See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/232563951

Meta-Analytic Review of LeaderMember

Exchange Theory: Correlates and Construct
Issues.

Article in Journal of Applied Psychology November 1997

Impact Factor: 4.31 DOI: 10.1037/0021-9010.82.6.827

CITATIONS READS

1,061 336

2 authors, including:

David V. Day
University of Western Australia
89 PUBLICATIONS 6,094 CITATIONS

SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: David V. Day
letting you access and read them immediately. Retrieved on: 19 June 2016
Journal of Applied Psychology Copyright 1997 by the American Psychological Association, Inc.
1997, Vol. 82. No. 6, 827-844 0021-9010/97/S3.00

Meta-Analytic Review of Leader-Member Exchange Theory:

Correlates and Construct Issues

Charlotte R. Gerstner and David V. Day

The Pennsylvania State University

The leader-member exchange (LMX) literature is reviewed using meta-analysis. Rela-

tionships between LMX and its correlates are examined, as are issues related to the
LMX construct, including measurement and leader-member agreement. Results suggest
significant relationships between LMX and job performance, satisfaction with supervi-
sion, overall satisfaction, commitment, role conflict, role clarity, member competence,
and turnover intentions. The relationship between LMX and actual turnover was not
significant. Leader and member LMX perceptions were only moderately related. Partial
support was found for measurement instrument and perspective (i.e., leader vs. member)
as moderators of the relationships between LMX and its correlates. Meta-analysis showed
that the LMX7 (7-item LMX) measure has the soundest psychometric properties of all
instruments and that LMX is congruent with numerous empirical relationships associated
with transformational leadership.

Within the broad area of organizational leadership,
leader-member exchange (LMX) theory has evolved into
one of the more interesting and useful approaches for
studying hypothesized linkages between leadership pro-
cesses and outcomes. First proposed by Graeri and col-
leagues (Dansereau, Cashman, & Graen, 1973; Danser-
eau, Graen, & Haga, 1975; Graen, 1976; Graen & Cash-
man, 1975), LMX is distinguished from other leadership
theories by its focus on the dyadic relationship between
a leader and a member. Unlike traditional theories that
seek to explain leadership as a function of personal char-
acteristics of the leader, features of the situation, or an
interaction between the two, LMX is unique in its adop-
tion of the dyadic relationship as the level of analysis.
Although the theory has been modified and expanded
Charlotte R. Gerstner and David V. Day, Department of Psy-
chology, The Pennsylvania State University.
We gratefully acknowledge the contributions made by Dan
Brass, Jim Farr, John Mathieu, Janet Swim, and Francis Yam-
marino on earlier versions of this article. We also thank Ste-
phanie Klein for coding all the studies used in these analyses
and Chuck Pierce for assisting us with data analysis. This article
is based on Charlotte R. Gerstner's master's thesis. An earlier
version was presented at the 10th annual conference of the Soci-
ety for Industrial and Organizational Psychology, May 1995,
Orlando, Florida.
Correspondence concerning this article should be addressed
to Charlotte R. Gerstner, Department of Psychology, The Penn-
sylvania State University, 429 Bruce V. Moore Building, Univer-
sity Park, Pennsylvania, 16802-3104.
since first proposed, this basic unit of analysis has re-
mained unchanged.
According to LMX, the quality of the relationship that
develops between a leader and a follower is predictive of
outcomes at the individual, group, and organizational lev-
els of analysis. Dyadic relationship development is
grounded in role and exchange theories (see Graen &
Uhl-Bien, 1995; Liden, Sparrowe, & Wayne, 1997; Uhl-
Bien, Graen, & Scandura, 1997, for more detailed discus-
sions of the theoretical background and development of
LMX). After more than 25 years of empirical research
and theoretical development, LMX continues to provide
an operable alternative to the traditional leadership ap-
proaches focused on leader traits and behaviors (e.g.,
Bass, 1990; Mintzberg, 1973; Stogdill, 1948).
Although researchers remain enthusiastic about LMX,
there is unresolved ambiguity about the nature of the con-
struct, its measurement, and its relationships with other
organizational variables. This ambiguity may be due in
part to the evolving nature of LMX theory. In a recent
review, Graen and Uhl-Bien (1995) classified the evolu-
tion of LMX theory into four stages: (a) work socializa-
tion and vertical dyad linkage where the focus was on the
discovery of differentiated dyads (i.e., in-groups and out-
groups), (b) LMX where the focus was on the relation-
ship quality and its outcomes, (c) a prescriptive approach
to dyadic partnership building, and (d) LMX as a sys-
tems-level perspective (i.e., moving beyond the dyad to
group and network levels). The latter two stages are fairly
recent developments, and most of the work associated with
them is theoretical. The majority of empirical research

