Conflicting relational models as a predictor of (in)justice perceptions and (un)cooperative behavior at work
Funding information
The research was supported by the German Research Foundation (SFB 768, TP A8)
Abstract
Humans are naturally social, and according to relational models theory (RMT), they use cognitively represented and motivationally operative models (i.e., relational models) to structure and understand their social interactions. RMT proposes that the fit between the expected and perceived relational model (i.e., RM fit) in a given social interactive situation is related to perceptions of justice, while an RM misfit is related to injustice perceptions. The experience of RM fit/misfit is motivationally operative for generating behavior intended to either strengthen a just relationship or transform an unjust relationship. Building on these theoretical considerations, it is argued that RM fit (misfit) is positively (negatively) related to perceived justice which in turn is positively related to willingness to help and negatively related to willingness to hide one's knowledge from an interaction partner. Willingness to help and sharing information are of particular practical importance in the context of teamwork and for cooperative relationships in organizations more generally. Three experimental studies (n1 = 441, n2 = 618, n3 = 455) were conducted in which RM fit/misfit was manipulated as an independent variable in three different work scenarios (vignettes). We assessed participants' justice perceptions and willingness to exhibit (un)cooperative behavior (i.e., more or less helping and knowledge hiding) toward their interaction partners. All three experiments confirmed the hypothesized relationships. The results are discussed with respect to the theoretical relevance of RMT for explaining mechanisms underlying justice perceptions, helping behavior, and knowledge hiding at work in teams and organizations.
1 INTRODUCTION
In today's dynamic and knowledge-intensive work environments, organizational success has become increasingly dependent on employees' willingness to exhibit cooperative behavior toward their co-workers and to refrain from uncooperative actions (Bridoux & Stoelhorst, 2016; Mossholder et al., 2011). One form of cooperation that has been intensively studied in organizational settings is helping behavior among co-workers (Podsakoff et al., 2009, 2014). Scholars have also recently begun to explore employee behaviors that are explicitly uncooperative in nature, such as intentionally withholding knowledge (i.e., knowledge hiding, Connelly et al., 2012).
Perceived justice is a construct that has been repeatedly linked to both helping behavior (Barclay & Kiefer, 2014) and knowledge hiding (Connelly et al., 2012; Huo et al., 2016; Tsay et al., 2014). Often, a wide range of different forms of behavior can be perceived as fair from one perspective and unfair from another, leaving room for substantial (mis)alignment in interaction partners' expectations about appropriate behavior in social interactive situations. While a large body of theoretical and empirical research has shed light on the consequences of perceived (in)justice (for an overview, see e.g., Cropanzano & Ambrose, 2015), much less is known about the antecedent factors and processes which shape a person's perception of what is fair1 and unfair in a given social interactive situation and how (mis)alignment among these antecedents relates to (un)cooperative behaviors such as helping behavior and knowledge hiding.
Relational models theory (RMT, Fiske, 1992) offers explanations for how antecedent conditions to justice perceptions, including social cognitive and motivational variables and processes, might be further related to behavioral consequences like helping and knowledge hiding. The theory posits the existence of four universal and distinct mental schemata (i.e., relational models) which people use to interpret, understand, and regulate their social interactions and make any necessary behavioral adjustments. The four relational models include distinct moral motives (Rai & Fiske, 2011) encompassing distinct principles of fairness (Fiske, 1992; Rai & Fiske, 2011).
According to RMT, in cases of misalignment, that is, when interaction partners apply different relational models in a given social interactive situation, they are guided by different moral motives and fairness principles and have different expectations of what behavior is appropriate or fair. In this way, RMT not only offers theoretical explanations for the consequences of perceived injustice, but also, the origins of varying expectations in what behavior is “fair” in a given social interaction. As such, it provides a promising framework for the examination of (in)justice perceptions. However, although RMT has stimulated a growing body of theoretical research that applies the theory to various aspects of social interaction (e.g., Bridoux & Stoelhorst, 2016; Mossholder et al., 2011; Vodosek, 2000), empirical work examining some of its central predictions remain limited and extant studies which have explored them, failed to support the hypothesized effects (Poulson, 2005). Therein lies our study's purpose. Drawing upon RMT, we propose that when interaction partners apply different relational models in the same social interactive situation, a misfit occurs between the expected and perceived relational models (i.e., RM misfit), resulting in perceptions of injustice. Perceptions of injustice, in turn, are likely to affect one's willingness to exhibit (un)cooperative behavior toward an interaction partner.
Our research aims to make two primary contributions to theory and research. First, we contribute to further develop RMT by specifying and substantiating central—but thus far rather general and unspecific—predictions of RMT. Namely, we explore and operationalize the proposition that applying different relational models (among two or more social interaction partners) leads to perceptions of injustice. In more detail, we hypothesize and show that the exact same behavior is perceived as just or unjust depending on the situation-specific salient “standards” for relationship regulation and respective expectations thus tied to the situation at hand by an individual. Furthermore, we specify and explore behavioral reactions, such as reducing helping behaviors and knowledge hiding behaviors as consequences of such injustice perceptions.
Second, by testing the aforementioned predictions of RMT in the context of co-worker interactions at work, we connect the general predictions of RMT to theory and research in the area of organizational psychology. We theoretically outline and empirically explore social cognitive and motivational causes of (in)justice perceptions and exemplary behavioral consequences of great relevance for the applied context of organizational psychology (i.e., helping behavior or knowledge hiding behaviors). While exploring consequences of perceived (in)justice has received a lot of attention by theory and research in organizational psychology, a theoretical understanding of antecedents of (in)justice perceptions in specific social interaction are largely lacking. By using RMT to explain injustice perceptions, we move beyond theorizing that emphasizes injustice perceptions as a factor with many important consequences, but forces on conditions that foster injustice. Of particular interest for theory in organizational psychology should be, that the causes of injustice that we propose are not per se unjust or unmoral but rather perceived unjust because they contradict individual expectations and standards of relationship regulation within a specific context.
2 RELATIONAL MODELS THEORY
Relational models theory (RMT, Fiske, 1992) posits the existence of four distinct, fundamental mental representations of social relationships (i.e., relational models) that people use to structure and regulate their social interactions. People (often unconsciously) use these relational models “to plan and construct action; to anticipate and interpret others' actions; to encode, process, and remember social experience; to evaluate and sanction their own and others' action” (Fiske & Haslam, 2005, p. 271). Relational models enable people to instantly appraise how they see themselves in relation to others and provide specific information about what behavior is (not) appropriate and (not) acceptable in a given situation. Specifically, each of the relational models contains a specific and distinct underlying moral motive (Rai & Fiske, 2011), which serves as the basis for perceptions of (un)fairness as well as moral outrage. The four relational models are communal sharing (CS), authority ranking (AR), equality matching (EM), and market pricing (MP).
When people apply a CS model to an interaction, they see themselves and their interaction partner(s) as sharing a common identity and are guided by feelings of belonging and solidarity. When adopting a CS model, people distribute resources in accordance with the principle of need (i.e., everyone gives what he or she can and receives what he or she needs); when making decisions, they try to reach consensus. In a CS model, people do not keep track of individuals' inputs and outputs; keeping an account of exchanges between interaction partners is perceived as inappropriate and morally reprehensible.
When people apply an AR model to an interaction, they situate each other in a hierarchical order along a certain dimension, such as formal rank, seniority, or expertise. When adopting an AR model, people distribute resources such that each person's share conforms to his/her rank; it is socially accepted that higher-ranking people will receive a larger share than lower-ranking people. When making decisions, it is perceived as appropriate for higher-ranking people to make decisions for the whole group, but they are also expected to bear sole responsibility for these decisions. In an AR model, higher-ranking persons decide on the appropriate level of each individual's contribution; it is sometimes considered fair for higher-ranking persons to contribute less and sometimes considered fair for them to contribute more (in the sense of noblesse oblige) than lower-ranking persons.
When people apply an EM model to an interaction, they perceive themselves and their interaction partner(s) as equal but distinct individuals with exactly the same rights and duties. When adopting an EM model, people distribute resources such that each group member receives exactly the same share; when making a decision, each individual's voice has exactly the same weight. In an EM model, people are expected to keep track of imbalances in support or favors and do their part to balance them out in a similar way.
When people apply an MP model to an interaction, they are guided by rational calculations of their individual inputs and outputs with respect to the relationship. When adopting an MP model, people distribute resources in accordance with the individual contribution of each group member; people who have invested more expect to receive a larger share. Decisions are made on the basis of individuals' inputs and outputs as well as rational cost-benefit calculations of their consequences. In an MP model, it is accepted and even expected for individuals to keep track of each group member's inputs and outputs and to return favors and support in an appropriate (but not necessarily in the same) manner.
These four relational models can be described as the fundamental cognitive building blocks or grammar of the social interactions that make up social relationships. They constitute the cognitive structures through which people interpret, evaluate, and sanction the behavior of interaction partners.
