REVIEW SPORT
Review of Literature
The purpose of this review is to provide the literature and theoretical frameworks
related to the objectives of the study. The review contains many subsections, but the
organization overall is rather uncomplicated. First, the importance of conducting dyad level
research in organizational settings is provided. Second, the literature describing the role of
demographics in work dyads and groups is reviewed and evaluated. Finally, the various
theoretical foundations for relational demography are described.
The Importance of Dyad Research
Tsui, Xin, and Egan (1995) have asserted that much of the research on demographic
diversity in the work place has been performed at the group rather than dyad level. This
contention is surprising given the great deal of research that has shown the importance to
understanding the relationship between the superior and subordinate. For example, Tsui,
Xin, and Egan (1995) contend, “an important factor in how well a team works is the
relationship that a team leader has with each individual team member” (p. 97). Further, Fahr,
Podsakoff, and Organ (1990) indicated that much of the contract between an individual
employee and an organization is derived from the relationship the employee enjoys with his
or her immediate supervisor.
Much of research that has occurred at the dyadic level has occurred over the past three
decades. One paradigm that produced numerous studies at the dyadic level was originally
termed the vertical dyad linkage model (Dansereau, Cashman, & Graen, 1973). Recently this
line of research, which focuses on the leadership exchanges between superiors and
subordinates, has been categorized as the leader-member exchange (LMX) theory (Graen,
Novak, & Sommerkamp, 1982). Tsui et al. (1995) describes the basic tenets of the theory as
“leaders categorize subordinates into two groups: the ingroup (characterized by high trust,
interaction, support, and formal and informal rewards) and the outgroup (characterized by low
trust, interaction, support, and formal and informal rewards)”(p. 99).
The research that has been conducted under LMX theory has been able to establish that
ingroup members enjoy better relationships with, and benefit greater from, their supervisors
than outgroup members. Further, according to Graen and Cashman (1975), supervisors trade
resources (both personal and positional) for cooperation from subordinates. In exchange for
this collaboration, ingroup members benefit from enhanced access to information, decision-
making latitude, supervisory support, opportunity for challenging tasks, and influence.
An example of a study using this framework by Liden and Graen (1980) on 41
superior-subordinate dyads tested the validity of the vertical dyad linkage model, and reported
findings indicating the importance of achieving high quality exchange relationships. Results
signified that subordinates in high quality leader-member relationships, or the ingroup,
indicated having greater job responsibility, having a greater contribution to their work units,
and received higher performance ratings than those in low quality leader-member relationships
(i.e., outgroup relationships). Another study by Vecchio, Griffeth, and Hom (1986) of hospital
employees showed a positive relationship between job satisfaction and a high quality
relationship. In yet another study, Vecchio and Gobdel (1984) confirmed that high quality
leader to member exchanges were important to work related outcomes. Their study of 45
dyads in a business organization showed that subordinates achieving ingroup status were rated
higher by superiors, had fewer intentions to quit, and showed greater satisfaction with
supervision than those in the outgroup.
Another study of 261 superior-subordinate dyads from a telephone company (Duarte,
Goodson, & Klich, 1993), however, failed to indicate conclusive evidence of a positive link
between the quality of exchange and actual objective job performance. Yet, the researchers
reported that subordinates in the ingroup received higher performance appraisal ratings
regardless of the actual objective performance. That is, regardless of the actual performance
level by a subordinate, those enjoying ingroup status received higher subjective performance
appraisals than those experiencing low quality exchanges with a supervisor, which indicates a
positive bias by supervisors toward ingroup members.
In summary, based on conceptual and empirical research conducted under the leader-
member exchange model, the literature generally indicates that members of an ingroup will
experience significantly better relationships with their respective supervisor than those
subordinates in the outgroup. However, what has yet to be established from this literature is
what factors may contribute to the categorization of subordinates into an ingroup or outgroup
by their superiors. Based on preliminary evidence in the literature, there appears to be
evidence that demographic factors alone may have an impact on whether a subordinate will
experience high-quality exchanges (i.e., be an in group member), or low-quality exchanges
with supervisors (i.e., be an outgroup member).
Review of Previous Relational Demography Research
Tusi et al. (1995) have contended that relational demography is the missing link toward
a greater understanding of vertical dyad research. This assertion is based on preliminary
research that suggest that relational demographic similarity between superiors and subordinates
can play an important role in the well being of subordinates. However, the research on
relational demography is in its infancy. Only a few studies have explicitly examined relational
demography at the dyad level, though many studies have analyzed the theory utilizing larger
group samples. Therefore the literature concerning both work groups and work dyads are
presented, however, the reviewed literature in this section heavily emphasizes the relevant
contributions at the dyad level, and only the most noteworthy group level analyses.