827
828 GERSTNER AND DAY
evaluating factors thought to contribute to high-quality
exchanges and analyzing the connection between LMX
and work-related outcomes occurred during the second
stage of LMX theory development.
It could be argued that the last two stages show the most
potential for describing leadership behaviors in complex
organizations. Clearly, more research is needed to under-
stand these extensions of LMX theory. The goal of this
article is to quantitatively summarize and evaluate the
existing LMX research base to build a foundation for
future empirical work. More specifically, we consider is-
sues related to the LMX construct, including measurement
scale properties and leader-member agreement, as well
as the correlates of LMX. We use the results of these
quantitative summaries to draw conclusions about LMX
and identify opportunities for future theoretical develop-
ment and empirical research.
An Analysis of LMX Issues
Measurement and Dimensionality
Instruments. Despite claims of an apparently robust
phenomenon (e.g., Graen & Uhl-Bien, 1995), there is
surprisingly little agreement on what LMX is or how it
should best be measured. The theoretical progression of
LMX described above is illustrated by the changes in
LMX measurement instruments over the years. The con-
struct of LMX has evolved from the two-item measure
of negotiating latitude (NL) to more elaborate, multidi-
mensional scales (e.g., Liden & Maslyn, in press; Schries-
heim, Neider, Scandura, & Tepper, 1992). Because differ-
ent studies use different LMX scales, it is unclear whether
conflicting results are due to deficiencies in the theory or
in the operationalization of the core construct. Through
the use of meta-analytic techniques, it was possible to
examine the type of measurement instrument as a potential
between-study moderator of outcomes. On the basis of
the amount of research that has gone into its development
and refinement, and the recent recommendation provided
by Graen and Uhl-Bien (1995) that it be adopted as the
standard measure of LMX, we expected that the seven-
item LMX measure (LMX7; Graen, Novak, & Sommer-
kamp, 1982) would demonstrate the highest reliability
(i.e., internal consistency) and largest correlations with
other variables, as compared with other LMX measures.
Internal consistency and unidimensionality. An issue
that has been raised by numerous researchers (e.g.,
Dienesch & Liden, 1986; Liden & Maslyn, in press; Liden
et al., 1997; Schriesheim et al., 1992) concerns the potential
multidimensionality of LMX. For example, Dienesch and
Liden proposed that LMX is comprised of the dimensions
of perceived contribution, loyalty, and affect. Graen and
Uhl-Bien (1995) argued that LMX is comprised of the
interrelated dimensions of respect, trust, and mutual obliga-
tion. We addressed the possibility of multiple dimensions
underpinning the LMX scales by examining aggregated in-
ternal consistency (i.e., coefficient alpha) estimates of
LMX.1 Graen and Uhl-Bien noted that Cronbach alphas for
multidimensional LMX measures were consistently in the
.80- .90 range. Therefore, they concluded that LMX may
comprise several dimensions, but they are al] highly related
and can be adequately measured with a unidimensional mea-
sure of LMX. Thus, we expected that averaged alphas
would be generally large (i.e., above .80), indicating the
likelihood of one underlying LMX dimension summarized
by the centroid item of the LMX7 scale (i.e., "How effec-
tive is your working relationship with your leader [fol-
lower]?"; Graen & Uhl-Bien, p. 236).
Perspective. LMX can be measured from both leader
and member perspectives, but is it the same construct
when measured from different perspectives? Empirical
support for the relationship between leader LMX and
member LMX has been equivocal. Graen and Cashman
(1975) found a correlation of .50 between leader LMX
and member LMX; however, others have reported much
lower correlations (e.g., Scandura, Graen, & Novak,
1986, reported a correlation of only .24 between perspec-
tives). The low correlation between leader and member
exchange is consistent with meta-analytic research on
self-supervisor agreement on performance ratings, which
has demonstrated relatively low agreement (r = .35, cor-
rected; Harris & Schaubroeck, 1988). Graen and Uhl-
Bien (1995) suggested that the degree of leader-member
agreement can be used as an index of the quality of data.
They implied that an aggregate, sample-weighted correla-
tion assessing the level of agreement between leader and
member reports would be positive and strong. We tested
this proposition by estimating the sample-weighted corre-
lation between corresponding leader and member LMX
ratings. We also examined measurement perspective as
a potential between-study moderator of the relationship
between LMX and its correlates.
Correlates
LMX is generally found to be associated with positive
performance-related and attitudinal variables, especially
1
Although a large alpha (e.g., > .80) cannot always be inter-
preted as evidence of scale unidimensionality (Cortina, 1993),
it does provide information regarding the average interitem cor-
relation. Thus, if the number of items is reasonably small (e.g.,
< 10), a large coefficient alpha estimate can be construed as
meaning that the average item intercorrelation is also relatively
large. It is possible that a scale with a relatively high alpha may
represent a multidimensional construct; however, the dimensions
are probably highly intercorrelated.
 LEADER-MEMBER EXCHANGE META-ANALYSIS
for members. These include (a) higher performance rat-
ings (e.g., Liden, Wayne, & Stilwell, 1993), (b) better
objective performance (e.g., Graen, Novak, & Sommer-
kamp, 1982; Vecchio & Gobdel, 1984), (c) higher overall
satisfaction (e.g., Graen, Novak, & Sommerkamp, 1982),
(d) greater satisfaction with supervisor (e.g., Duchon,
Green, & Taber, 1986), (e) stronger organizational com-
mitment (e.g., Nystrom, 1990), and (f) more positive
role perceptions (e.g., Snyder & Binning, 1985). Support
for other relationships, such as member competence (cf.
Kim & Organ, 1982; Liden et al., 1993) and turnover (cf.
Graen, Liden, & Hoel, 1982; Vecchio, 1985), has been
equivocal.
Our meta-analysis provided overall effect estimates for
previously investigated correlates of LMX.2 In general,
we expected that correlations between LMX and member
performance and attitudinal variables would be positive
and strong (the exceptions being turnover processes and
role conflict, in which cases the direction was expected
to be negative). We expected LMX scale reliability to be
generally acceptable and leader and member LMX ratings
to be only moderately correlated. We proposed and tested
specific moderating variables. In particular, we hypothe-
sized that the LMX7 scale would demonstrate the best
general predictive validity, although measures from the
different perspectives (leader vs. member) might demon-
strate divergent relationships due to variability in the con-
struct's meaning inherent in these perspectives.
Method
Literature Search
The studies included for analysis consisted of published arti-
cles, conference papers, doctoral dissertations, and unpublished
manuscripts. We identified relevant published articles primarily
through computer-based searches of the Psyclnfo (1984-1996),
PsycLit (1974-1984), and ABI/INFORM (1975-1996) data-
bases using the keywords leadermember exchange, vertical
dyad linkage, exchange quality, and dyadic leadership. We used
the Social Sciences Citations Index to identify articles that refer-
enced either of the seminal LMX articles (i.e., Dansereau et al.,
1975; Graen & Cashman, 1975). The reference lists of identified
empirical studies and theoretical reviews were also cross-
checked for additional references. Finally, we conducted a man-
ual search of Academy of Management Journal, Administrative
Science Quarterly, Group and Organization Management, Hu-
man Relations, Journal of Applied Psychology, Organizational
Behavior and Human Performance (later Organizational Be-
havior and Human Decision Processes), and Personnel Psy-
chology from 1975 to 1996 to identify studies that measured
LMX, although it may not have been the primary variable of
interest. We identified conference papers presented during the
period from 1990 to 1994 by manually searching the programs
for the annual Society for Industrial-Organizational Psychology
and Academy of Management meetings. We identified disserta-
829
tions using the Dissertations Abstracts International (1975-
1996) computer database. Although unpublished manuscripts
were not actively solicited, some authors sent unpublished manu-
scripts when communicating about other published studies.
These unpublished studies were also included in the analyses.
We identified a population of 164 studies using these proce-
dures. The criteria for inclusion in the analyses were (a) dyadic
exchange quality was measured in the study; (b) the reported
results were sufficient to calculate an effect size for the relation-
ship between exchange quality and a correlate, or between leader
and member LMX perceptions; and (c) the relationship reported
was also reported in at least five other studies. The last criterion
was necessary to ensure that the number of studies for each
meta-analysis was adequate for drawing generalizable conclu-
sions about the proposed relationships. Although it has been
suggested that a meta-analysis can be conducted with as few as
two studies (Hunter, Schmidt, & Jackson, 1982), second-order
sampling error poses a distinct threat to the validity of results
when only a small number of studies are included (Hunter &
Schmidt, 1990). There is no strict rule about the minimum
number of studies to include in a meta-analysis, but we chose
six as the cutoff for our study because of concerns about Type
I and Type II errors in the moderator analyses (Sackett, Harris, &
Orr, 1986; Spector & Levine, 1987). On the basis of these
criteria, we retained 79 studies for the analyses.3 Six studies
contained two samples; therefore, we worked with a total of 85
independent samples. The relationships analyzed are described
in the following section.
Constructs Included in the Meta-Analyses
LMX. Although this term is commonly found in the litera-
ture to refer to the dyadic relationship between a leader and a
member, construct operationalization is far from consistent. We
identified seven different versions of the LMX scales developed
by Graen and colleagues, two additional LMX measures (Li-
den & Maslyn, in press; Schriesheim et al., 1992), and several
modified versions of these scales. The LMX7 scale (Qraen,
Novak, & Sommerkamp, 1982) is by far the most frequently
used LMX measure. LMX7 differs from the earlier VDL mea-
sures in that it does not focus on the amount of negotiating
latitude a leader allows a member but rather on the nature of
their general working relationship.
To be included in the analyses, the study had to measure
dyadic exchange quality in a manner consistent with the general
themes found in the LMX7 scale. Because of the variability
^
From a conceptual standpoint, it makes sense to divide the
literature on LMX correlates into two categories: antecedents
(i.e., those variables hypothesized to affect the development of
leader-member relationships) and outcomes (i.e., those vari-
ables hypothesized to result from LMX). Because most of the
empirical research reviewed in this article is correlational and
cross-sectional, we avoid discussing them in terms of causal
inferences regarding the direction of these relationships. For
purposes of the present analyses, we treat them all as correlates.
3
A list of the reasons for study exclusion is available on
request from Charlotte R. Gerstner.
830 GERSTNER AND DAY
found in scales explicitly labeled "LMX," we also included a
few other measures that were used to assess the LMX construct
(e.g., Leader Behavior Description Questionnaire; LBDQ). We
included the LBDQ because it was used in the early empirical
research before the specific LMX scales were developed (e.g.,