While it is possible for people to use different relational models in different domains of a social relationship, research has revealed a tendency for individuals to use the same relational model across multiple domains within a given social relationship, both in dyadic relationships (Haslam & Fiske, 1999) and in groups (Vodosek, 2009). Accordingly, if interaction partners have applied a certain relational model to various domains of their relationship, they should also exhibit a tendency to apply this relational model to other social domains—and to expect their interaction partners to apply this relational model, too. For example, if two (or more) interaction partners have previously applied a CS model when allocating resources and making decisions, they are likely to also apply a CS model when helping and backing up each other.
3 RELATIONAL MODELS AND JUSTICE PERCEPTION
Each of the relational models contains a distinct moral motive and thus different principles and expectations about what is appropriate in different domains of social interaction and in relationships with different people. Thus, the perception of what is right and wrong, of what is fair and unfair is not stable across relationships and situations, but rather highly dependent on the relational model an individual considers valid in a given social interaction (Simpson & Laham, 2015). From the perspective of RMT, the relational rules and norms which people perceive in a social relationship stem from their individual perceptions and expectations of which relational model should be implemented (how) in a given domain within a given relationship.
Behavior that is considered highly appropriate from the perspective of one relational model is often considered highly inappropriate from the perspective of another relational model. For example, imagine that a bonus payment must be distributed among the members of a team. Different distribution mechanisms can appear fair depending on the relational model applied: According to the EM model, everyone should get an equal share; according to the CS model, the person who needs the bonus most should get the most; according to the AR model, the person who took on managerial functions should get the most; and according to the MP model, the person who invested the most should get the most.
Because the principles of justice inherent to the four relational models are usually incommensurable with one another, the adoption of different relational models by interaction partners in a given situation is likely to cause perceptions of injustice (Fiske, 1992; Poulson, 2005). This assumption received empirical support from a study conducted by Arendt et al. (2019). They explored the effects of (the extent of) team members' shared understanding of the relational models in their team on various aspects of team functioning. Specifically, they found a positive relationship between the degree of sharedness of relational models in teams and team members' justice perceptions. The higher the degree to which team members perceived same relational models to be “valid” in their teams, the greater justice they perceived. An explanation for this finding is that a lower degree of sharedness of relational models in a team means that team members are likely to apply different, conflicting relational models in social interactive situations. Building upon this explanation, we propose that conflicting relational models in a given social interactive situation negatively affect the interaction partners' justice perceptions.
We propose that if a given relationship between interaction partners tends to be predominated by a specific relational model and one interaction partner acts in a way that contradicts this relational model, while the other interaction partner expects to perceive RM-congruent behavior (hereafter RM misfit), the event will be perceived as less fair than an even in which the second interaction partner perceives behavior from the interaction partner congruent with his/her expectations (hereafter RM fit).
Thus, we predict the following:
Hypothesis 1.An RM fit leads to higher perceptions of justice in a social interactive situation compared to an RM misfit.
4 JUSTICE PERCEPTIONS AND HELPING BEHAVIOR
In Hypothesis 1, we propose RM fit/misfit to be an antecedent of perceived justice. However, justice perceptions have also been identified as an antecedent of various aspects of social behavior in organizations (Cropanzano & Ambrose, 2015). In particular, justice perceptions have been repeatedly linked to cooperative behavior at work (Ambrose et al., 2015). One type of cooperative behavior of particular interest to the study of social interactions at work is helping behavior toward co-workers, usually described as individual-oriented organizational citizenship behaviors (“OCBI”). OCBI have been defined as behaviors which “immediately benefit specific individuals and indirectly through this means contribute to the organization” (Williams & Anderson, 1991, p. 602). In recent decades, helping behavior has been intensively studied in different types organizations, and a growing body of empirical studies has demonstrated its positive effects on various aspects of organizational behavior and performance (for an overview, see Colquitt et al., 2013; Podsakoff et al., 2014).
Helping behavior—whether at work or in other areas of life—is usually associated with some form of effort on the part of the helper, who provides the person helped with some resource (i.e., time, labor, expertise, knowledge). This exchange of resources occurs with the expectation that the other party (i.e., the interaction partner) will adhere to the rules of the respective relationship, or, in other words, that he/she will behave fairly. When interaction partners are perceived as behaving unfairly, one's willingness to exchange resources with them is likely to decrease. This assumption is supported by a large number of studies reporting positive relationships between justice perceptions and various forms of cooperative behavior (e.g., Arendt et al., 2019; Colquitt et al., 2013; Naumann & Bennett, 2002). Therefore, we predict the following:
Hypothesis 2.The more justice is perceived in a social interactive situation, the higher the willingness to exhibit helping behavior toward the interaction partner(s).
Taken together with Hypothesis 1, this leads to the assumption that an RM fit in a given social interaction, as compared to an RM misfit (hereafter: RM fit/misfit), is indirectly related to a greater willingness to exhibit helping behavior toward the interaction partner via justice perceptions as the mediating variable.
Hypothesis 3.In a social interactive situation, perceived justice mediates an indirect relationship between RM fit/misfit and the willingness to exhibit helping behavior toward the interaction partner(s).
5 JUSTICE PERCEPTIONS AND KNOWLEDGE HIDING
A specific form of cooperative behavior that is attracting increasing interest from both scientists and practitioners, and which has been repeatedly linked to various forms of organizational performance, is the exchange of knowledge among co-workers. While knowledge sharing has received a great deal of research attention in recent decades (for an overview, see Mesmer-Magnus & DeChurch, 2009; Wang & Noe, 2010; Witherspoon et al., 2013), only in recent years have scholars also begun to explore its counterpart, namely knowledge hiding (Connelly et al., 2012). Knowledge hiding has been defined as “an intentional attempt by an individual to withhold or conceal knowledge that has been requested by another person” (Connelly et al., 2012, p. 65). We assume justice perceptions in a social interaction will affect the willingness to engage in knowledge hiding behavior in two ways:
First, as described above for helping behavior, persons who perceive an interaction as unfair may no longer be willing to invest resources (i.e., knowledge) in the relationship with their interaction partner because they cannot be sure that this person will not fail to meet their expectations and break the “relational rules” again.
Second, a person who perceives unfairness in a social interaction may feel the impulse to punish the interaction partner who caused this alleged unfairness by breaking the relational rules of the relationship (Fiske, 1991). From this point of view, knowledge hiding can be understood not only as a refusal to invest but also as a form of punishing the person whose behavior is perceived as unfair. RMT posits that people strongly believe that they and their interaction partner(s) should respect the rules of the relational model they have applied to the social interaction (Fiske, 1992) and that they usually have a strong desire to punish the violation of this relational model (Fiske, 1991). Intentionally withholding knowledge their interaction partner needs can be one such form of punishment. Thus, we expect that perceived unfairness in a social interaction is related to a higher willingness to engage in knowledge hiding behavior toward the interaction partner(s). The assumed relationship between justice perceptions and the willingness to engage in knowledge hiding behavior also received empirical support in Arendt et al. (2019), who linked justice perceptions to knowledge hiding among team members.
Thus, we predict the following:
Hypothesis 4.The less justice is perceived in a social interactive situation, the higher the willingness to engage in knowledge hiding behavior toward the interaction partner(s).
Taken together with Hypothesis 1, this leads to the assumption that an RM fit in a given social interaction, as compared to an RM misfit, (RM fit/misfit) is indirectly related to a greater willingness to engage in knowledge hiding behavior toward the interaction partner. Thus, the following prediction is made:
Hypothesis 5.In a social interactive situation, perceived justice mediates an indirect relationship between RM fit/misfit and the willingness to engage in knowledge hiding behavior toward the interaction partner(s).
6 STUDY OVERVIEW
We conducted three experimental vignette studies2 to test our hypotheses in different scenarios. In Study 1 and Study 2, we tested the proposed effects of RM fit/misfit on participants' justice perceptions (H1) and willingness to engage in helping behavior (H2, H3) in a team setting (Study 1) and a dyadic setting (Study 2). In Study 3, we tested the same propositions (H1–H3) in a dyadic setting, while further testing the proposed effects on participants' willingness to engage in knowledge hiding behavior (H4, H5).
7 STUDY 1
7.1 Method
7.1.1 Sample
Participants were recruited from social networks as well as through a German university's student and graduate mailing lists. As an incentive, participants had the option to take part in a raffle for 150€, and psychology students could receive course credit for their participation. Overall, 502 people participated in the study. Sixty-one participants were excluded from the sample due to a high number of missing values (more than 5%) or not fulfilling the age requirements (a minimum age of 18 years old).
Our final sample consisted of 441 individuals (335 female, 106 male) with an average age of 25.26 years (SD = 7.31) and ranging from 18 to 64 years old. The majority of our participants were German (91.4%) and were university students (72.6%).
7.1.2 Design and procedures
We applied an experimental vignette methodology using an online questionnaire. The experimental vignette methodology allows for the investigation of causal relationships and combines the internal validity of an experimental design with the external validity of field research (Aguinis & Bradley, 2014; Atzmüller & Steiner, 2010). The experiment had a 4x4 between-subject design and participants were randomly assigned to one of the 16 combinations of vignettes.
First, the participants were presented with one of four vignettes that included a short description of a fictive working team. Participants were asked to imagine that they worked in this team and that their fellow team members usually behave exactly as described. In each of the four vignettes, the team was described in a manner congruent with one of the four relational models (see Appendix A). The aim of this description was to manipulate the participants' expectations with respect to the relational model that guides social interactions within the described team.