The origin of relational demography research stems from the broader work on
organizational demography (Pfeffer, 1983). Organizational demography contends that the
distributional properties of both individual and group demographic characteristics in an
organization can have immense meaning beyond that associated with a demographic attribute
considered in isolation (Pfeffer, 1983). Tsui and O’Reilly (1989) were the first to posit the term
relational demography by suggesting that demographic variation could be analyzed even
further than that proposed by Pfeffer, and in the context of interacting group members. Tsui
and O’Reilly (1989) defined relational demography as the comparative demographic
characteristics of group members, including dyads, who engage in interactions on a regular
basis. They described the conceptualization in detail as:
We propose that knowing the comparative similarity or dissimilarity in given
demographic attributes of a superior and a subordinate or of the members of an
interacting work team may provide additional information about the members’
characteristic attitudes and behaviors and, more important, insight into the processes
through which demography affects job outcomes (Tsui & O’Reilly, 1989, p. 403).
Although Tsui and O’Reilly (1989) were the first to coin the term relational
demography and analyze specific research questions from this framework, previous research
had indicated the importance of relational demographics on interacting members. For example,
the work by Pfeffer and his associates under the organizational demography methodology
produced findings consistent with relational demographic studies. One such study by McCain,
O’Reilly, and Pfeffer (1983) indicated that organizational turnover was related to the
demographics of the group. Results suggested that those group members that belonged to
more homogeneous groups in terms of tenure experienced fewer turnovers. A similar study by
Wagner, Pfeffer, and O’Reilly (1984) examined turnover in top-management groups. Their
findings showed that individuals belonging to groups that were more heterogeneous in terms of
age were more likely to turnover than those in more homogeneous groups.
Another more thorough study by O’Reilly, Caldwell, and Barnett (1989) examined
work group cohesion and turnover. The research on field sales representatives included 20
different work groups consisting of 3 to 6 members each (N = 79). Their findings showed that
work groups that were most similar in terms of tenure, reported greater group cohesion among
the members and lower turnover. Further, findings at the individual level indicated that the
more similar members were in terms of tenure with other members, the less likely they were to
turnover and were more integrated.
There were a few early studies that focused on the relational demographic framework at
the dyad level as well. For example, one of the first studies to discuss a demographic effect in
interacting dyads was completed by Larwood and Blackmore (1978). Larwood and Blackmore
used 60 male and female students in an experiment to understand the behavior of soliciting
volunteer leaders. The study reported that the students tended to solicit leadership toward
members of their own sex more so than the opposite sex.
Liden (1985) studied 35 female bank employees in an attempt to measure the subjects’
reactions to female and male managers. Liden reported that 80% of the female subordinates in
the study actually showed a preference for a male manager. While this finding might
demonstrate that homogenous work teams provide no advantage in the work place, the author
drew a different conclusion. Liden concluded that the relationship was based on situational
variables, and not to gender differences. That is, because the female superiors in the banks
possessed less experience and the male superiors reported having more influence than female
managers, the preference for the male supervisors was based more on rank or position and not
gender of supervisor.
Tsui and O’Reilly (1989) were the first to perform a study under what is now
considered relational demography. The researchers framed their study under the similarity-
attraction paradigm and analyzed the effects of age, gender, education, company tenure, and
job tenure dissimilarities on four outcome variables. The outcome variables included
reputational effectiveness, supervisory affect, role ambiguity, and role conflict. The study
analyzed superior-subordinate dyads (N = 272) from a Fortune 500 multidivisional
corporation. Analysis included the use of blocked regressions that included the superior’s
demographics (block one), the subordinate’s demographics (block two), and relational
demographics (block three). The results of the study indicated a relational demographic effect
on three of the four outcome variables, and significant beta weights were yielded on 13 of the
24 possible relational demographic variables. Some of the most noteworthy findings included
subordinates in mixed-gender dyads were rated to perform worse and were liked less well than
those subordinates in a same-gender dyad. Subordinates in the mixed-gender dyads also
reported higher levels of role conflict and role ambiguity. Supervisors also indicated a greater
liking for subordinates with shorter job tenures than themselves, than those with the same or a
greater amount of tenure.