Graen, Dansereau, Minami, & Cashman, 1973). However, we
included studies using the LBDQ only if the item referents were
changed to reflect a dyadic perspective. We did not include
studies that used the Multifactor Leadership Questionnaire
(Bass, 1985) because these relationships are not consistently
dyadic in nature and have recently been reviewed by Lowe,
Kroeck, and Sivasubramaniam (1996). We evaluated studies on
a case-by-case basis to determine if the constructs used could
be classified as LMX. In cases of uncertainty, we contacted the
author(s) to determine if they would classify their measure as
LMX.
Performance ratings. This construct was operationalized as
subjective, supervisory ratings of a member's job performance.
We excluded global measures of competence (i.e., not related
to performance in a specific job) from this analysis but analyzed
them separately in the relationship between competence and
LMX.
Objective performance. Objective performance referred to
all measures of performance that did not rely on a subjective
rating, which typically meant indexes of the quantity or quality
of work, such as the total number of cases processed or the total
sales in dollars (e.g., Graen, Novak, & Sommerkamp, 1982;
Tanner & Castleberry, 1990).
Satisfaction. We examined two constructs related to work
satisfaction in the meta-analysis: overall job satisfaction and
satisfaction with the supervisor. Overall satisfaction referred to
how satisfied a member was with the job in general or the
organization, whereas satisfaction with the supervisor referred
more explicitly to how satisfied a member was with his or her
working relationship with a supervisor. Several different scales
were used to measure satisfaction, including the Job Description
Survey, Hoppock Scale, and Minnesota Satisfaction Question-
naire. Satisfaction with supervisor was usually measured using
one of the facet scales associated with these general measures
of satisfaction.
Organizational commitment. Although there are many dif-
ferent types of commitment discussed in the literature, our anal-
ysis included only measures of members' commitment to the
employing organization. Measures of both attitudinal and calcu-
lated (or behavioral) commitment were included. With two ex-
ceptions, all studies in this analysis used a version of the com-
mitment scales developed by Mowday, Steers, and Porter
(1979).
Role perceptions. Many different aspects of role perceptions
have been linked to LMX; however, there was an insufficient
number of studies available to meta-analyze all role perceptions.
We conducted meta-analyses to assess the relationships between
LMX and role conflict and role clarity. The analysis for role
clarity also included studies that measured role ambiguity (re-
verse coded). The most common measure of role perceptions
was a version of the scale developed by Rizzo, House, and
Lirtzman(1970).
Turnover processes. We examined the relationship between
LMX and turnover in two separate analyses. One analysis was
conducted using actual turnover data whereas the second analy-
sis included member self-reports of intentions to leave the
organization.
Member competence. Measures of competence included
both self- and supervisor ratings of general ability or expertise,
cognitive ability tests, and experimental manipulations of com-
petence. Because competence was conceptualized as a more
global measure of ability, we did not use specific performance
ratings in this analysis.
Leader-member agreement. This analysis did not refer spe-
cifically to a correlate of LMX but rather to the agreement
between leader and member regarding LMX quality. All studies
that reported a correlation between leader LMX and member
LMX were included in this analysis.
Coded Variables
In addition to the correlations between LMX and the con-
structs described above, the following information was coded
for each study: (a) date of publication, (b) publication form
(i.e., article, conference paper, dissertation, book chapter, or
unpublished manuscript), (c) type of sample (lab versus field,
type of field sample), (d) sample size, (e) demographic infor-
mation regarding leaders and members, ( f ) LMX measurement
instrument, (g) LMX measurement perspective (i.e., leader or
member), (h) scale properties of LMX measure (i.e., reliability
and range), and (i) measurement information for criterion vari-
ables (i.e., measurement instrument and reliability). All studies
were coded independently by two raters (Charlotte R. Gerstner
and a psychology doctoral student). A total of 1,075 coding
decisions were made, with 86% agreement between the two
raters. All disagreements were resolved through discussion with
David V. Day.
Computation and Analysis of Effect Sizes
We used correlation coefficients in the computation of all effect
sizes. If a correlation coefficient could not be calculated from the
reported statistics, the study was dropped from further analyses.
In order to preserve the independence of samples, we included
only one effect size from each sample in each meta-analysis. We
computed estimates of overall effect sizes and homogeneity of
effect sizes according to the formulas set forth by Hedges and
Olkin (1985). Before combining effect sizes, we converted rs to
ds and corrected for sampling error. We then averaged the rfs to
obtain an estimate of the overall effect size, d+.
When measurement reliability information was available for
at least 75% of the effect sizes in a given analysis, we also
corrected the corresponding sample-weighted correlations for
measurement error. Although the application of corrections is
traditionally associated with the Hunter and Schmidt (1990)
approach to meta-analysis, Hedges and Olkin (1985) stated that
this is an acceptable procedure provided that sufficient reliabil-
ity information is available. In such instances, a closer estimate
of the true population correlation can be attained by applying
correction formulas to the study-level correlations, given that
statistical artifacts such as measurement error can attenuate
wilhin-study correlations (Huffcutt, Arthur, & Bennet, 1993;
Hunter & Schmidt, 1990). In the present meta-analyses, we
 LEADER-MEMBER EXCHANGE META-ANALYSIS
calculated summary effect sizes with both uncorrected and cor-
rected correlations.
In addition to the estimates of overall effect size, we also
calculated a homogeneity statistic (Q) for each analysis. A sig-
nificant Q statistic indicates that the overall mean effect size
does not adequately describe all the effect sizes and further
moderator analyses are warranted. Because the formulas for
homogeneity of effect sizes are inappropriate for disattenuated
effect sizes, we did not interpret the Q statistics associated with
the correlations corrected for measurement unreliability. We
used the corrected correlations only to estimate a mean weighted
effect size.
Tt is fairly common for overall effect sizes to be heterogeneous
(Hedges, 1987). Although Hunter and Schmidt (1990) argued
that nearly all heterogeneity can be attributed to statistical arti-
facts, heterogeneity may also result from other factors, such as
between-study moderators and statistical outliers. Some re-
searchers have advocated the removal of statistical outliers in
an attempt to achieve homogeneity of effect sizes, as well as a
more reliable estimate of the true population effect size (Hedges,
1987; Hedges & Olkin, 1985; Huffcutt & Arthur, 1995). In the
present meta-analyses, we conducted both categorical moderator
analyses and statistical outlier analyses to examine potential
sources of heterogeneity.
On the basis of our review of the literature, we identified two
a priori categorical moderators of overall effect sizes: (a) the
LMX measurement instrument (i.e., LMX7 vs. all others), and
(b) LMX perspective (i.e., leader vs. member), which we used
to examine the impact of these two variables on the homogeneity
of the overall effect sizes. These analyses are analogous to a
one-way analysis of variance that compares effect sizes by a
specified class variable determined on the basis of study charac-
teristics. If a categorical moderator model completely fits the
data, the between-class effect will be significant whereas the
within-class effect will be nonsignificant (indicating homogene-
ity within classes; Johnson & Turco, 1992).
In addition to the categorical moderator analyses, we con-
ducted outlier analyses to attain homogeneity of effect sizes.
This was accomplished by ordering the effect sizes in each
meta-analysis with regard to their deviations from the mean
effect size and then sequentially eliminating the largest outlier
until the overall Q statistic was nonsignificant (indicating homo-
geneity of effect sizes), or until 20% of the original studies
were removed (see Hedges & Olkin, 1985, pp. 256-257). On
the basis of the work of many statisticians, Hedges (1987)
suggested a 20% cutoff as a rule of thumb for concluding
whether the heterogeneity of effect sizes could be attributed to
a few aberrant values. According to Hedges, when homogeneity
can be achieved by removing fewer than 20% of the outliers,
the overall effect size that is calculated from the remaining
studies may better represent the population distribution of ef-
fects, as compared with the mean effect size based on all studies.
Results and Discussion
A summary of the relationships included in our study
is reported in Table 1.4 We calculated overall effect sizes
for the 10 LMX correlates, the relationship between leader
and member reports of LMX, and the internal consistency
831
estimates (i.e., coefficient alpha) of leader and member
LMX reports. Overall effect sizes for all relationships are
summarized and reported in Table 2. Moderator and out-
lier analyses are summarized in Tables 3-5. The results
are organized into the following subsections: (a) reliabil-
ity analysis, (b) leader-member agreement, and (c) LMX
correlates.
Reliability
We analyzed the internal consistency (i.e., Cronbach's
alpha) reliabilities of member LMX and leader LMX. For
the members, the mean sample-weighted alpha was .85,
with the average number of scale items estimated at 7.57.
This estimated effect size was significantly heterogeneous
and could not be rendered homogeneous with the removal
of fewer than 20% of the effects. Thus, there appears to
be sufficient variability in effects to pursue an analysis
of potential moderators. The categorical moderator analy-
sis for measurement instrument indicated a generally
higher alpha for the LMX7 category (.89) than for the
mean of all other instruments (.83). Overall, however,
these results indicate generally acceptable internal consis-
tency estimates from the members' perspective regardless
of the type of measurement instrument.
For leaders' LMX reports, the average number of items
was slightly greater (M = 8.36) than the average item
number for members, although the average sample-
weighted alpha was smaller (.77). As with the member
reports, this effect was heterogeneous and could not be
explained by outliers. Unlike the member results, however,
there was no statistically significant difference between
LMX7 internal consistency (.78) and the alpha estimated
for the category of other measures (.76). Overall, results
of these analyses suggest somewhat better reliability for
members' (as compared with leaders') reports of LMX.
Potential implications of this finding are examined along
with recommendations for better LMX measurement in
the final section, which addresses conclusions and future
directions.
Leader-Member Agreement
The mean sample-weighted correlation between leader
and member reports of LMX was .29 (.37 corrected for
leader and member LMX unreliability). The moderating
effect of measurement instrument was not supported for
this relationship. It was necessary to remove nine outliers
(38%) before a homogeneous effect size was obtained.
Although the overall effect size was significant, the mag-
4
A detailed summary table describing individual study char-
acteristics is available on request from Charlotte R. Gerstner.
832 GERSTNER AND DAY