Second, the participants read that one of four different events took place in the team. Specifically, participants learned that a bonus had been paid out and needed to be allocated among the members of the team. In each of the four events, the bonus payment was allocated in accordance with the justice principles inherent to one of the four relational models (see Appendix A). To avoid outcome favorability effects (Skitka, 2002), the participants were only told how the bonus was allocated and not whether or to what degree they themselves would profit from this allocation. Thus, the participants knew which distributional rule was implemented, but not the particular role they would take in the distributional outcome.
After having read the description of the team (i.e., one of the four conditions) and the event (i.e., one of the four conditions), participants were asked to rate their justice perceptions with respect to the event (i.e., to what degree was the bonus allocated in a fair manner?) as well as their willingness to help other team members in the future.
At the very end, participants were asked directly how appropriate they considered the distribution system for their specific team (when answering this question, participants could read both the team and event descriptions once more). This question (hereafter perceived degree of fit) served as a manipulation check.
7.1.3 Stimulus material and measures
Team description
The four vignettes including team descriptions were formulated based on the relational models scale by Haslam and Fiske (1999) as well as Vodosek's (2009) adaptation of it to the work team context. Each vignette was formulated in accordance with one relational model and addressed the following domains of teamwork: the distribution of resources, decision making, the allocation of tasks and responsibilities, and the general nature of social relationships in the team (see Appendix A). To ensure that each vignette unequivocally described just one of the four relational models (i.e., CS, AR, EM, MP), the vignettes were independently rated by four experts who were familiar with RMT. All raters correctly identified the intended relational model for each vignette.
Event vignettes
Due to favorable developments, your team will now be provided with an impressive bonus that can be distributed within the team. The team decides that…
The second sentence ended with a short description of how the team had decided to allocate the bonus. In each condition, the bonus was allocated according to the justice principle underlying one of the four relational models. Thus, the bonus was allocated according to the principle of need in the CS condition, the principle of hierarchy in the AR condition, the principle of equality in the EM condition, and the principle of proportionality in the MP condition (see Appendix A).
Perceived justice
Participants' justice perceptions were measured with seven items adapted from Ambrose and Schminke (2009). A sample item is “I would feel treated fairly.” Cronbach's alpha was α = .93.
Helping behavior
Participants' willingness to engage in helping behavior toward their fellow team members was measured with eight items from the OCBI subscale from Lee and Allen (2002). The scale was translated into German by individuals fluent in both German and English. A sample item is “I would willingly give my time to help others who have work-related problems.” Cronbach's alpha was α = .92.
Perceived degree of fit (manipulation check)
Perceived degree of fit between the two presented vignettes, which served as a manipulation check, was assessed with the item “How suitable did you find the described distribution system for the described team?”
7.2 Results
7.2.1 Manipulation check
As a manipulation check, we conducted an independent t test for the perceived degree of fit item. We assumed that the team's decision in the event condition should be perceived as more appropriate when it applied the same relational model as in the team description as opposed to a different relational model. Levene's test indicated unequal variances (F = 36.54, p < .01); thus, the degrees of freedom were adjusted from 439 to 158. The distribution decision was perceived as more appropriate in the fit condition (n = 82, M = 3.85; SD = 1.11) in the misfit condition (n = 359, M = 2.67; SD = 1.52), t(158) = 8.10, p < .001, g = 0.81 (95% CI [0.56, 1.06]). This indicates that our RM fit/misfit manipulation was successful.
7.2.2 Effects of demographic variables
Student and non-student participants did not differ with regard to perceived justice (t(439) = .286, p = .775, g = 0.031, 95% CI [−0.18, 0.24]) and helping behavior (t(439) = .868, p = .386, g = 0.031, 95% CI [−0.12, 0.30]). Similarly there was no significant effect of gender, neither on perceived justice (t(439) = .474, p = .636, g = 0.053, 95% CI [−0.27, 0.17]), nor on helping behavior (t(439) = .152, p = .879, g = 0.017, 95% CI [−0.24, 0.20]). The age of the participants also had no effect on the perceived justice (r = −0.07, p = 0.165, 95% CI [−0.16, 0.03]) and helping behavior (r = −0.04, p = 0.40, 95% CI [−0.13, 0.05]).
7.2.3 Hypothesis testing
Correlations, means, and standard deviations of all variables are shown in Table 1.
| Variable | M | SD | 1 | 2 | 3 | 4 | |
|---|---|---|---|---|---|---|---|
| 1 | RM fit/misfit | .19 | .39 | – | |||
| 2 | Justice | 3.27 | .91 | .27** | (.93) | ||
| 3 | Helping | 3.12 | 1.22 | .08 | .40*** | (.92) | |
| 4 | Perceived degree of fit | 2.89 | 1.52 | .30*** | .71*** | .35*** | – |
Note
- Reliabilities (Cronbach's alpha) are indicated on the diagonal in parentheses.
- ** p < .01;
- *** p < .001.
Hypothesis 1 proposed that an RM fit in a social interactive situation leads to a higher perception of justice than an RM misfit. To test Hypothesis 1, we conducted an independent t test. Levene's test indicated unequal variances (F = 18.04, p < .01); thus, the degrees of freedom were adjusted from 439 to 157. Supporting Hypothesis 1, participants in RM fit conditions reported higher perceptions of justice (M = 3.80; SD = .91) than participants in RM misfit conditions (M = 2.96; SD = 1.23), t(157) = 6.99, p < 0.001, g = 0.72 (95% CI [0.47, 0.96]). Participants presented with an event vignette employing the same relational model as the previous description of the team reported higher perceptions of justice than participants presented with an event vignette employing a different relational model than in the previous description of the team.
Supporting 2, participants' perception of justice was positively related to participants' willingness to engage in helping behavior toward other team members (r = .40, p < .001, 95% CI [0.31, 0.48]). The higher the perception of justice, the higher the participants' willingness to help other team members in the future.
Hypothesis 3 proposed an indirect effect of RM fit/misfit on helping behavior via perceived justice. To test this indirect effect, we created a dummy variable (see Iacobucci, 2012) coded 1 for RM fit and 0 for RM misfit. Then, a mediation analysis was conducted using process (Hayes, 2013) with 20,000 bootstrapping iterations. The results revealed a significant indirect effect of RM fit/misfit on willingness to engage in helping behavior via perceived justice (95% CI [0.068; 0.153]), in support of Hypothesis 3. The results of the mediation analysis are depicted in Figure 1.
8 STUDY 2
The results of Study 1 supported our propositions regarding the effects of RM fit/misfit in a social interactive situation on justice perceptions and willingness to engage in helping behavior. The purpose of Study 2 was to find additional empirical evidence for these effects in a different scenario, while additionally considering the schematic nature of relational models. Relational models are defined as cognitive schemata that affect various social domains, and research on RMT suggests that individuals have a tendency to use the same relational model within a given relationship across diverse social domains (Haslam & Fiske, 1999; Vodosek, 2009). Accordingly, it should not be necessary to explicitly describe every social domain of a given social relationship in order to generate expectations concerning the general application of a specific relational model in a given social interactive situastion. In other words: if people are informed that social interactions within a given social relationship are predominantly guided by one specific relational model, this should shape their expectations for this relationship across multiple social domains (e.g., decision making, resource allocation, resource exchange).
8.1 Method
8.1.1 Sample
Participants were recruited through student and graduate mailings lists of two universities in Germany and Austria. As an incentive, participants had the option to take part in a raffle for 150 €, and psychology students could receive course credit for their participation. Overall, 635 people participated in the study. Two participants were excluded from the sample because they had too many missing values. Fifteen participants were excluded from the sample because they did not fulfill the age requirements (a minimum age of 18 years old).
Our final sample consisted of 618 individuals (423 female, 193 male) with an average age of 23.38 years (SD = 5.78) and ranging from 18 to 57 years old. The majority of participants were Austrian (41.7%) or German (39%) and were university students (82.5%).
8.1.2 Design and procedures
Just like in Study 1, we used an experimental vignette design to test our hypotheses.
Our experimental design differed in the following ways from Study 1:
First, the vignettes in Study 2 referred to a dyadic relationship. Thus, we created descriptions of a relationship and an event (i.e., the interaction partner's behavior) that referred to only one fictive colleague. In the course of this, we split the AR model into two separate conditions in which the participant had a lower (AR−) versus higher (AR+) status than his/her interaction partner. Since our hypotheses refer to conflicts between relational models, the logic of our questionnaire made it impossible to directly combine the two AR conditions in the relationship and event description.
Second, while in Study 1 it remained unclear whether and to what degree the participant would share in the scarce resource, the vignettes in Study 2 were described such that each event bore the (fictive) risk that the participant would be disadvantaged in the resource allocation.
Third, we removed the resource allocation domain from our framing vignettes. Similarly to Study 1, we described the relationship in each condition in accordance with one of the relational models, but we did not mention how resources are usually allocated in this relationship (see Appendix B).