Weslowski and Mossholder (1997) conducted a more recent study under the concept of
relational demography, and framed the analysis under self-categorization theory. The
researchers tested if demographic dissimilarity between the dyad for the variables of race,
gender, age, and education affected subordinates job attitudes of job satisfaction, burnout, and
perceived procedural justice. The researchers collected data from 124 superior-subordinate
dyads working at two different service-oriented companies, and primarily used polynomial
regressions for analyses. The results of the study yielded significant relational effects for the
race variable only. Specifically, relational race was found to correlate with perceptions of
procedural justice and job satisfaction, but not for burnout. That is, those in mixed-race dyads
indicated significantly lower means for job satisfaction and procedural justice than those in
same-race dyads.
Another dyad level study that utilized a relational demography methodology was
conducted by Epitropaki and Martin (1999). The researchers analyzed the impact of
differences in age, organizational tenure, and gender between subordinates and their managers
as a potential moderator between the quality of leader-member exchanges, organizational
commitment, job satisfaction, and job-related well-being. The findings of the study did not
show direct relational demographic effects on any of the work-related outcomes. However, the
researchers did reveal some evidence of the moderating effect of relational demographics on
work outcomes. For example, when LMX was low, a high age difference was associated with
lower well being than when the age differences were low. That is, employees with a high
difference in age to their manager and low LMX, indicated lower well-being. Organizational
tenure differences between the manger and subordinate also moderated the relationship
between LMX and organizational commitment, job satisfaction, and well-being. Therefore,
the study indicated that those experiencing low LMX and having large organizational tenure
differences reported the lowest organizational commitment, job satisfaction, and well-being.
Any gender differences between the manager and subordinate were not found to moderate the
LMX relationship and the work outcomes. While this study fails to provide evidence of a
direct relational demographic effect on the work outcomes, it does demonstrate the importance
of relational demography in understanding work outcomes through the moderation in the
leader-member exchanges of dyads.
Judge and Ferris (1993) studied the extent that a demographic dissimilarity between a
superior and subordinate would affect the performance appraisal process on 81 registered
nurses and their supervisors from a hospital in central Illinois. The researchers chose just two
demographic variables for analysis, age and tenure, and hypothesized that the more similar a
the supervisor and subordinate were with respect to theses two variables, the more the
supervisor reported liking the subordinate, which would indirectly have an effect on a positive
performance rating. Results of the study supported the hypothesis in that increased similarity
between the dyad on a composite score of both age and tenure, positively affected supervisors'
affect toward subordinates, and therefore, indirectly affected a positive performance appraisal.
Another dyad level study by Green, Anderson, and Shivers (1996) assessed the effects
of organizational (e.g., work unit size) and demographic characteristics (e.g., age, gender, and
education differences) on the quality of leader-member exchanges (LMX) among a sample of
208 public library employees. Furthermore, the researchers assessed the relative contributions
of the organizational characteristics, relational demographics, and LMX on a subordinate’s
work attitudes (i.e., job satisfaction and organizational commitment). The results indicated that
a gender difference was evident on LMX and that LMX was of lower quality when the
subordinate and superior were of different genders. Further, the relational gender difference
almost always took the form of a female subordinate and a male supervisor. Therein, the
authors contend, “the presence of a male manager with a female subordinate may have taken
on special significance in this work setting and altered the LMX development process” (Green
et al., 1996, p. 210). The negative effect on the quality of LMX was subsequently found to
affect the job satisfaction variable indirectly through a positive effect of LMX on satisfaction.
The relational education variable was found to have a direct effect on organizational
commitment. The authors also report that organizational characteristics (unit size and work
load experienced) were negatively related to LMX quality.
An important group level, as opposed to dyad level, study by was conducted by Fields
and Blum (1997). The study analyzed the relationship between an employee’s job satisfaction,
and the gender composition of his or her work group. The authors surveyed a total of 820 men
and 814 women representing employed persons from across the United States. Results of the
study indicated that both men and women working in a gender-balanced group (similar
amounts of male and females) had higher job satisfaction levels than those working in more
homogeneous groups (i.e., mostly male or mostly female). Further, employees that worked in
groups containing mostly men indicated the lowest levels of job satisfaction from the other
groups. Those employees working in groups of mostly females, indicated job satisfaction
levels in the middle of the continuum. Although this study did not use relational demography
as a framework for analysis, the study further iterates the importance of demographic
characteristics on the well being of employees.