Table 1
Number of Samples, Aggregate Sample Sizes, and Reliability Estimates far All Relationships

Samples Aggregate sample size

Relationship (*) (AO LMX a Correlate a

Correlate
Performance ratings (leader LMX)' 12 1,909 .76 .83
Performance ratings (member LMX)a 30 4,218 .85 .89
Objective performance 8 982 .80
Satisfaction with supervision 27 5,302 .82 .88
Overall satisfaction 33 6,887 .84 .81
Organizational commitment 17 3,006 .84 .87
Role conflict 12 3,728 .86 .80
Role clarity 14 4,105 .86 .74
Turnover 7 856 .81

Turnover intentions 8 1,074 .82 .85
Member competence 15 3,880 .84 .75
Construct issues
Leader -member agreement 24 3,460 .84b .77
Leader LMX reliability 22 3,329

Member LMX reliability 69 13,885

Note. Dashes indicate that the statistic could not be computed, k number of studies that included the
relationship of interest; A' = total number of individuals across the k samples; a ~ average reliability
coefficient; LMX = leader-member exchange.
" In both cases, the performance measure was a supervisor's rating of the member's performance. The
difference between me two analyses is the source of the LMX measure. * Member LMX reliability.
" Leader LMX reliability.

nitude was relatively small for a relationship that is pur-
portedly dyadic.
LMX Correlates
Overall effect sizes for the analyses between LMX and
its correlates are summarized in Table 2. Moderator and
outlier analyses are summarized in Tables 3-5.
they are somewhat unique compared with the other analy-
ses, the results for the performance ratings analysis is
described in more detail in the following paragraphs.
Performance ratings. We conducted two
analyses to assess the relationship between LMX and
member performance ratings, one for member-reported
LMX and one for leader-reported LMX. It was necessary
to do separate analyses for this relationship because many
of the studies included both leader and member reports
of LMX. Therefore, to preserve sample independence, we
included only one effect size from each study in each
analysis. (In both cases, however, the performance mea-
sure was a supervisor's rating of the member's perfor-
mance.) The mean sample-weighted correlation between
leader reports of LMX and member performance ratings
was .41 (.55 corrected for LMX unreliability).
The moderator analysis for measurement instrument re-
sulted in a significant between-class effect for measure-
ment instrument. The within-class effect sizes for both
classes were also heterogeneous, indicating that unex-
plained variance remained between study outcomes after
accounting for the measurement instrument. It would be
necessary to remove four (33%) of the effect sizes to
achieve homogeneity.
The mean sample-weighted correlation between mem-
ber reports of LMX and member performance ratings was
.28 (.30 corrected for LMX unreliability). These results
demonstrate that member LMX perceptions are also cor-
related with leader performance ratings, although the rela-
Because
tionship is not as strong as the correlation between leader
LMX and performance ratings. It should be noted that the
leader LMX-performance correlation may be confounded
by same-source bias and the resulting effect size estimate
separate
may be artificially inflated. However, this bias is poten-
tially of theoretical interest. Supervisory ratings may be
influenced by many factors, but the perceptions of a super-
visor regarding an employee's performance become crite-
ria on which important decisions are made. A tendency
for a supervisor to rate someone favorably as a result of
a positive LMX relationship can translate into favorable
outcomes for the employee. Furthermore, creating positive
or negative expectations about an employee through the
development of LMX perceptions may change the actual
performance level of employees (i.e., self-fulfilling proph-
ecy) and not just affect performance ratings (Feldman,
1986). Although the correlation between leader LMX and
ratings of member performance was high, our results also
indicate they are empirically distinct constructs.
The moderating effect of measurement instrument for
the correlations between member LMX and corresponding
performance ratings was very similar to that estimated
 LEADER-MEMBER EXCHANGE META-ANALYSIS 833

Table 2
Summary of Effect Sizes

95% CI d+

Relationship Lower Upper Corrected r

Correlate
Performance ratings (leader LMX) 0.91 0.84 0.98 .41 .55'
Performance ratings (member LMX) 0.58 0.54 0.63 .28 .30*
Objective performance 0.19 0.11 0.28 .10 .11"
Satisfaction with supervision 1.59 1.55 1.64 .62 71'
Overall satisfaction 1.03 1.00 1.07 .46 .50"
Organizational commitment 0.75 0.70 0.80 .35 .42"
Role conflict -0.53 -0.58 -0.49 -.26 -.31"
Role clarity 0.73 0.69 0.78 .34 .43'
Turnover -0.07 -0.16 0.03 -.03 -.04"
Turnover intentions -0.58 -0.67 -0.50 -.28 -.31"
Member competence 0.53 0.48 0.57 .26 .28"
Construct issues
Leader-member agreement .62 .57 .67 .29 .37"
Leader LMX reliability 2.40 2.33 2.46 .77

Member LMX reliability 3.25 3.21 3.29 .85

Note. Significant effect sizes are indicated by confidence intervals that do not include 0. Dashes indicate
that the statistic could not be computed. d+ - mean sample-weighted effect size; CI = confidence interval;
r = mean sample-weighted correlation; Corrected r = mean weighted correlation corrected for measurement
unreliability; LMX = leader-member exchange.
" Correlations were corrected for unreliability of both LMX and criterion measures. b Correlations were
corrected for only unreliability of LMX measure.

from the leader's perspective. Although there was a sig- analysis using the two categories of member and leader
nificant between-class difference, the effect sizes within LMX. In order to do this, it was necessary to combine
each of the measurement classes remained heterogeneous. the two analyses for leader and member LMX and perfor-
With regard to outliers, it was necessary to remove six mance ratings. Although this violates the assumption of
studies (20%) to achieve homogeneity of effect sizes. independence of effect sizes, some researchers have ar-
When these studies were excluded, the mean sample- gued that this is an acceptable procedure when testing
weighted correlation based on 22 studies was .27 (p < moderators of theoretical interest (e.g., Rosenthal, 1991).
.01), thus suggesting that the outliers were symmetrically We calculated an overall effect size (r = .32) and exam-
distributed around the mean. ined perspective as a categorical moderator. This analysis
We adopted two different approaches to test whether showed a significant between-class effect for perspective,
the difference between correlations of LMX with perfor- with a significantly larger correlation between leader
mance varied significantly as a function of measurement LMX and performance ratings (r = .41) than between
perspective. First, we conducted a categorical moderator member LMX and performance ratings (r = .28). How-
ever, this model did not completely fit the data, as indi-
cated by the significant within-class variation. Although
Table 3 there may be consistent differences in effect sizes attribut-
Categorical Moderator Models for Perspective able to the perspective from which LMX is measured,
there are probably other moderating influences that con-
Variable and class Qb k
tribute to heterogeneous effect sizes within perspective'.
Performance ratings (combined) 55.10** 42 .32 We also examined the impact of perspective by testing
Member LMX 30 .28 128.52** the difference between leader LMX-performance and
Leader LMX 12 .41 343.80**
Member competence 9.03* 14 .25 member LMX-performance correlations. We conducted a
Member LMX 8 .23 37.68** /-test for dependent correlations using sample size and
Leader LMX 6 .31 266.92** the correlations between (a) leader LMX and perfor-
Note. Significance of Q statistics indicates rejection of the hypothesis mance, (b) member LMX and performance, and (c)
of homogeneity, k = number of studies in each analysis; r = mean leader and member LMX (see Cohen & Cohen, 1983, p.
sample-weighted correlation; Q, = between-classes goodness-of-fit sta- 57). With the overall sample of studies (i.e., meta-analytic
tistic; Qv = homogeneity of effect sizes within each class; LMX =
leader-member exchange. correlations; N = 1,909), these correlations were .41, .28,
*p < .01. and .29, respectively. For the subset of studies (k = 9; N
834 GERSTNER AND DAY

Table 4
Categorical Moderator Models for Measurement Instrument

Variable and class

Performance ratings (LLMX) 19.49** 12 .41

LMX7 6 .49 15.10*
Other 6 .37 293.60**
Performance ratings (MLMX) 4.63* 30 .28
LMX7 12 .31 74.36**
Other 18 .26 70.57**
Objective performance 0.60 8 .10
LMX7 4 .11 11.97*
Other 4 .07 3.23
Satisfaction with supervision 202.87** 27 .62
LMX7 8 .74 154.07**
Other 19 .58 753.43**
Overall satisfaction 12.94** 33 .46
LMX7 11 .50 124.27**
Other 22 .45 203.48**
Organizational commitment 5.87* 17 .35
LMX7 9 .38 37.51**
Other 8 .32 62.65**
Role conflict 29.05** 12 -.26
LMX7 4 -.40 14.28*
Other 8 -.24 40.80**
Role clarity 13.70** 14 .34
LMX7 4 .44 76.71**
Other 10 .33 79.78**
Turnover intentions 2.31 8 -.28
LMX7 6 -.27 11.43
Other 2 -.37 1.21
Leader-member agreement 0.17 24 .29
LMX7 11 .30 89.68**
Other 13 .29 104.26**
Member competence 1.85 15 .26
LMX7 7 .23 59.26**
Other 8 .26 259.35**
Leader LMX reliability 1.63 22 .77
LMX7 10 .78 120.91**
Other 12 .76 721.45**
Member LMX reliability 392.91** 69 .85
LMX7 28 .89 476.98**
Other 41 .83 1,412.04**

Note. Significance of Q statistics indicates rejection of the hypothesis of homogeneity, k = number of

studies in each analysis; r = mean sample-weighted correlation; gb = between-classes goodness-of-fit
statistics; Q. = homogeneity of effect sizes within each class; LLMX = leader-reported LMX; LMX7 =
seven-item LMX measure; MLMX = member-reported LMX; LMX = leader-member exchange.
*p<.05. **p<.01.