8.1.3 Stimulus material
Description of the relationship
The five vignettes including descriptions of a dyadic coworker relationship were formulated based on the relational models scale by Haslam and Fiske (1999) as well as Vodosek's (2009) adaptation of it. Each vignette was formulated in accordance with one relational model and addressed the following domains of a co-worker relationship: decision making, the allocation of tasks and responsibilities, and the general nature of the social relationship between the participant and his/her fictive colleague (“Mr. Miller”). Unlike in Study 1, how resources are typically distributed in this relationship was not addressed in the description of the relationship. To ensure that each vignette unequivocally described just one relational model, the vignettes were independently rated by six experts who were familiar with RMT. All raters correctly identified the intended relational model for each vignette.
Event vignettes
You and Mr. Miller are offered the opportunity to take part in a training program that is of great interest to both you and Mr. Miller. However, you are informed that there is only one free spot left and that you and Mr. Miller will have to decide which one of you can participate. You ask Mr. Miller about his position on this. Mr. Miller reacts in the following way:
In all conditions, Mr. Miller argues that he should attend and states that he thinks it would be fair for the study participant to give him the spot in the training program. However, his justification varied across the five conditions in accordance with the moral motive underlying the respective relational model. Thus, his argument is based on the principle of need in the CS condition, the principle of hierarchy in the two AR conditions, the principle of equality in the EM condition, and the principle of proportionality in the MP condition (see Appendix B).
To ensure that each justification reflected the moral motive underlying the intended relational model, the event vignettes were also independently rated by six experts familiar with RMT. All raters correctly identified the intended relational model in each vignette.
8.1.4 Measures
We used the same measures as in Study 1. All items were adapted to the new scenario and referred to the dyadic relationship described in the vignettes. The reliabilities of the scales are shown in Table 2.
| Variable | M | SD | 1 | 2 | 3 | 4 | |
|---|---|---|---|---|---|---|---|
| 1 | RM fit/misfit | .53 | .50 | – | |||
| 2 | Justice | 2.92 | 1.00 | .13** | (.92) | ||
| 3 | Helping | 3.13 | .82 | −.01 | .43*** | (.87) | |
| 4 | Perceived degree of fit | 2.80 | 1.20 | .13** | .69*** | .37*** | – |
Note
- Reliabilities (Cronbach's alpha) are indicated on the diagonal in parentheses.
- ** p < .01;
- *** p < .001.
8.2 Results
8.2.1 Manipulation check
We again used the perceived degree of fit between framing and vignette as a manipulation check and subjected it to an independent t test. Levene's test indicated unequal variances (F = 10.28, p < .01); thus, the degrees of freedom were adjusted from 616 to 593. Mr. Miller's argumentation was perceived as more appropriate (M = 2.94; SD = 1.15) in the fit condition than in the misfit condition (M = 2.65; SD = 1.24), t(593) = 3.05, d = 0.25 (95% CI [0.09, 0.41], indicating that our RM fit/misfit manipulation was successful.
8.2.2 Effects of demographic variables
Student and non-student participants did not differ with regard to perceived justice (t(616) = −0.088, p = 0.930, g = 0.009. 95% CI [−0.20, 0.22]) and helping behavior (t(170.8) = 0.503, p = 0.615, g = 0.049, 95% CI [−0.26, 0.16]). Similarly there was no significant effect of gender, neither on perceived justice (t(369.34) = 0.832, p = 0.406, g = 0.072, 95% CI [−0.10, 0.24]). There was a significant effect of gender on helping behavior (t(614) = 2.131, p = 0.033, g = 0.185, 95% CI [0.01, 0.35]) with men reporting higher willingness to show helping behavior (M = 3.23, SD = 0.86) than women (M = 3.08, SD = 0.80). However, since in the sample of Study 2 male and female participants were equally distributed in the fit condition and the misfit condition (X2(1) = 0.248, p = 0.619, φ = 0.020), there was no ground to assume that this should have affected our results. The age of the participants had no effect on the perceived justice (r = −0.02, p = .634, 95% CI [−0.10, 0.06]) and helping behavior (r = −0.04, p = 0.38, 95% CI [−0.11, 0.04]).
8.2.3 Hypothesis testing
Correlations, means, and standard deviations for all variables are shown in Table 2.
Hypothesis 1 proposed that an RM fit in a social interactive situation leads to a higher perception of justice than an RM misfit. To test Hypothesis 1, we conducted an independent t test. Supporting Hypothesis 1, participants in RM fit conditions reported higher perceptions of justice (M = 3.04; SD = .99) than participants in RM misfit conditions (M = 2.78; SD = 1.01), t(618) = 3.17, p = .002, d = 0.26 (95% CI [0.10, 0.42]. Participants presented with an event vignette applying the same relational model as the previous description of the relationship reported higher perceptions of justice than participants presented with an event vignette applying a different relational model than in the previous description of the team.
Supporting 2, participants' justice perceptions were positively related to their willingness to engage in helping behavior toward their co-worker (r = .43, p < .001, 95% CI [0.36, 0.50]). The higher the perceived justice, the higher the participants' willingness to help their co-worker in the future.
Hypothesis 3 proposed an indirect effect of RM fit/misfit on helping behavior via perceived justice. We again created a dummy variable (see Iacobucci, 2012) coded 1 for RM fit and 0 for RM misfit and conducted a mediation analysis using process (Hayes, 2013) with 20,000 bootstrapping iterations. The results indicated a significant indirect effect of RM fit/misfit on willingness to engage in helping behavior via perceived justice (95% CI [0.022; 0.092]), in support of Hypothesis 1. The results of the mediation analysis are depicted in Figure 2.
9 STUDY 3
The aim of Study 3 was to replicate the findings of the first two studies with a different scenario and to test H4 and H5, that is, the proposed effect of RM fit/misfit and justice perceptions on participants' willingness to engage in knowledge hiding behavior.
9.1 Method
9.1.1 Sample
Participants were recruited through mailings lists of a German university as well as through the authors' personal networks, social media networks, and by handing out flyers. As an incentive, participants had the option to take part in a lottery of 150 €, and psychology students could receive course credit for their participation. A total of 459 people participated in the study. Four participants were excluded from the sample because they completed the questionnaire in an unrealistically short time (less than 20% of the average time calculated in test runs.)
Our final sample consisted of 455 individuals (256 female, 199 male) with an average age of 30.01 years (SD = 12.67) and ranging from 18 to 76 years old. The majority of participants (96.3%) were German and were university students (51.9%).
9.1.2 Design and procedures
The experimental procedure was the same as in Study 2, except for the usage of different scenario vignettes, the additional assessment of participants' willingness to engage in knowledge hiding behavior, and the fact that the description of the relationship also referenced the social domain in which the event took place.
9.1.3 Stimulus material
Description of the relationship
The five vignettes including descriptions of a dyadic relationship were formulated based on the relational models scale by Haslam and Fiske (1999), its adaption by Vodosek (2009), and descriptions of the four relational models in the pertinent literature (e.g., Fiske, 1992, 2004). Each vignette was once again formulated in accordance with one relational model and addressed the following domains of a co-worker relationship: the distribution of resources, decision making, the allocation of tasks and responsibilities, and the general nature of the social relationship between the participant and his/her fictive colleague (“Mr. Meier”). To ensure that each vignette unequivocally described just one relational model, the vignettes were independently rated by three experts who were familiar with RMT. All raters correctly identified the intended relational model for each vignette.
Event vignettes
One day, the following happens: In your team you have successfully driven a project forward for many weeks. You and your team colleague, Mr. Meier, are now offered the opportunity to attend a management board meeting to complete the project. This is an excellent opportunity to present yourself to the management board and receive positive feedback on your performance. Both you and Mr. Meier would like to take on this role. However, since only one person can attend the meeting, Mr. Meier and you will have to decide among yourselves which of you will get to present the positive results. Immediately, Mr. Meier claims the right to attend the meeting, giving the following reasons:
In all conditions, Mr. Meier argues that he should attend and states that he thinks it would be fair for the study participant to allow him to attend the management meeting. However, his justification varied across the five conditions in accordance with the moral motive underlying the respective relational model. Thus, his argument is based on the principle of need in the CS model, the principle of hierarchy in the two AR conditions, the principle of equality in the EM condition, and the principle of proportionality in the MP condition (see Appendix C).
9.1.4 Measures
Justice perceptions and anticipated helping behavior were assessed using the same measures as in Study 1 and Study 2. The reliabilities of these scales are shown in Table 3.
| Variable | M | SD | 1 | 2 | 3 | 4 | 5 | |
|---|---|---|---|---|---|---|---|---|
| 1 | RM fit/misfit | .52 | .50 | – | ||||
| 2 | Justice | 2.46 | .91 | .20*** | (.91) | |||
| 3 | Helping | 4.71 | 2.80 | .14** | .22*** | (.83) | ||
| 4 | Knowledge hiding | 1.95 | .81 | −.07 | −.10* | −.41*** | (.92) | |
| 5 | Perceived degree of fit | 2.35 | 1.22 | .18*** | .57*** | .23*** | −.16*** | – |
Note
- Reliabilities (Cronbach's alpha) are indicated on the diagonal in parentheses.
- * p < .05;
- ** p < .01;
- *** p < .001.