A more recent study by Lichtenstien and Alexander (2000) did use relational
demography as a framework for analysis. The study utilized data from 38 hospitals and
hospital administrative offices (N = 1,795). The authors hypothesized that perceptions of
advancement opportunities of employees with regard to demographic dissimilarity to the work
group in public sector organizations (i.e., VA hospital employees) would differ from previous
research utilizing private sector organizations. That is, the authors contended that being
demographically dissimilar to co-workers in a public sector organization would result in much
different results on perceived advancement opportunities than other relational demography
research that indicate a negative effect on the construct. Results of the study partially
supported the hypotheses, at least with respect to relational age and race. The results indicated
that the more dissimilar an employee was with regard to age and race, the greater the
perception of advancement opportunity was. The authors contend that these findings, which
contradict previous studies, could be attributed to the many equal opportunity policies that
public sector organizations pursue, which altered the expected relationship between
dissimilarity in demographics and perceptions of advancement opportunity in ones job.
Pelled (1996) conducted a study of 233 blue collar workers and assessed if a
demographic dissimilarity from those in a work group (n = 42) affected how individual’s
perceived the groups performance and conflict. Pelled used a relational demography
framework to shape the study hypotheses and assessed differences among the interacting
members on the demographic variables of gender, organizational tenure, and race. The model
included two hypotheses and assessed the effects of relational demography on the outcome
variables of perceived emotional conflict and perceptions of group performance. Results
indicated that both gender and tenure dissimilarity had positive relationships with the perceived
emotional conflict construct. Demographic dissimilarity was also negatively related to the
ratings of group performance indirectly through the conflict perception variable. That is,
although the demographic dissimilarities did not have a direct effect on the perceived
productivity of the group, the negative relationships toward the emotional conflict variable,
which subsequently predicted less perceived productivity, indicates that relational demography
can affect the confidence members have toward their group.
Jackson et al. (1991) produced a study that analyzed both the effects of demographic
similarity to a group on individuals and groups. The researchers examined the demographic
differences among the variables of age, organizational tenure, educational level, college
curriculum, industry experience, and military experience on a sample of 93 top management
teams (totaling 625 individuals) in the banking industry. The study was conducted under
similar theories—the attraction-selection-attrition model (ASA) (Schneider, 1987) and the
organizational demography model (Pfeffer, 1983). The effects of individual dissimilarity and
group heterogeneity on the outcome variables of recruitment, promotion, and turnover were
assessed using analysis of variance, correlations, and regressions. The group level analyses
indicated that group heterogeneity predicted turnover. That is, the more dissimilar the work
group was in terms of the demographic variables, the more turnover the team experienced over
the four-year period under study. The results at the individual level indicated that a
dissimilarity between the individual and the work group with respect to the demographic
variables predicted higher turnover. The results lent support for both models under
investigation in that demographic similarity (organizational demography) and psychological
similarity (ASA) effects on the outcomes were noted.
A group level study by Mueller et al. (1999) studied teachers in 405 urban school
district schools under relational demography theory, a racial prejudice framework, and status
characteristics theory. The studied assessed the direct effects of the racial composition of the
schools teachers and students on job satisfaction, organizational commitment, and career
commitment. Furthermore, the researchers were interested in assessing what variables would
act as mediators of the group composition differences effect on satisfaction and commitment
constructs. Specifically, the authors main hypotheses was that teachers that worked in schools
in which there own race was dominant would experience greater commitment and job
satisfaction than teachers in schools where a race other than their own was primary. Further,
the authors tested whether White teachers in schools with predominately White colleagues
would experience greater coworker support, role conflict, and autonomy, a contention that is
grounded in the nonsymmetry hypothesis. Results indicated that the racial composition of
schools affected the White teachers but not the Black teachers, which lent support for the
nonsymmetry hypothesis that Whites in Black-dominant settings often react more negatively
than Blacks in White-dominant settings. The specific results indicated that White teachers in
“mismatched” settings (e.g., in a Black-dominant school) experienced greater role conflict,
inadequate resources, and less job autonomy. These negative effects in turn shaped less job
satisfaction and organizational commitment. However, no racial composition effects on career
commitment were found. As such, the authors assert that relational demography effects appear
to have “more short-term than long-term effects on teachers” (Mueller et al., 1999, p. 211).