= 1,551) that included all three necessary correlations,
the correlations were .36, .24, and .24, respectively. We
tested both sets of correlations as a means of determining
whether a different conclusion would be reached using
the overall estimates as compared with those calculated
from the subset. For the overall analysis, the difference
in correlations was significant, t (1,909) = 5.16,p < .01;
the difference in correlations for the subset of studies
that reported all three necessary correlations was also
significant, ( (1,551) = 3.61, p < .01. Thus, it can be
concluded that correlations between LMX and perfor-
mance differ significantly depending on whether LMX is
measured from the leader's or member's perspective. The
association is stronger when measured from the leader's
point of view.
In summary, it seems that LMX is positively related to
performance ratings. This is consistent with recent work
by Borman and colleagues demonstrating that the inclu-
sion of interpersonal factors can increase the variance
explained in performance ratings by as much as two times
that explained by ability, job knowledge, and technical
factors alone (Borman, White, & Dorsey, 1995). LMX
is one example of such an interpersonal factor. However,
the relationship between LMX and performance ratings
may not be as straightforward as was first proposed. Our
results indicate that the strength of this relationship de-
pends on the perspective from which LMX is measured,
as well as the type of measurement instrument used. Even
after accounting for these moderators, there is consider-
able variability across effect sizes, suggesting the possibil-
 LEADER-MEMBER EXCHANGE META-ANALYSIS 835

Table 5
Summary of Attempts to Reduce Heterogeneity

Support for
measurement Number of Overall r
instrument outliers (after removing
Relationship Q moderator (%K) outliers)

Performance ratings (leader LMX) 873.85** Partial 4 (33%)

Performance ratings (member LMX) 149.57* Partial 6(20%) .27
Objective performance 15.80* None 1 (13%) .07
Satisfaction with supervision 1,110.37** Partial 1 8 (67%)

Overall satisfaction 340.69** Partial 14 (42%)
Organizational commitment 106.03** Partial 4 (24%)
84.13** Partial 4 (33%)

Role conflict
Role clarity 170.19** Partial 6 (43%)

Turnover 31.83** 1 (14%) -.01
Turnover intentions 14.95* None 1 (13%) -.30
Member competence 320.47** None 7 (47%)
Leader-member agreement 194.11** None 9 (38%)

Leader LMX reliability 844.00** None 16 (68%)
Member LMX reliability 2,281.94** Partial 45 (65%)

Note. Q = homogeneity statistic; significance of the Q statistic indicates rejection of the hypothesis of
homogeneity. Partial support for the moderator analyses means that the between-class effect (Qh) was
significant, but the within-class effects (Qw) were also significant. Number of outliers refers to the number
of extreme cases that had to be removed before a homogeneous effect size was achieved. Overall rs after
removing outliers were calculated only for analyses in which 20% or fewer outliers had to be removed to
achieve a homogeneous effect size. Dashes indicate that the statistic could not be computed. LMX = leader-
member exchange.
*p < .05. **p < .01.