Knowledge hiding
Participants' anticipated knowledge hiding behavior toward the interaction partner was measured with a German version of the 12-item scale developed by Connelly et al. (2012) translated by Knipfer and Schmid (2019). A sample item is “When Mr. Meier requests knowledge from me, I would offer him some other information instead of what he really wants.” One item had a very low item-total correlation (r = .34) and was, therefore, excluded from the scale. Cronbach's alpha for the remaining items was α = .92.
9.2 Results
9.2.1 Manipulation check
We again used the perceived degree of fit between framing and vignette as a manipulation check and subjected it to an independent t test. The justification was perceived as more appropriate in the RM fit condition (M = 2.56; SD = 1.25) than in the RM misfit condition (M = 2.13; SD = 1.14), t(453) = 3.78, d = 0.36 (95% CI [0.174, 0.545]). This indicated that our RM fit/misfit manipulation was successful.
9.2.2 Effects of demographic variables
Student and non-student participants did not differ with regard to perceived justice (t(432.48) = .999, p = .3180, g = 0.094, 95% CI [−0.28, 0.09]) and helping behavior (t(453) = −1.065, p = .287, g = 0.100, 95% CI [−0.08, 0.28]). However, there was a significant effect of gender on perceived justice (t(397.39) = 2.003, p = .046, g = 0.192, 95% CI [0.01, 0.38]) with men perceiving more justice (M = 2.56, SD = .69) than women (M = 2.39, SD = .05). Furthermore, there was a significant effect of gender on helping behavior (t(453) = 2.003, p = .046, g = 0.189, 95% CI [0.001, 0.374]) with men reporting higher willingness to show helping behavior (M = 2.88, SD = .79) than women (M = 2.73, SD = .77). However, since in the sample of Study 3 male and female participants were equally distributed in the fit condition and the misfit condition (X2(1) = .324, p = .345, φ = .046), there was no ground to assume that this should have affected our results. The age of the participants had no effect on the perceived justice (r = −.052, p = .273) and helping behavior (r = −.002, p = .971, 95% CI [−0.09, 0.09])
9.2.3 Hypothesis testing
Correlations, means, and standard deviations for all variables are shown in Table 3.
Hypothesis 1 proposed that an RM fit in a social interactive situation leads to a higher perception of justice than an RM misfit. To test Hypothesis 1, we conducted an independent t test. Levene's test indicated unequal variances (F = 5.82, p < .05); thus, the degrees of freedom were adjusted from 453 to 450. Supporting Hypothesis 1, participants in RM fit conditions reported higher perceptions of justice (M = 2.64; SD = .96) than participants in RM misfit conditions (M = 2.28; SD = .83), t(450) = 4.29, p < .001, d = 0.40 (95% CI [0.213, 0.585]). Participants presented with an event vignette applying the same relational model as the previous description of the relationship reported higher perceptions of justice than participants presented with an event vignette applying a different relational model than in the previous description of the relationship.
Supporting 2, participants' justice perception was positively related to their willingness to engage in helping behavior toward their co-worker (r = .20, p < .001, 95% CI [0.129, 0.309]). The higher the perception of justice, the higher the participants' willingness to help their co-worker in the future.
Hypothesis 3 proposed an indirect effect of RM fit/misfit on helping behavior via perceived justice. We again created a dummy variable (see Iacobucci, 2012) coded 1 for RM fit and 0 for RM misfit and conducted a mediation analysis using process (Hayes, 2013) with 20,000 bootstrapping iterations. The results indicated a significant indirect effect of RM fit/misfit on willingness to engage in helping behavior via perceived justice (95% CI [0.016, 0.069]), in support of Hypothesis 3. The results of the mediation analysis are depicted in Figure 3.
Supporting 4, participants' justice perception was negatively related to their willingness to engage in knowledge hiding behavior toward their co-worker (r = −.10, p = .04, 95% CI [−0.189, −0.003]). The higher the perception of justice, the lower the participants' willingness to engage in knowledge hiding behavior toward their co-worker in the future.
Hypothesis 5 proposed an indirect effect of RM fit/misfit on knowledge hiding behavior via perceived justice. A mediation analysis using process (Hayes, 2013) with 20,000 bootstrapping iterations revealed a significant indirect effect of RM fit/misfit on willingness to engage in knowledge hiding behavior via perceived justice (95% CI [−0.037 −0.001]), in support of Hypothesis 5. The results of the mediation analysis are depicted in Figure 3.
10 EXPLORATORY ANALYSES
In our studies, we compared an RM fit to an RM misfit in a social interactive situation. In more detail, we introduced one of the four RMs as the predominant RM of a team (Study 1) or a dyadic relationship (Studies 2 and 3) and then described a behavior of the interaction partner(s), which was either congruent with the described predominate RM (i.e., RM fit) or incongruent with the predominant RM (any one of the other RMs; i.e., RM misfit). In our analyses we compared the RM fit / misfit across all RMs, no matter which was the predominant RM of the relationship and based on which RM the (in)congruent behavior was shown.
In additional analyses we wanted to explore, whether we could find the effect for RM fit versus RM misfit for all four RMs being the predominant RM of a relationship (i.e., for all relationship description conditions) and for all four RMs being the basis for the (in)congruent behavior (i.e., for all event conditions). For example: Was there a fit–misfit effect within the scenarios, with a CS predominant relationship? Or was there a fit–misfit effect when the interaction partner showed an (in)congruent AR behavior within the relationship? Etc. To do so, we compared RM fit with RM misfit using perceptions of justice as a dependent variable, given it was the variable theoretically immediately influenced by fit versus misfit.
We conducted the suggested explorative analyses across all three studies (and not for each study separately) in order to increase the power (and the N) of the analyses. Looking at subsamples (e.g., only those participants having received a CS frame) would have otherwise led to very small sample sizes in each cell. We used meta-analytic techniques to combine and aggregate across the three studies; in more detail, we used the computer program Comprehensive Meta-Analysis, which follows the method suggested by Borenstein et al. (2011). We calculated fixed effects models for comparing two independent means (RM fit vs. RM misfit). Please also note that we pooled the conditions AR− and AR+ in one, given that the distinction was only made in Study 2 and Study 3, but not in Study 1.
Calculating at the fit–misfit effect within each of the four RMs as predominant RM of the given relationship(see Appendix D for means and standard deviations), we found the following results: Participants experiencing a CS behavior within a CS relationship (i.e., RM fit) reported significantly higher perceptions of justice than participants, experiencing an AR, EM, or MP behavior within a CS relationship (i.e., RM misfit) (i.e., RM misfit): d = 0.24, p = .043, 95% CI [0.007, 0.473]. The fit / misfit effect was also found for participants having imagined themselves in an EM relationship (d = 1.19, p < .001, 95% CI [0.948, 1.432]). For participants having received the AR relationship (d = 0.11, p = .220, 95% CI [−0.067, 0.290]) or an MP relationship (d = 0.20, p = .115, 95% CI [−0.048, 0.440]) we found only small effects, which were not significant; however, they pointed into the predicted direction.
Calculating the RM fit / misfit effect within the four RMs being the basis for the (in)congruent behavior(i.e., within the four event conditions) in the social interactive situation (see Appendix D for means and standard deviations), we found the following results: Participants experiencing a CS behavior within a CS relationship (i.e., RM fit) reported significantly higher perceptions of justice than participants, experiencing an CS behavior within an AR, EM, or MP relationship (i.e., RM misfit): d = 0.24, p = .043, 95% CI [0.008, 0.457]. The fit / misfit effect was also found for those participants having experienced an AR event (d = 0.76, p < .001, 95% CI [0.566, 0.960]), an EM event (d = 0.49, p < .001, 95% CI [0.262, 0.712]), and an MP event (d = 0.23, p = .053, 95% CI [−0.003, 0.467]). Please note that for MP events the effect was only marginally significant or rather only significant for directed hypotheses, which we had.
We conclude that descriptively we found the predicted RM fit / misfit effect on perceptions of justice for all RM relationships (i.e., for all relationship description conditions) and for all RMs being the basis for the (in)congruent behavior (i.e., for all event conditions). Even though most of the fit / misfit effects were also statistically significant, for a predominant AR relationship and a predominant MP relationship the effect was small and thus, insignificant. Thus, future research is asked to explore different constellations of norm violations (norm violations in different relational contexts and norm violations through different actions with respect to their relational models) in more detail.
11 OVERALL DISCUSSION
The present study explored the effects of a relational model misfit in a given social interactive situation on justice perceptions and subsequently on cooperative (i.e., helping) and uncooperative (i.e., knowledge hiding) behaviors toward co-workers. We proposed that describing a social interactive situation in which a person's fictive interaction partner violates the relational model perceived as predominant in the fictive relationship and that was, therefore, expected to be applied in the social interactive situation at hand leads to feelings of injustice. Furthermore, we proposed that perceptions of injustice, in turn, are negatively related to willingness to engage in helping behavior and positively related to willingness to engage in knowledge hiding behavior toward the interaction partner.
Overall, three experimental vignette studies provided support for our hypotheses regarding the effects of an RM fit/misfit on justice perceptions (Studies 1–3) and willingness to engage in helping (Studies 1–3) and knowledge hiding behavior (Study 3).