Tsui, Egan, O’Reilly (1992) constructed a framework built on self-categorization
theory to test a series of hypotheses regarding relational demographic differences with respect
to age, race, tenure, education, and gender. The researchers tested the effects of relational
demography on the outcome variables of commitment, attendance behavior, and tenure
intentions among 1,705 workers across three different industries. The results indicated that an
increase in work-unit diversity among group members negatively affected the psychological
attachment of the individuals. Specifically, the researchers reported general relational
demography effects on three of the five difference variables. The tenure, gender, and race
variables all accounted for a difference in all three of the outcome variables. For the gender
and race relational scores, the direction of the relationship was as hypothesized. Thus, the
greater the difference in gender and race of the individual toward the work group, the lower
levels of commitment, the higher the frequency of absences, and the lower stay intentions were
experienced by the employees. However, the hypothesized effects for education and tenure
were not supported. In fact, the opposite effects to those hypothesized were actually found to
exist.
Numerous interesting results were noted in the study by Tsui et al. (1992). The most
noteworthy of these emerged in the nonsymmetrical effects analysis for the gender variable
(i.e., separate analysis for each gender). For the men in this analysis, an increase in the gender
composition of the work group was actually associated with less psychological attachment,
increased absence, and fewer stay intentions. However, for women, an increase in the gender
from others in the group was associated with greater levels of organizational attachment. Thus,
it appears that men are more affected by an increase in the heterogeneity of a work group, and
would have more positive psychological outcomes in a male-dominated or all male setting.
However, females appear to be unaffected by an increase in the gender heterogeneity a work
group.
Theoretical Frameworks Explaining Demographic Effects
Relational demography effects can be best explained by self-categorization theory, a
theory grounded in Tajfel’s (1974) social identity theory. Hogg and Terry (2000) contend that
Tajfel developed the theory to indicate “how beliefs about the nature of relationships between
groups (status, stability, permeability, legitimacy) influence the way that individuals or groups
pursue positive social identity” (p. 122-123). One principle of an individual’s social identity is
that of self-enhancement, which is related to one’s self-esteem (Riordon, 1995). Individuals
are expected to desire to establish a high level of self-esteem (e.g., Brockner, 1988), this in turn
will motivate them to achieve a favorable self-identity (Tajfel & Turner, 1986). However, in
order for individuals to identify how they feel about others, they are required to identify
themselves first (Tsui et al., 1992).
To establish this identity, individuals are expected to partake in a self-categorization
process (Turner, 1987). Self-categorization elaborates on social identity theory and contends
that individuals “classify themselves and others into social categories using characteristics such
as organizational membership, age, race, status, or religion” (Tsui et al., 1992). These social
categories allow individuals to define themselves in terms of social identity (Miklos, 1999).
Furthermore, the process allows individuals to assume a more positive self-identity and s/he
may consequently seek to maximize their ingroup uniqueness and disfavor the outgroups
distinctiveness (Kramer, 1991). Stephan and Stephan (1985) have asserted, “people who are
regarded as superior experience anxiety concerning interaction with others who are regarded as
inferior” (p. 163). This anxiety can in turn challenge one’s self-esteem and enable people to
avoid contact with members of an outgroup, and to increase the stereotyping behavior toward
the outgroup (Tsui et al., 1992).
The theory is ideal for analysis in an organizational context because existence of the
numerous groups in these setting (e.g., work groups, supervisor-subordinate dyads,
management groups) and the research that suggests that individuals prefer to function in
homogeneous groups of similar others rather than in a group of dissimilar others (e.g.,
Schneider, 1987). Relational demographics are relevant because individuals often classify
themselves and others into categories using various characteristics such as gender, race, age,
tenure, and education (Riordan, 1995; Tsui et al., 1992; Zenger & Lawrence, 1989). Thus, if an
individual’s demographic background and characteristics (such as age, gender, tenure, race, or
religion) make them distinct, she or he may engage in social identification and subsequent self-
categorization based on the particular background or characteristic (Pelled, 1996).
At the dyad level, self-categorization theory and relational demographic effects
contends that demographic dissimilarities between the two members can lead to an increase in
polarization between the members based on definition of the social group as a whole (i.e.,
conflicting outgroup or ingroup group memberships when compared to the broader group as a
whole) (Turner & Oakes, 1989). Further, when demographic dissimilarities exist within a dyad,
subordinates and superiors may tend to stereotype each other and emphasize their differences
(Weslowski & Mossholder, 1997). These contentions, along with the literature reviewed
indicating the importance of ingroup and outgroup categorizations in dyad studies, and the
reviewed relational demographic studies, suggest that dyads that differ demographically may
negatively affect individual work-related variables.