ity of additional contingencies. Overall, our results sug-
gest that LMX is related to supervisory performance
evaluations.
Other correlates. We found significant positive corre-
lations between LMX and objective performance, satisfac-
tion with supervision, overall satisfaction, organizational
commitment, and role clarity. We found significant nega-
tive correlations between LMX and role conflict and turn-
over intentions. Overall, the results suggest that having a
high-quality relationship with one's supervisor can affect
the entire work experience in a positive manner, including
performance and affective outcomes. Our results also sug-
gest that LMX is more strongly related to subjective per-
formance ratings and member affective outcomes than to
objective measures, such as productivity and turnover. The
two smallest overall effect sizes were for objective perfor-
mance (r = .10,p < .05) and actual turnover (r = .03,
ns). Although the effect size for objective performance
was statistically significant, its practical meaningfulness
is questionable. Furthermore, both overall effect sizes
could be rendered homogeneous by the removal of one
outlier (objective performance: Graen, Novak, & Som-
merkamp, 1982; actual turnover: Graen, Liden, & Hoel,
1982). When these studies were removed from the analy-
ses, the overall correlations were adjusted to .07 (p <
.05) for objective performance and - .01 (n s) for turnover.
Although there is not enough evidence at present to
support strong correlations between LMX and the objec-
tive outcomes of performance and turnover, we believe
that these relationships are worthy of further examination.
More complex models may be necessary to specify the
nature of the relationships between LMX and objective
outcomes. For example, actual turnover is a complex pro-
cess that depends on numerous variables, such as em-
ployee attitudes and labor market conditions. As Vecchio,
Griffeth, and Horn (1986) suggested, the relationship be-
tween LMX and turnover should not be abandoned but
rather should be examined more closely by searching for
variables that mediate and moderate this process. Given
our results, it is reasonable to propose that LMX affects
turnover through other attitudes, such as satisfaction and
commitment.
There may also be methodological explanations for the
differences in magnitude between subjective and objective
outcome measures. Whereas the challenges inherent in the
study of subjective phenomena (e.g., same-source bias)
may serve to artificially inflate the correlations, those as-
sociated with objective phenomena may serve to attenuate
the correlations. For example, many turnover studies are
plagued by range restriction and low power to detect sig-
nificant relationships. The design of the study may also
impact the outcomes. The Graen, Novak, and Sommer-
kamp (1982) study consisted of a field experiment in
which leaders were trained to maintain high-quality rela-
tionships with their members. In this case, the LMX inter-
vention was very successful at increasing performance
836 GERSTNER AND DAY
(and satisfaction); however, there have been few subse-
quent attempts to replicate this field experiment.5 More
research is necessary to determine whether these effects
will persist after the intervention and whether it is crucial
to have an intervention in order to impact objective perfor-
mance indices.
Summary of Results
Arguably, one of the most provocative findings from
these analyses concerns leader-member agreement on rat-
ings of LMX qualityor lack of it. The average sample-
weighted correlation was estimated to be .29, which is
only slightly higher than the correlation noted between
self- and supervisor performance ratings (mean r = .22),
corrected for measurement error only (Harris & Schau-
broeck, 1988). Our finding regarding leader-member
agreement suggests a couple of directions for future re-
search. First, as suggested by other researchers (e.g.,
Scandura & Schriesheim, 1994), LMX should always
be measured from both leader and member perspectives.
Second, leader-member agreement should be examined
as a relevant independent or dependent variable. Finally,
leader-member agreement should be assessed using lon-
gitudinal designs. Graen and Scandura (1987) argued that
range restriction in leader LMX may attenuate agreement
between perspectives; however, leader reports should be-
come less restricted over time and therefore demonstrate
higher levels of agreement with member reports.
Our results also noted a systematic difference between
leaders' and members' reports of LMX and the correlation
with performance ratings. As stated above, the higher cor-
relation between leader (as compared with member)
LMX and ratings of members' performance may not be
entirely attributable to same-source bias. A leader's ex-
pectations are likely shaped by the quality of LMX that
is developed with a member, which can influence actual
performance levels (Feldman, 1986). More research is
certainly needed, however, to understand the degree to
which same-source bias inflates the correlation between
LMX and performance ratings. An interesting approach
would be to compare the LMX-performance correlations
for dyads in which LMX agreement (i.e., correlation be-
tween leader and member LMX) is high with dyads in
which agreement is low. High agreement should reduce
the difference in correlations between leader and member
LMX reports and performance. In other words, member
LMX correlations with performance may be attenuated
as a result of disagreement with a leader's LMX rating.
There is an additional aspect to our results that should
be considered. All overall effect sizes were heterogeneous,
suggesting that mean correlations do not adequately de-
scribe all corresponding correlations. We conducted mod-
erator and outlier analyses for the reported relationships
in an attempt to reduce the heterogeneity of effect sizes.
Although these analyses did explain some of the variation
in effects, complete homogeneity was achieved for only a
few of the relationships.6 Uhl-Bien et al. (1997) suggested
several situational moderators of LMX-outcome relation-
ships, such as task characteristics, time constraints, re-
sources, physical setting, organizational climate, and cul-
ture. Unfortunately, there were not enough studies that
reported on these variables to examine them as between-
study moderators in our analyses.
Conclusions and Future Directions
The results of the present meta-analyses suggest that
LMX contributes both theoretically and empirically to the
organizational leadership literature. LMX is consistently
correlated with member job performance, satisfaction
(overall and supervisory), commitment, role perceptions,
and turnover intentions. In addition to summarizing the
extant literature, one purpose of the present meta-analysis
was to identify issues that guide further theory develop-
ment and empirical testing. Thus, the final section presents
areas identified for additional theoretical and research at-
tention, as well as our recommendations.
Measurement and Dimensionality
The lack of consistency with which the LMX construct
is operationalized makes it difficult to argue that all mea-
sures are associated with an identical core construct. In
particular, more work is needed to determine whether
LMX is best considered to be multidimensional, or
whether it can be adequately defined with a unidimen-
sional scale. There are two conclusions from the aggregate
reliability results that have relevance to this discussion.
First, we found higher average alphas for the LMX7 mea-
sure as compared with the average reliability for all other
LMX measures. In addition to its higher average alpha,
5
Two notable exceptions are the LMX intervention studies
reported by Scandura and Graen (1984) and Graen, Scandura,
and Graen (1986) in which LMX was significantly related to
objective performance. Bivariate correlations between LMX and
performance could not be calculated from the results reported
in these studies; therefore, it was not possible to include them
in the present analysis.
6
We also conducted an exploratory analysis to examine publi-
cation status as a source of heterogeneity in study results. This
analysis did not result in overall homogeneity; however, we
found that effect sizes were generally larger in unpublished
studies (as compared with those published in refereed journals).
Although we cannot offer a compelling reason for this result,
future research should consider both methodological and theo-
retical moderators of the relationships between LMX and
outcomes.
 LEADER-MEMBER EXCHANGE META-ANALYSIS
studies using the LMX7 measure also tended to obtain
higher correlations with outcomes than those using other
measures (see Table 4). This difference may be partly
due to smaller estimated measurement error for LMX7.
Thus, one implication of these findings is that LMX7
appears to provide the soundest psychometric properties
of all available LMX measures. As such, the LMX7 mea-
sure is recommended to researchers interested in assessing
an overall (i.e., unidimensional) exchange quality.
Second, our results suggest that LMX is more reliably
assessed from a member's perspective than from a leader's
perspective. An implication of this finding is that leaders
may have a somewhat more complex, multidimensional
construction of exchange quality than members, as indi-
cated by the lower overall LMX alpha estimate from a
leader's perspective. Future attempts to revise and develop
alternative LMX measures should examine this notion of
exchange quality separately from leaders' and members'
perspectives. In addition to better measurement properties,
separate leader and member LMX measures might also
be able to improve overall agreement. It is conceivable
that if leaders' LMX is multidimensional whereas mem-
bers' LMX is unidimensional, then correlating the facet
of leader LMX that best matches member LMX would
likely improve resulting agreement correlations.
Agreement
It is unlikely that problems of low agreement between
leaders and members will be completely eliminated with
the development of new measures. Instead, we believe
that leader-member LMX agreement is an interesting and
potentially valuable outcome variable in its own right.
Although they did not explicitly address agreement on
LMX perceptions, Graen and Schiemann (1978) found
that agreement between leaders and members on mutually
experienced events (e.g., member job problems, leader
support to member) varied as a function of LMX quality.
More specifically, agreement with leaders was stronger
for members reporting higher LMX. More research is
needed to understand the complexities of leader-member
agreement.
Interventions directed at improving agreement could be
developed and tested. Fahr and Dobbins (1989) found
that self-evaluations tended to be more strongly correlated
with supervisor ratings when comparative performance
information was made available prior to the self-evalua-
tions. Correspondingly, Mabe and West (1982) found in
a meta-analysis of 55 studies that self-ratings were more
highly correlated with performance when self-evaluation
scale anchors used social comparison terminology such
as "better than average" or "as compared with fellow
workers" than when phrased in absolute terms. Thus,
encouraging both leaders and members to think of ex-
837
change quality in relative terms might enhance rating
agreement Although such agreement may not be expected
to translate directly into performance and affective out-
comes, it could serve to open a dialogue on how to build
a better working relationship, thus providing a catalyst
for developing a stronger partnership, and moving to a
more mature stage of the leadership relationship (Graen &
Uhl-Bien, 1995).
Antecedents of LMX Quality
Researchers have only recently begun to devote atten-
tion to the development of LMX. Therefore, there is very
little cumulative knowledge currently available regarding
the antecedents of LMX. Of those relationships we re-
viewed, only member competence was examined as an
antecedent of LMX in the literature. Although some stud-
ies have examined demographic variables as antecedents
of LMX (e.g., Duchon et al., 1986), there appears to be
little theoretical or empirical justification for the develop-
ment of LMX based on simple demographics.7 Theoreti-
cally, it makes more sense to consider demographic simi-
larity (i.e., relational demography) as an antecedent of
LMX, but empirical support for the role of relational
demography in predicting LMX quality has been mixed
(cf. Bauer & Green, 1996; Green, Anderson, & Shivers,
1996; Liden et al., 1993). More research is needed in
order to clarify the contributions of relational demogra-
phy, as well as other variables that have been examined