In Study 1, we presented our participants with a fictive team described in accordance with one of the four relational models. This description sought to evoke participants' expectations about which relational model is usually applied in the team. Since each relational model incorporates a distinct fairness principle, participants' expectations of what behavior is regarded as fair in the team should depend on which relational model was described as predominant in the team. Participants were subsequently presented with the description of a social interactive situation in which the team applies either the expected or a different relational model in order to create fit or misfit between the expected and the perceived relational model (i.e., RM fit/misfit). The finding that an RM fit leads to higher perceptions of justice than an RM misfit and that higher perceived justice was related to participants' willingness to engage in helping behavior toward the described team supported our theoretical model. It is in line with our proposition that people feel treated unfairly when co-workers break the “relational rules” (stemming from the expected relational model) in a social interactive situation and that this perceived injustice leads to a lower willingness to engage in future cooperative behavior in the relationship.
In Study 2, we aimed to find additional empirical evidence for the proposed effects in a different setting in order to confirm the generalizability of our results to different types of social interactive situations. Furthermore, we aimed to take into account the schematic nature of relational models and the tendency to use the same relational model for a given relationship across diverse social domains (Haslam & Fiske, 1999; Vodosek, 2009). The experimental setup was the same as in Study 1 with the exception of two key differences: First, the participants were presented with a description of a different social interactive situation and of a dyadic relationship instead of a team. Second, the description did not include any information about the social domain of the subsequently described event (i.e., the distribution of resources). Since relational models are proposed to be cognitive schemata, perceiving a given relational model to be dominant in a social relationship should lead to the expectation that this relational model will be applied in all social domains of this relationship (including social domains on which no information has been made available). The results of Study 2 supported this assumption: Even though the relationship description did not include any information about the allocation of resources, an RM misfit in this social domain led to lower perceptions of justice than an RM fit. As in Study 1, perceived justice was related to willingness to engage in future helping behavior toward one's fictive co-worker.
While Study 1 and Study 2 provided empirical support for the proposed effects of an RM fit/misfit on perceived justice and willingness to engage in cooperative behavior, Study 3 additionally examined effects on study participants' willingness to engage in uncooperative behavior. The experimental design was the same as in Study 2 with the exception of this additional dependent variable and a different scenario. As expected, an RM fit lead to higher justice perceptions than an RM misfit. Supporting our hypotheses, perceived justice was both positively related to study participants' willingness to engage in helping behavior and negatively related to their willingness to engage in knowledge hiding behavior toward their interaction partner.
11.1 Theoretical contributions
The results of all three studies provide empirical evidence for a central proposition of RMT (Fiske, 1992; Rai & Fiske, 2011), namely that conflicting relational models lead to perceptions of injustice. In all three studies, we operationalized conflicting relational models by creating a misfit between an expected and a perceived relational model. We manipulated participants' expectations regarding the relational model that is typically applied in the described relationships by describing the relationships as being dominated by just one relational model. The participants then received a situational description in which their interaction partners were described as exhibiting behavior rooted in a relational model that either fit or did not fit their previously manipulated expectations. Consequently, our subjects' expectations either matched or did not match the relational model our subjects perceived (inferred) on the basis of their interaction partner's behavior.
The results of our studies show that an individual's justice perceptions concerning the behavior of an interaction partner in a social interactive situation depend on whether this behavior is in accordance with or in contradiction to the relational model perceived as predominant and thus expected in the relationship. In other words: an individual's justice expectations in a social interactive situation are shaped by the relational background of the respective relationship formed through earlier social interactions.
As the results of Study 2 indicate, the description of the relational model that is typically applied did not have to include a given social domain to raise expectations with respect to this social domain. In the description vignettes for Study 2, participants did not receive any information about how resources are usually allocated—the social domain of the event vignette. Despite this lack of information on the resource allocation domain, an RM fit/misfit had the same effects as in Study 1 and Study 3 (although the effect sizes were smaller). This indicates that describing a relationship in accordance with a single relational model evoked expectations regarding a broad range of social domains, including those that were not part of the original description. In other words: when a relational model is applied to social interactions in one domain (e.g., decision making), this can evoke expectations that the same relational model will be applied in other social domains (e.g., resource allocation) within the relationship. This finding supports the claim that relational models are cognitive schemata that refer to social interaction in multiple domains (Fiske, 1992) and dovetails with empirical research on RMT that revealed a pronounced tendency for individuals to use the same relational model within a social relationship across various domains (Haslam & Fiske, 1999).
By linking perceived justice in co-worker relationships to willingness to engage in helping behavior toward an interaction partner and—albeit with only a small effect size—willingness to withhold knowledge from an interaction partner, the presented studies also contribute to research on cooperative (i.e., helping) and uncooperative (i.e., knowledge hiding) behaviors at work. The less justice participants perceived in the described social interactive situation, the lower their willingness to engage in future helping behavior, and the higher their willingness to engage in future knowledge hiding behavior toward their interaction partner. These results can be interpreted in at least two ways:
First, interaction partners who are perceived as breaking the relational rules and thus evoke perceptions of injustice may be seen as unreliable, causing people to refrain from future interaction and exchange processes with them. This rationale is in line with a “classical” social exchange perspective in which individuals exchange resources on the basis of reciprocity and mutually accepted agreements (Cropanzano & Mitchell, 2005). The less reliable people perceive their interaction partner to be, the lower the perceived probability that this person will respect their agreements in future interactions. Consequently, people who perceive injustice in an interaction may no longer be willing to invest resources (i.e., knowledge, time, effort) in the relationship with their interaction partner because they cannot be sure that this person will not fail to meet their expectations and break the “relational rules” again.
Second, our results can be interpreted as implying that interaction partners who break the relational rules and evoke feelings of injustice may cause behavioral responses that (with regard to intention) go beyond merely reducing one's inputs in terms of a rational social exchange process. According to RMT, people expect their interaction partners to respect the rules of the relational model they apply in a social interactive situation (Fiske, 1992). If their interaction partner violates this relational model (in our studies, by applying another and thus conflicting relational model), people have a strong desire to punish this transgression (Fiske, 1991). From this moral perspective, reducing helping behavior and hiding knowledge can be seen as a form of punishment and sanctioning behavior toward the interaction partners who have evoked feelings of injustice by breaking the relational rules (see below, Future Research).
By examining the effects of conflicting relational models on justice perceptions and (un)cooperative behaviors, the present studies build upon and extend existing research on conflicting relational models in the workplace. For instance, Vodosek (2000) theoretically discussed the effects of the application of different relational models in teams on intragroup conflict. Intragroup conflict, in turn, has been repeatedly linked to employees' justice perceptions (for an overview see Shapiro & Sherf, 2015). An empirical study by Arendt et al. (2019) found the degree of sharedness of relational models in work teams to be related to perceived justice and (un)cooperative behavior among team members. In this study, it was argued that a high degree of sharedness of relational models would lead to social interaction situations in which team members applied different relational models. Arendt et al.'s (2019) finding that the degree of sharedness of relational models is positively related to justice perceptions is in line with the results of the present series of studies that RM fit leads to higher perceptions of justice than RM misfit.
11.2 Future research
The findings of the present studies suggest several avenues for future research.
As discussed previously, the reduced engagement in helping behavior and higher engagement in knowledge hiding behavior can be interpreted as a form of punishment toward interaction partners who have evoked feelings of injustice by breaking the relational rules. This interpretation is in line with the RMT proposition (Fiske, 1992) that people desire to punish interaction partners who have violated the standards of the relational model they applied to the respective interaction. However, more research is needed to examine the extent to which these behavioral responses actually take place with the intention to punish one's interaction partner. Future research could employ a qualitative approach such as interviews to examine people's own perceptions and attributions of their behavioral reactions to the violation of relational rules.
In the present studies, we intentionally focused on misfit in general and not on differences between different combinations of relational models. We did so in order to test the proposition of RMT that all relational models should be incommensurable with each other in social interactions within various domains (Fiske, 1992). While all three studies provide consistent support for this proposition, it is reasonable to assume that some combinations of relational models may be more detrimental to social relationships than others (Simpson et al., 2016). For instance, the application of an MP model in a social context in which another person perceives a CS model to be appropriate seems to be perceived as particularly reprehensible (Fiske, 1992; Rai & Fiske, 2011; Simpson et al., 2016; Tetlock et al., 2000). Adopting an MP model in a (perceived) CS relationship is expected to be much more detrimental to the relationship and cause much more moral outrage than the application of a CS model in a (perceived) MP relationship (Fiske & Tetlock, 1997) or the application of an EM model in a (perceived) CS relationship—two relational models that seem to occur more often in combination with one another (Haslam & Fiske, 1999). Thus, future studies on relational models in organizations could examine whether and to what degree different combinations of expected and perceived relational models are more or less detrimental for future interactions.
