in a dyadic relationship. This research adds to the slowly accumulating reports that partner power is also relevant for understanding an actor’s experiences and behaviors in intimate social relationships as has been shown for romantic relationships (Körner & Schütz, 2021; Overall et al., 2023). Considering that friendships are not as close or as interdependent as romantic relationships, one may expect to find effects of partner power in other contexts beyond romantic contexts as well (e.g., school or work contexts).
The present findings also reveal that power research that was based on hierarchical (e.g., work) or romantic contexts does not automatically generalize to friendships. Indeed, we used research on romantic relationships to guide our hypotheses on power in friendships because both romantic partners and friends are intimate individuals. However, we found that neither actor nor partner power in friendships was related to forgiveness, even though power in general and in romantic contexts has been reported to be strongly positively linked to forgiveness (Karremans & Smith, 2010; Körner et al., 2022). Among unacquainted undergraduates, power was negatively linked to compassion (van Kleef et al., 2008). By contrast, we found that power was positively linked to compassion in friendship dyads. Friendships seem to provide a unique type of relationship in which processes of influence and control work differently than in other relationships. Therefore, theories of power should more fully account for different correlates and consequences of power depending on the type of relationship.
Finally, the present findings demonstrate that experienced and desired power reflect two distinct constructs in the friendship context. Some measurement tools and theoretical analyses do not clearly distinguish between these two aspects (for overviews see Körner, Overbeck et al., 2025; Murphy et al., 2022). This lack of differentiation is problematic, as people who desire power show different relationship behaviors than people who feel powerful (regarding romantic relationships, see Kim et al., 2019). We found that greater experienced power was associated with higher ratings of pro-relationship characteristics and more positive global friendship evaluations—for both friendship members (i.e., actor and partner). This contrasts with the finding that greater desired power was negatively associated with relationship functioning, but only for the actor. Thus, not only are experienced and desired power primarily correlated in opposite directions for actors, but it is also only experienced power that matters for partners’ friendship evaluations. This suggests that desired power is less determined by social context and more by individual characteristics that shape an actor’s thoughts and feelings.
Limitations and future research directions
Despite identifying both intra- and interpersonal links between power and relationship functioning, the current study relied on cross-sectional data, which limits causal conclusions. Both directions of effects between power and relationship functioning are possible. As theories of power in romantic relationships (e.g., Overall et al., 2023; Simpson et al., 2015) or general contexts (Keltner et al., 2003; Magee & Smith, 2013) conceive of power as an antecedent of concepts such as satisfaction or forgiveness, we modeled power as the predictor of relationship functioning. Nevertheless, future experimental and longitudinal research is needed to tackle the question of causal influences.
Our analyses are the first to distinguish between various aspects of power in friendships (e.g., actor power, partner power). To do so, we analyzed dyadic relationships and accounted for the nonindependence of friends by computing APIMs. Yet, people typically have more than one friend (Altmann, 2020). Although our research can speak to the interplay between power and relationship functioning in specific friendship dyads, it may be interesting to examine whether people also possess more abstract evaluations of their experienced power across friendships and how such perceptions are related to more general friendship evaluations. For example, to what degree is experienced power similar across friendships versus friendship-specific? Can power perceptions based on various friendships predict general but also relationship-specific levels of satisfaction? Examining power across various friendships can advance the understanding of how power operates not only in dyadic settings but also in multiperson contexts.
Further, we instructed participants to recruit a friend without defining the term “friend.” Previous research has demonstrated substantial variability in friendship definitions across cultures, groups, individuals within groups, and even within individuals over time (Baumgarte, 2016; Merkin & Gareis, 2021; Tesch, 1983). By not providing a definition, we allowed for this variability to be reflected in our sample, avoiding restrictions that might otherwise distort the findings. This approach aligns with a current large-scale study on friendship, in which participants were also allowed to determine for themselves whom they consider a friend, without having a friendship definition imposed on them (Pennington et al., 2024).
However, this approach may have also led to the inclusion of a broader variety of dyads compared with studies that provided specific friendship definitions, which often exclude certain types of relationships (e.g., online-only friendships, relationships with relatives). For example, our sample included a small number of participants who nominated a relative, a former or current romantic partner, or someone with whom they had a sexual relationship as a friend. While these cases were not common, they raise important considerations for interpreting our findings in light of different relational contexts. Some of our findings replicated results from power research conducted in romantic and general contexts, which may be due to the small proportion of participants who selected a romantic partner or family member as their friend in our study. Still, we expect findings between power on friendships and sexual, romantic, or familial relationships to differ slightly because friendships show less intimacy than these relationships. For example, we found that the links between power, compassion, and forgiveness in our study differed from those reported in studies focusing on romantic relationships or general relational contexts. Future research could systematically compare correlates of power across different types of close relationships, using fixed and exclusive definitions for categories such as friend, romantic partner, or relative.
Other limitations include the low reliability of the perceived partner power scale and the social desirability scale. Replication studies may benefit from using more established and longer social desirability scales with better psychometric properties (see Larson, 2019; Tan et al., 2022). On another note, the forgiveness measure might also be criticized for presenting a hypothetical situation that some participants may relate to, but others might not. Upcoming studies should use other forgiveness measures for assessing dispositional and dyadic forgiveness, as well as offense-specific forgiveness concerning real experienced situations (McElroy-Heltzel et al., 2019), to examine the replicability and generalizability of our null finding.
Finally, future studies could expand on the generalizability of the findings by drawing samples from other cultural backgrounds. Friendships are differently valued and fulfill different functions in different cultures (Altmann, 2021). In specific cultures (e.g., collectivistic countries), the links between power and friendship evaluations might differ from the individualistic culture the present sample was drawn from (Baumgarte, 2016). Future studies should also collect more detailed data on sociodemographic variables (e.g., percentage of students, socioeconomic status, disability status) and might benefit from distinguishing between different types of friendships (long-distance vs. short-distance; adolescent vs. elderly; same-sex vs. different-sex) to test how robust the links we reported are.
Conclusion
Power pervades all human relationships. We illustrated the importance of power in friendships by showing that greater experienced power (by both friendship members) as well as greater perceived partner power are associated with better relationship functioning. By contrast, greater desired power is associated with low levels of relationship functioning. These findings advance existing power theories and research by demonstrating the dyadic nature of power and highlighting that friendships provide a unique context in which power matters. The findings also support the distinction between experienced and desired power and highlight that findings of power from general, work, or romantic contexts might not fully generalize to friendships. We hope our research will stimulate new avenues of inquiry linking two fundamental relational topics: power and friendship.
Supplemental Material
Supplemental Material - Power in friendships: How experienced and desired power are associated with relationship functioning
Supplemental Material for Power in friendships: How experienced and desired power are associated with relationship functioning by Robert Körner and Tobias Altmann in Journal of Social and Personal Relationships.
Footnotes
Author contribution
Conception and design of the work: RK. Data collection: RK, TA. Data analysis: RK. Drafting the manuscript: RK. Critical revision of the manuscript: RK, TA. Funding (or proofreading): TA.
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Open research statement
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