as antecedents of LMX, such as leader and member per-
sonality traits, upward influence behavior, leader delega-
tion, and leader-member similarity (see Liden et al.,
1997, for a review of LMX antecedents).
Our results for the moderating effect of perspective
on the relationship between LMX and competence show
preliminary evidence for differential weighting of pre-
dictors for leaders and members. In other words, the ante-
cedents of LMX may be different (or at least differentially
important) for leaders and members. Further examination
of the antecedents of LMX is clearly an area of opportu-
nity for future empirical research.
Development Over Time
Another way to better understand the full range of
LMX, and possibly isolate LMX antecedents, is to care-
fully examine its longitudinal development. According to
7
We computed overall effect sizes for the relationships be-
tween LMX and member education, sex, and age but did not
find meaningful results. This finding is consistent with several
of Graen and colleagues' longitudinal field studies in which
leader and member demographics have never shown consistent
patterns with LMX (G. B. Graen, personal communication,
1997).
838 GERSTNER AND DAY
Graen's (1976) model, the LMX relationship develops as
a result of negotiation between the two parlies, but which
factors affect this negotiation process? It could be that
some people are more likely to maintain high-quality rela-
tionships with their leaders across jobs and situations.
Additional research investigating robust dispositional
characteristics that are theoretically associated with LMX
development (e.g., member conscientiousness) is recom-
mended. There may also be leader characteristics associ-
ated with LMX. For example, leaders who can appreciate
multiple points of view may initiate high-quality relation-
ships with all members, regardless of demographics, per-
sonality, or ability. One proposition is that the more trans-
formational the dyadic relationship, the more it depends
on leader characteristics, whereas transactional relation-
ships depend on both leader and member (i.e., relational)
characteristics, given the inherent exchange process that
is proposed in the latter. These possibilities should be
studied using longitudinal designs.
Much of the original theoretical work on LMX pro-
posed that exchange quality is determined early in the
dyadic relationship and remains relatively stable over
time. This assumption, however, has received little em-
pirical attention. Dansereau et al. (1975) reported fairly
consistent NL means for both in- and out-group members
over a 9-month time period. Three years into the Japa-
nese Career Progress Study, Wakabayashi, Graen, Graen,
and Graen (1988) reported a median LMX stability of
.60, representing the average of all possible test-retest
correlations over six time periods. In a recent longitudi-
nal field study, Liden et al. (1993) studied the develop-
ment of LMX from 2 weeks to 6 months into the leader-
member relationship. Results demonstrated general sta-
bility over time, although the consistency of LMX (for
both leaders and members) decreased with greater time
intervals.
Initial impressions may be difficult to overcome, but
results suggest that LMX can change over the course of
a relationship. With the exception of the Japanese Career
Progress Study, the few existing longitudinal studies have
been conducted over relatively short time spans. Many
working relationships may last several years; therefore, a
measure of LMX taken at 6 months may still be relatively
early in the development of a dyadic relationship. As lead-
ers and members have new and different experiences to-
gether and learn more about each other, it seems likely
that this relationship will evolve and grow. It is unlikely
that the LMX relationship changes in a simple linear fash-
ion. Change patterns may differ according to the perspec-
tive (i.e., leader or member) or the nature of the dyadic
relationship (i.e., transactional or transformational). In-
vestigating models of LMX change is an interesting area
for future research.
Integration With Transformational Leadership
In addition to addressing criticisms regarding the nature
of LMX and its relationships with other variables, a goal
of our synthesis was to clarify the role of LMX in the
larger context of leadership theory and to provide direc-
tion for future research. One of the more popular emerging
frameworks for studying organizational leadership has
been the transactional-transformational distinction pro-
posed by Burns (1978), and further refined by Bass and
his colleagues (Bass, 1985; Bass & Avolio, 1994). Trans-
actional leaders exchange highly valued rewards (e.g.,
promotions, good assignments) for greater quality or
quantity of work, or higher levels of loyalty and commit-
ment from members. They are motivated primarily to sat-
isfy their own self-interests rather than to develop
members or bring about meaningful change. In contrast,
transformational leaders have been described as "self-
defining" (Kuhnert, 1994) in that they are motivated by
higher order values and beliefs and strive to develop fol-
lowers. Transformational leaders are other directed to the
extent that they "continually adjust their behavior to the
level to which the follower has been developed"
(Avolio & Bass, 1995, p. 207), which seems conceptually
similar to the process of developing a unique exchange
relationship that is central to LMX.
If the transactional-transformational framework pro-
posed by Bass and Avolio (1994) encompasses the full
range of leadership behaviors, then where does LMX fit
in the model? Our review supports the suggestion made
by others (Graen & Uhl-Bien, 1995; Scandura & Schries-
heim, 1994) that LMX should incorporate both transac-
tional and transformational processes. We identified sev-
eral factors that support this suggestion, including
measurement issues, the content of LMX training inter-
ventions, and the effects on followers.
LMX is conceptually described as an exchange process,
but it is not usually measured this way (Liden et al., 1997;
Sparrowe & Liden, 1997). Members are seldom asked
what is expected in return for the rewards they are given as
part of a high-quality exchange. Therefore, the exchange
process is inferred but not directly measured. If leaders
do not make explicit demands on members in the form
of harder work or greater commitment for these rewards,
then the relationship might be better characterized as
transformational (Kuhnert, 1994). To the extent that most
LMX measures tap mutual respect, trust, and the overall
quality of the working relationship, LMX is more oriented
toward transformational leadership.
Another indication of the transformational potential of
LMX can be found in the content of a training intervention
designed by Graen, Novak, and Sommerkamp (1982). In
this field experiment, leaders were taught to maintain
high-quality LMX with all members by (a) spending time
 LEADER-MEMBER EXCHANGE META-ANALYSIS
talking about each person's problems, concerns, and ex-
pectations; (b) using "active" listening skills and being
sensitive to the particular issues raised by each member;
(c) refraining from imposing the leader's or manage-
ment's frame of reference on issues discussed; and (d)
sharing some of the leader's expectations about his or her
own job, the member's job, and their working relationship.
These training directives were aimed at producing high-
quality exchanges, a purportedly transactional process;
however, such efforts might be better conceptualized as
transformational in nature in that they encourage leaders
to adopt the viewpoints of their subordinates. The ability
to adopt and appreciate multiple points of viewreferred
to as perspective takingis a recently introduced con-
struct to the transformational leadership literature (Kuhn-
ert, 1994; Kuhnert & Lewis, 1987; Russell & Kuhnert,
1992). To the extent that leaders with high perspective-
taking abilities initiate high LMX relationships with all
members, the predictive power of LMX may expand to
outcomes beyond the individual or dyad level, such as
work group and organizational performance.
In terms of the effects on members, the results of the
present meta-analyses highlight a number of positive out-
comes for members associated with high-quality LMX.
These outcomes (e.g., job performance, organizational
commitment, satisfaction with supervision) are also those
traditionally associated with transformational and charis-
matic leadership (e.g., Bass, 1985; House, 1977). In addi-
tion, LMX has been linked to outcomes associated with
member development, such as increased delegation
(Leana, 1986), empowerment (Keller & Dansereau,
1995), mentoring (Scandura & Schriesheim, 1994), and
career progression (Wakabayashi et al., 1988).
Deluga (1992) explicitly tested the hypothesis that high
LMX was associated with transformational leadership.
He found support for individualized consideration and
charisma as predictors of LMX quality. Basu (1992) also
found a strong positive correlation between LMX and
transformational leadership (r = .87). In addition, Basu
factor analyzed LMX and the four transformational lead-
ership scales and found no support for the distinction
between LMX and transformational leadership. Thus,
there is emerging support for the notion that LMX may
be transformational, at least at certain times and under
certain conditions.
A Comprehensive LMX Model
In order to continue to expand the nomological network
of LMX and demonstrate empirical support for the inte-
grated theoretical model proposed by Uhl-Bien et al.
(1997), additional antecedents and consequences should
be investigated. We have discussed some possible future
research directions in terms of antecedents. Researchers
839
have also begun to study additional consequences of
LMX, including climate perceptions (e.g., Graen et al.,
1973; Kozlowski & Doherty, 1989), organizational citi-
zenship behaviors (e.g., Manogran & Conlon, 1993;
Wayne & Green, 1993), decision influence (e.g., Scand-
ura et al., 1986), leader effectiveness (e.g., Graen et al.,
1973; Deluga & Perry, 1991), member empowerment
(Keller & Dansereau, 1995), and creativity (Tierney,
1992). These outcomes could not be included in the pres-
ent analyses because of an insufficient number of studies
but are important to the continued development of a com-
prehensive LMX model.
Researchers should also continue to examine the role
of organizational context in the LMX model. In their
complete model of relationship-based leadership, Uhl-
Bien et al. (1997) illustrated how situational moderators
can affect both the development of the dyadic leadership
relationship and the linkage between LMX and organiza-
tional outcomes. Finally, much empirical research is
needed to understand how the LMX model operates at
different levels of analysis (Graen'& Uhl-Bien, 1995).
One promising avenue for increasing our understanding
of levels issues is to examine the patterns of LMX and
the relationship between LMX and relevant outcomes in
work groups and teams. Many organizations have adopted
team-based structures, yet we know very little about the
dynamics of leadership relationships within teams and
their impact on team and organizational level outcomes.
Toward a Prescriptive Model of LMX
One of the most interesting challenges for future re-
search on LMX is to move beyond a descriptive leadership
approach to a more prescriptive foundation (Yukl, 1989).
Given that high-quality exchanges are consistently related
to favorable individual outcomes, it may prove beneficial
to devote greater attention to developing and evaluating
LMX training models. Using an organizational case study,
Graen (1989) described a model for conducting leader-
ship training that focuses on the dyadic relationship. How-
ell and Frost (1989) have also shown that it is possible
to train individuals to exhibit aspects of transformational
leadership. Focusing on the development of high-quality
dyadic relationships may be valuable both as an addition
to current models of leadership training and as an alterna-
tive to these methods. More work is needed to specify
the content of such training programs and the boundary
conditions for their effectiveness.
The demonstrated efficacy of Graen, Novak, and Som-
merkamp's (1982) intervention makes that approach
worth considering as a prototype in moving toward a
prescriptive model of LMX. This is an important study
in demonstrating that the underlying perspective-taking
ability addressed in their intervention can be trained. Such
840 GERSTNER AND DAY
efforts are especially important under present economic
pressures to cut organizational costs while maximizing
human-resource potential. In light of the demonstrated
benefits of transformational leadership in organizations