Scholars could also try to identify possible moderators of both the effect of RM fit/misfit on justice perceptions as well as the effect of perceived justice on willingness to engage in (un)cooperative behavior. A growing body of research has provided evidence for a trait-like sensitivity toward justice with regard to both (in-)justice perceptions as well reactions to injustice (Baumert & Schmitt, 2016; Schmitt et al., 2010). A recent study examining justice sensitivity in the organizational context identified this construct as a moderator of the effect of workplace stressors on counterproductive working behaviors (Schulte-Braucks et al., 2019). It seems reasonable that justice sensitivity may also moderate the effect of relational model violations on perceived justice as well as the effect of perceived justice on individuals' sanctioning behavior toward their interaction partners. Thus, integrating justice sensitivity as a moderating variable into our research model may help to further explore the relationships among our studies' variables and could be a promising path for future research.
11.3 Limitations
It is commonly suspected that the use of an experimental vignette methodology, in which participants report imagined expectations and behavior, results in limited generalizability and ecological validity. However, for research on cognitions, work attitudes, expectations and behavioral intentions in the realm of organizational behavior, the field in which our three studies are located, experimental vignettes have not only become an accepted methodology (Aguinis & Bradley, 2014) but have also been meta-analytically shown to not differ significantly from field studies (Shaw et al. (2003).
A second issue concerns our samples, which mainly consisted of university students (70% overall), a fact that also affected our participants' average age and level of work experience. Since the scenarios for all three studies referred to an organizational context, these demographic characteristics may have biased our results and may have led to an over- or underestimation of the examined effects, particularly with regard to our outcome variables (i.e., co-worker helping behavior and knowledge hiding). However, we think that these behaviors also play a role in the daily life of university students to a certain extent. For instance, teamwork on student projects is likely to include similar situations in which perceptions of justice play a role and in which individuals have to decide whether to help each other or hide knowledge from their classmates. Nevertheless, future studies would benefit from including more working participants in their samples.
A third limitation of our study concerns the fact that all of the event vignettes presented in the three studies referred to the allocation of some kind of resource (i.e., a bonus, a valuable training program, a valuable opportunity to make a good impression on senior management). We chose this social domain because it allowed us to construct the vignettes in such a way that every misfit between relational models in the framing and the event created a conflict, while still ensuring that the vignettes had very similar and consistent descriptions. However, other social domains (e.g., decision making) could be addressed in a similar way, and future research would profit from replicating our results with a more diverse set of social domains.
11.4 Practical implications
Since both helping behavior and sharing knowledge have been identified as antecedents of various forms of performance-related organizational behavior (Podsakoff et al., 2014; Wang & Noe, 2010), the results of the present study also have relevance for practitioners.
Our finding that justice expectations result from the relational models an individual expects to be applied in a social relationship may help employees gain a better understanding of their co-workers' different views of what is fair in workplace social interactions. Knowledge of such mechanisms may help practitioners—whether in their role as managers or as employees—become aware of different fairness expectations stemming from earlier events and circumstances that made specific relational models salient. The findings of Study 2 are particularly relevant here: Even though participants in this study did not receive any information regarding resource allocation in the described relationship, they perceived a misfit when their interaction partner applied a conflicting relational model (i.e., a relational model different from the relational model in the relationship description) with respect to this social domain. In other words, fairness expectations in a given domain of a relationship can stem from earlier experiences not directly involving this domain. Since the salience of different relational models at work is proposed to be influenced by aspects of the organizational context, such as the HR system (cf., Batistič et al., 2016; Mossholder et al., 2011), organizations may unwillingly raise expectations regarding relational models and thus unwittingly cause relational model conflicts, which are in turn likely to affect (un)cooperative behavior among employees.
COMPLIANCE WITH ETHICAL STANDARDS
The authors certify that the research presented in this manuscript was conducted in compliance with the ethical standards of the DGPs (German Psychological Society) regarding research with human participants and scientific integrity. Participants were free to not participate and to terminate participation at any time without any consequence or any loss they were otherwise entitled to receive. All subjects gave written informed consent in accordance with the Declaration of Helsinki.
CONFLICT OF INTEREST
The authors declare that there are no conflicts of interest.
ACKNOWLEDGMENT
We thank Fiona Kunz, Julia Obermeier, Marcel Obermeier, and Lisa Straßer for their support in conducting Study 1. We thank Simon Böhm for his support in conducting Study 3.
APPENDIX A
TEAM DESCRIPTION VIGNETTES FOR STUDY 1
Instructions
In the following, a fictitious team is described. Please imagine being part of this team. Assume that the described team behaves exactly as described and try to imagine yourself in the situation as vividly as possible.
CS description
Imagine you are working in a team that can be described as follows:
Your team understands tasks as a collective responsibility to which each of you contributes as much as you can. Individual contributions are not tracked or specified because you work as a team. You and your colleagues see yourself as a community with similar attitudes and values. Your team has common traditions; you feel close to your colleagues and support each other. The well-being of others is just as important as your own, and when you need help, everyone does their utmost to help you. You receive this help without being asked for anything in return. Decisions are made in your team according to the principle of consensus. If decisions have to be made, your team will discuss them until all members agree. In addition, you use the resources provided to you as a team by your organization together in the team. Individual demands do not play a role in your team.
AR description
Imagine you are working in a team that can be described as follows:
Tasks are directed and distributed by a specific team member while you and your colleagues follow that person's instructions. In the team, responsibilities are distributed in a differential manner and hierarchies are recognizable. You perceive your team leader to be on a different level than yourself because he is also the one who bears more responsibility than you and your colleagues. Your team leader leads the team professionally and / or disciplinarily. He supports and encourages you and your colleagues. Decisions in your team are always made by your team leader. Your colleagues and you agree with this decision-making process. In addition, the resources provided to you as a team by your organization are distributed in your team according to rank and responsibility. Accordingly, your team leader will receive a larger share than you because of his or her higher position.
EM description
Imagine you are working in a team that can be described as follows:
Tasks are distributed among you and your colleagues in such a way that each team member has equal work to do. When you take over a task from your colleague, you expect your colleague to complete one of your tasks for you in the foreseeable future. If your colleague does you a favor or helps you, you will feel obliged to do him an equal favor or to help him as well. In what you do for each other, you always try to maintain as balanced a relationship as possible. Decisions in your team are made by a majority vote. When decisions have to be made, your team prefers the alternative that gets the most approval. In addition, the resources provided to you as a team by your organization are distributed in such a way that everyone receives exactly the same share.
MP description
Imagine you are working in a team that can be described as follows:
You and your colleagues contribute only so much to the completion of tasks as is worthwhile for you. All team members expect to individually profit from cooperation. This benefit of teamwork is weighed against its costs, such as the effort or time invested. Decisions are made in your team on the basis of balancing profit and costs. When making decisions, every contribution is seen as an investment that should pay off. In addition, the resources provided to you as a team by your organization are distributed so that everyone in the team receives a share in accordance with his or her performance or investment.
Event vignettes for Study 1
CS event
Due to favorable developments, your team will now be provided with an impressive bonus, which can be distributed within the team. In the team, it is decided that this bonus will be distributed according to individual needs. (Example: The team member who is in financial distress through no fault of his own receives a larger share.)
AR event
Due to favorable developments, your team will now be provided with an impressive bonus, which can be distributed within the team. In the team, it is decided that this bonus will be distributed according to each team member's position in the team's hierarchy. (Example: The team leader receives the largest share.)
EM event
Due to favorable developments, your team will now be provided with an impressive bonus, which can be distributed within the team. In the team, it is decided that this bonus will be distributed evenly to all persons. (Example: All team members receive the same share.)
MP event
Due to favorable developments, your team will now be provided with an impressive bonus, which can be distributed within the team. In the team, it is decided that this bonus will be distributed according to individual performance. (Example: The team member who has made the largest contribution receives the largest share.)
APPENDIX B
RELATIONSHIP DESCRIPTION VIGNETTES FOR STUDY 2
Instructions
The following section describes a fictitious colleague, Mr. Miller, and your relationship with him. Imagine that you work with Mr. Miller. Assume that your relationship with him is exactly as described below and try to put yourself as vividly as possible in the situation described.
CS description
Your relationship with Mr. Miller can be described as follows:
In this relationship, you see the tasks that arise as a collective responsibility to which each of you contributes as much as you can. Individual contributions are not tracked or specified because you work as a team. You and Mr. Miller see yourself as a community with similar attitudes and values. You have common traditions; you feel close to each other and support each other. The well-being of each other is as important as one's own and if one of you needs help, the other does his best to help him. This help is provided without anything required in return. Decisions in this relationship are made according to the consensus principle. If decisions have to be made, you discuss until you both agree.
AR+ description
Your relationship with Mr. Miller can be described as follows:
Tasks that arise are usually led by Mr. Miller, while you follow his instructions. Responsibility between you and Mr. Miller is distributed differentially and a hierarchy is recognizable. You perceive Mr. Miller to be on a different level than yourself, because he is also the one who bears more responsibility than you. Mr. Miller supports and encourages you. Decisions are made by Mr. Miller.
AR− description
Your relationship with Mr. Miller can be described as follows:
Usually, you are in charge of the tasks that arise, while Mr. Miller follows your instructions. Responsibility between you and Mr. Miller is distributed differentially and a hierarchy is recognizable. You perceive Mr. Miller to be on a different level than yourself, because you are also the one who has more responsibility than Mr. Miller. You encourage and support Mr. Miller. Decisions are made by you.
EM description
Your relationship with Mr. Miller can be described as follows:
Any tasks that arise will be distributed among you and Mr. Miller in such a way that each of you has the same amount of work to do. When you take over a task from Mr. Miller, you expect him to complete one of your tasks for you in the foreseeable future. If Mr. Miller does you a favor or helps you, you feel obliged to do him an equal favor or to help him as well when the opportunity arises. In what you do for each other, you always try to maintain a balanced relationship. When decisions are made, every vote counts equally; if no agreement is reached, you draw lots or you alternate making decisions.
MP description
Your relationship with Mr. Miller can be described as follows:
You and Mr. Miller contribute only so much to the completion of tasks as is worthwhile for the individual. You both expect to individually profit from the cooperation. This benefit of cooperation is weighed against its costs, such as the effort or time invested. Decisions are made on the basis of balancing profit and costs. When making decisions, every contribution is seen as an investment that should pay off.
Event vignettes Study 2
Introduction
Now please imagine the following scenario:
You and Mr. Miller are offered the opportunity to take part in a training program that is of great interest to both you and Mr. Miller. However, you are informed that there is only one free spot left and that you and Mr. Miller will have to decide which one of you can participate. You ask Mr. Miller about his position on this. Mr. Miller reacts in the following way:
CS event
Mr. Miller argues that he should attend because he needs this training for upcoming projects. He thinks it would be fair for you to give him the spot in the training program.
AR+ event
Mr. Miller argues that he should attend, as he bears significantly more responsibility in day-to-day business and makes more decisions than you. He thinks it would be fair for you to give him the spot in the training program.
AR− event
Mr. Miller argues that he should attend, as he usually carries out your instructions in day-to-day business and “keeps your back clear”. He thinks it would be fair for you to give him the spot in the training program.
EM event
Mr. Miller argues that he should attend and give you priority at the next comparable training course. He thinks it would be fair for you to give him the spot in the training program.
MP event
Mr. Miller argues that he should attend, as he has recently worked significantly more overtime and invested more in the work than you have. He thinks it would be fair for you to give him the spot in the training program.
APPENDIX C
RELATIONSHIP DESCRIPTION VIGNETTES FOR STUDY 3
Instructions
In the following, you will be introduced to a scenario in which you are to imagine yourself in the role of an employee in an organization. Please try to imagine the situation.
You will then be asked to imagine your behavior in this scenario. Remember that this is not an evaluation of your own experiences from professional or everyday life, but only your assessment of the situation presented here.
Please imagine now that you are employed in the company described. In your unit, you work together with several colleagues on various projects. The work and team climate is characterized mainly by the following principles:
CS description
Tasks that arise in your team are handled jointly by you and your colleague Mr. Meier. Each of you contributes as much as individually possible. Carefulness and mutual support are the most important things for you. For this reason, it can happen, for example, that from time to time you perform and help more than Mr. Meier or vice versa, without one of you being better paid or remunerated than the other. When decisions are made, they are discussed until a consensus has been reached between Mr. Meier, you and your other colleagues, i.e., until you all agree. For Mr. Meier and you, the progress of the team as a whole is more important than your own individual progress.
Keywords that characterize your team are community, similarity, sharing, unity, solidarity, compassion, selflessness, and consensus.
AR+ description
Tasks that arise in the team are distributed according to your role in the team. Your colleague, Mr. Meier, assumes more responsibility than you and therefore leads many projects independently. Hierarchy and responsibility are central characteristics of the team. Mr. Meier is willing to exhibit substantial engagement and commitment to his job and this is respected by you and other colleagues. Mr. Meier is paid more than you are for his work. In return, you can expect good guidance from Mr. Meier and willingness from him to take responsibility for unpleasant decisions.
Keywords that characterize your team: authority, rank, hierarchy, superior, leader, subordination, respect, and power.
AR− description
Tasks that arise in the team are distributed according to each person's role in the team. You have more responsibility than your colleague, Mr. Meier, and other team members and therefore lead many projects independently. Hierarchy and responsibility are central characteristics of the team. You are willing to exhibit great engagement and commitment to your profession and for this, you are respected by Mr. Meier and other colleagues. You receive a higher salary for your work than Mr. Meier. In return, Mr. Meier expects you to provide him with good guidance and to be willing to take responsibility for unpleasant decisions.
Keywords that characterize your team: authority, rank, hierarchy, superior, leader, subordination, respect, and power.
EM description
Any tasks that arise in the team are distributed among you, your colleague Mr. Meier and the other team members in such a way that everyone has the same amount of work to do. Balance and equality are important to all of you. If you help Mr. Meier, you expect him to return the favor equally in the foreseeable future. Your vote and that of Mr. Meier are equally important when it comes to votes or decisions. You and Mr. Meier are paid equally for the work you do.
Keywords that characterize your team: equality, alignment, equilibrium, balance, reciprocity, and alternation.
MP description
You and your colleague Mr. Meier contribute only as much to the completion of tasks that arise in the team as pays off for each of you. You expect to individually profit from your cooperation. If you invest more time and effort in a project than Mr. Meier, you are also eligible to benefit more from the success of the project than Mr. Meier. According to this principle, profits, bonuses, and other rewards may be distributed quite differently between Mr. Meier and you, depending on each of your concrete work performance.
Keywords that characterize your team are proportionality, ratio, cost-benefit calculation, economy, share, appropriateness, and pay-off.
Event vignettes for Study 3
Introduction
One day, the following happens:
In your team, you have successfully driven a project forward for many weeks. You and your team colleague, Mr. Meier, are now offered the opportunity to attend a management board meeting to complete the project. This is an excellent opportunity to present yourself to the management board and receive positive feedback on your performance. Both you and Mr. Meier would like to take on this role.
However, since only one person can attend the meeting, Mr. Meier and you will have to decide among yourselves which of you gets to present the positive results. Immediately, Mr. Meier claims the right to attend the meeting, giving the following reasons:
CS event
Mr. Meier thinks it would be fair for him to go because he has not had many opportunities in recent weeks to present himself positively to management. He would like to take this opportunity now.
AR+ event
Although you played a crucial role in shaping the content of the project, Mr. Meier thinks it would be fair for him to go. Because of the high degree of responsibility, he bears for the success of the project, he would be the more suitable contact person for the executive board.
AR− event
Mr. Meier thinks it would be fair for him to go, because participating in the meeting offers him an optimal opportunity for development and promotion.
EM event
Mr. Meier thinks it would be fair for him to go; in return, he would let you go to a comparable meeting at the next opportunity—in the sense of equalizing your advantages.
MP event
Mr. Meier thinks it would be fair for him to go. His extraordinary engagement in the project deserves to be rewarded appropriately by him attending the management meeting.
APPENDIX D
| Experimental Condition | Study 1 | Study 2 | Study 3 | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fit | Misfit | Fit | Misfit | Fit | Misfit | |||||||
| M | SD | M | SD | M | SD | M | SD | M | SD | M | SD | |
| Relationship | ||||||||||||
| CS | 3.49 | .94 | 3.28 | 1.46 | 2.94 | 1.10 | 2.67 | 1.11 | 2.34 | .85 | 2.11 | .75 |
| AR | 3.47 | .77 | 3.15 | .93 | 3.00 | .90 | 2.94 | .87 | 2.70 | .94 | 2.62 | .81 |
| EM | 4.35 | .90 | 2.69 | 1.11 | 3.48 | .94 | 2.56 | 1.17 | 2.97 | .93 | 1.83 | .80 |
| MP | 3.72 | .63 | 2.73 | 1.18 | 2.77 | .95 | 2.78 | .96 | 2.44 | 1.00 | 2.40 | .70 |
| Event | ||||||||||||
| CS | 3.49 | .94 | 2.67 | 1.00 | 2.94 | 1.10 | 2.96 | .87 | 2.34 | .85 | 2.46 | .83 |
| AR | 3.47 | .77 | 2.18 | 1.06 | 3.00 | .90 | 2.43 | 1.07 | 2.70 | .94 | 2.01 | .75 |
| EM | 4.35 | .90 | 3.84 | 1.11 | 3.48 | .94 | 3.00 | .92 | 2.97 | .93 | 2.59 | .78 |
| MP | 3.72 | .63 | 3.02 | 1.07 | 2.77 | .95 | 2.76 | 1.07 | 2.44 | 1.00 | 2.17 | .86 |
Note
- In Study 2 and Study 3, AR+ and AR− conditions were pooled in one condition given that the distinction was only made in Study 2 and Study 3, but not in Study 1.
Open Research
DATA AVAILABILITY STATEMENT
The data that support the findings of all three studies are openly available at the Open Science Framework: https://osf.io/cau7y/.
REFERENCES
- 1 As is common in the pertinent literature (e.g., Cropanzano & Ambrose, 2015), we use the terms justice and fairness interchangeably.
- 2 In the pertinent literature, the terms vignette study and scenario study are used interchangeably. In accordance with recent methodological works (e.g., Aguinis & Bradley, 2014; Atzmüller & Steiner, 2010), we use the term vignette study.