(e.g., Hater & Bass, 1988; Howell & Avolio, 1993), this
is a promising area for future research.
On the basis of our review of the literature, we view
the relationship with one's supervisor as a lens through
which the entire work experience is viewed. LMX is con-
sistently linked to favorable outcomes for members and
should continue to play a major role in models of dyadic
perceptions and individual performance. Building from
these results, we propose that LMX has the potential to
play an even larger role in the context of organization
theory if expanded and refined in several recommended
ways. Construct specification is crucial. Researchers must
agree on the meaning of LMX, the proper unit of measure-
ment (i.e., dyad), and how it should be measured. Addi-
tional longitudinal field research is needed to explore the
dynamics of leader-member relationships over time and
their consequences for individuals, groups, and organiza-
tions. Aspects of transformational leadership should be
formally integrated into LMX so that a comprehensive
model of dyadic leadership may be developed and tested.
Finally, LMX should be incorporated into leadership
training programs and the success of these programs eval-
uated in terms of leader effectiveness, acceptance, and
member development. These suggestions will enhance
both the descriptive and prescriptive value of LMX as a
theory of leadership.
References
References marked with an asterisk indicate studies included in
the meta-analysis.
Avolio, B. J., & Bass, B. M. (1995). Individual consideration
viewed at multiple levels of analysis: A multi-level framework
for examining diffusion of transformational leadership. Lead-
ership Quarterly, 6, 199-218.
Bass, B. M. (1985). Leadership and performance beyond ex-
pectations. New York: Free Press.
Bass.B. M. (1990). Bass and Stogdill's handbook of leadership
(3rd ed.). New >rk: Free Press.
Bass, B. M., & Avolio, B. J. (1994). Improving organizational
effectiveness through transformational leadership. Thousand
Oaks, CA: Sage.
*Basu, R. (1992). An empirical examination of LMX and trans-
formational leadership as predictors of innovative behavior
(Doctoral dissertation, Purdue University, 1991). Disserta-
tion Abstracts International, 52, 2980.
*Bauer, T. N., & Green, S.G. (1996). The development of
leader-member exchange: A longitudinal test. Academy of
Management Journal, 39, 1538-1567.
*Baugh, G. (1992). Interpersonal influences in a project organi-
zation: A role set analysis (Doctoral dissertation, University
of Cincinnati, 1992). Dissertation Abstracts International,
53, 3589.
*Blau, G. (1988). An investigation of the apprenticeship organi-
zational socialization strategy. Journal of Vocational Behav-
ior, 32, 176-195.
Borman, W. C., White, L. A., & Dorsey, D. W. (1995). Effects
of ratee task performance and interpersonal factors on super-
visor and peer performance ratings. Journal of Applied Psy-
chology, 80, 168-177.
*Bumpus, M. A., & Ravlin, E. C. (1994). The roles of similarity
and task competence in the development of LMX relation-
ships. Manuscript submitted for publication.
Burns, J. M. (1978). Leadership. New 'Vbrk: Free Press.
"Carter, M. L., & Foti, R. J. (1994, April). Manager-subordi-
nate exchange relationships: Investigation of a manager be-
havior model. Paper presented at the 9th annual meeting of
the Society for Industrial and Organizational Psychology,
Nashville, TN.
*Castleberry, S. B., & Tanner, J. F. (1989). Salesperson's com-
mitment to the organization: Associations with performance,
motivation, conflict, satisfaction, and relationship with the
manager. Journal of Applied Business Research, 5, 8489.
Cohen, I., & Cohen, P. (1983). Applied multiple regression/
correlation analysis for the behavioral sciences (2nd ed.).
Hillsdale, NJ: Erlbaum.
Cortina, J. M. (1993). What is coefficient alpha? An examina-
tion of theory and applications. Journal of Applied Psychol-
ogy, 78, 98-104.
Dansereau, E, Cashman, J., & Graen, G. (1973). Instrumental-
ity theory and equity theory as complementary approaches in
predicting the relationship of leadership and turnover among
managers. Organizational Behavior and Human Perfor-
mance, 10, 184-200.
Dansereau, E, Graen, G., & Haga, W. J. (1975). A vertical dyad
linkage approach to leadership within formal organizations:
A longitudinal investigation of the role making process. Orga-
nizational Behavior and Human Performance, 13, 46-78.
*Day, D. V., & Grain, E. (1992). The role of affect and ability in
initial exchange quality perceptions. Group and Organization
Management, 17, 380-397.
*Deluga, R. J. (1992). The relationship of leader-member ex-
change with laissez-faire, transactional, and transformational
leadership in naval environments. In K. E. Clark, M. B.
Clark, & D. P. Campbell (Eds.), Impact of leadership (pp.
237-247). Greensboro, NC: Center for Creative Leadership.
*Deluga, R. J. (1994). Supervisor trust building, leader-mem-
ber exchange, and organizational citizenship behavior. Jour-
nal of Occupational & Organizational Psychology, 67, 315
326.
*Deluga, R. J., & Perry, J. T. (1991). The relationship of subor-
dinate upward influencing behavior, satisfaction, and per-
ceived superior effectiveness with leader-member exchanges.
Journal of Occupational Psychology, 64, 239-252.
*Deluga, R. J., & Perry, J. T. (1994). The role of subordinate
performance and ingratiation in leader-member exchanges.
Group and Organization Management, 19, 67-86.
*Dienesch, R. M. (1985). A test of the dimensionality of
leader-member exchange. Unpublished manuscript.
*Dienesch, R. M. (1988). An empirical investigation of the
relationship between quality of leader-member exchange and
subordinate performance and satisfaction (Doctoral disserta-
 LEADER-MEMBER EXCHANGE META-ANALYSIS
tion, Georgia Institute of Technology, 1987). Dissertation
Abstracts International, 48, 2922.
Dienesch, R. M., & Liden, R. C. (1986). Leader-member ex-
change model of leadership: A critique and further develop-
ment. Academy of Management Review, 11, 618634.
*Dobbins, G. H., Cardy, R. L., & Platz-Vieno, S. J. (1990).
A contingency approach to appraisal satisfaction: An initial
investigation of the joint effects of organizational variables
and appraisal characteristics. Journal of Management, 16,
619-632.
*Dockery, T. M., & Steiner, D. D. (1990). The role of the initial
interaction in leader-member exchange. Group and Organi-
zation Studies, 15, 395-413.
*Duarte, N. T, Goodson, J. R., & Klich, N. R. (1994). Effects
of dyadic quality and duration on performance appraisal.
Academy of Management Journal, 37, 499-521.
*Duchon, D., Green, S. G., &Taber, T. D. (1986). Vertical dyad
linkage: A longitudinal assessment of antecedents, measures,
and consequences. Journal of Applied Psychology, 71, 56-
60.
*Dunegan, K. J., Duchon, D., & Uhl-Bien, M. (1992). Examin-
ing the link between leader-member exchange and subordi-
nate performance: The role of task analyzability and variety
as moderators. Journal of Management, 18, 5976.
*Dunegan, K. J., Uhl-Bien, M., & Duchon, D. (1994, August).
Task-level climate (TLC) and leader-member exchange
(LMX) as interactive predictors of subordinate performance.
Paper presented at the 54th Academy of Management meeting,
Dallas, TX.
Fahr, J.-L., & Dobbins, G. H. (1989). Effects of comparative
performance information on the accuracy of self-ratings and
agreement between self- and supervisor ratings. Journal of
Applied Psychology, 74, 606-610.
Feldman, J. M. (1986). A note on the statistical correction of
halo error. Journal of Applied Psychology, 71, 173-176.
* Ferris, G. R. (1985). Role of leadership in the employee with-
drawal process: A constructive replication. Journal of Applied
Psychology, 70, 777-781.
*Rikami, C. V, & Larson, E. W. (1984). Commitment to com-
pany and union: Parallel models. Journal of Applied Psychol-
ogy, 69, 367-371.
*Gast, I. (1987). Leader cognitive complexity and its effects
on the quality of exchange relationships with subordinates
(Doctoral dissertation, George Washington University, 1987).
Dissertation Abstracts International, 47, 5082.
*Gerras, S. (1993). The effect of cognitive busyness and non-
verbal behaviors on trait inferences and LMX judgments
(Doctoral dissertation, The Pennsylvania State University,
1992). Dissertation Abstracts International, 53, 3819.
*Gessner, J. (1993). An interpersonal attraction approach to
leader-member exchange: Predicting the predictor (Doctoral
dissertation, University of Maryland, 1993). Dissertation Ab-
stracts International, 53, 3820.
Graen, G. B. (1976). Role-making processes within complex
organizations. In M. D. Dunnette (Ed.), Handbook of indus-
trial and organizational psychology (pp. 1201-1245). Chi-
cago: Rand McNally.
Graen, G. B. (1989). Unwritten rules for your career. New
York: Wiley.
841
*Graen, G. B., & Cashman, J. (1975). A role-making model of
leadership in formal organizations: A development approach.
In J. G. Hunt & L. L. Larson (Eds.), Leadership frontiers
(pp. 143-165). Kent, OH: Kent State University.
*Graen, G. B., Dansereau, E, Minanii, T., & Cashman, J.
(1973). Leadership behaviors as cues to performance evalua-
tion. Academy of Management Journal, 16, 611-623.
*Graen,G. B., Liden, R., &Hoel,W. (1982). Role of leadership
in the employee withdrawal process. Journal of Applied Psy-
chology, 67, 868-872.
*Graen, G. B., Novak, M. A., & Sommerkamp, P. (1982). The
effects of leader-member exchange and job design on pro-
ductivity and satisfaction: Testing a dual attachment model.
Organizational Behavior and Human Performance, 30, 109-
131.
Graen, G. B., & Scandura, T. A. (1987). Toward a psychology
of dyadic organizing. Research in Organizational Behavior,
9, 175-208.
Graen, G. B., Scandura, T. A., & Graen, M. R. (1986). A field
experiment test of the moderating effects of growth need
strength on productivity. Journal of Applied Psychology, 71,
484-491.
Graen, G. B., & Schiemann, W. (1978). Leader-member agree-
ment: A vertical dyad linkage approach. Journal of Applied
Psychology, 63, 206-212.
Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based ap-
proach to leadership: Development of leader-member ex-
change (LMX) theory of leadership over 25 years: Applying
a multi-level multi-domain perspective. Leadership Quarterly,
6, 219-247.
*Graen, G. B., Wakabayashi, M., Graen, M. R., & Graen, M. G.
(1990). International generalizability of American hypotheses
about Japanese management progress: A strong inference in-
vestigation. Leadership Quarterly, 1, 1-23.
*Green, S. G., Anderson, S. E., & Shivers, S. L. (1996). Demo-
graphic and organizational influences on leader-member ex-
change and related work attitudes. Organizational Behavior
and Human Decision Processes, 66, 203-214.
*Green, S. G., Blank, W, & Liden, R. C. (1983). Market and
organizational influences on bank employees' work attitudes
and behaviors. Journal of Applied Psychology, 68, 298-306.
Harris, M. M., & Schaubroeck, J. (1988). A meta-analysis of
self-supervisor, self-peer, and peer-supervisor ratings. Per-
sonnel Psychology, 41, 43-62.
Hater, J. J., & Bass, B. M. (1988). Superiors' evaluations and
subordinates' perceptions of transformational and transac-
tional leadership. Journal of Applied Psychology, 73, 695-
702.
Hedges, L. V. (1987). How hard is hard science, how soft is soft
science? The empirical cumulativeness of research. American
Psychologist, 42, 443-455.
Hedges, L. V, & Olkin, I. (1985). Statistical methods for meta-
analysis. New \brk: Academic Press.
House, R. J. (1977). A 1976 theory of charismatic leadership.
In J. G. Hunt & L. L. Larson (Eds.), Leadership: The cut-
ting edge (pp. 189-207). Carbondale: Southern Illinois
University.
Howell, J. M., & Avolio, B. J. (1993). Transformational leader-
ship, transactional leadership, locus of control, and support
842 GERSTNER AND DAY
for innovation: Key predictors of consolidated-business-unit
performance. Journal of Applied Psychology, 78, 891902.
Howell, J. M., & Frost, P. J. (1989). A laboratory study of
charismatic leadership. Organizational Behavior and Human
Decision Processes, 43, 243-269.
Huffcutt, A. I., & Arthur, W., Jr. (1995). Development of a new
outlier statistic for meta-analytic data. Journal of Applied
Psychology, 80, 327-334.
Huffcutt, A. I., Arthur, W., Jr., & Bennet, W. (1993). Conducting
meta-analysis using the proc means procedure in SAS. Educa-
tional and Psychological Measurement, S3, 119-131.
Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analy-
sis: Correcting error and bias in research findings. Newbury
Park, CA: Sage.
Hunter, J. E., Schmidt, F. L., & Jackson, G. B. (1982). Meta-
analysis: Cumulating research findings across studies. Bev-
erly Hills, CA: Sage.
*James, L. R., Gent, M. J., Hater, J. J., & Coray, K. E. (1979).
Correlates of psychological influence: An illustration of the
psychological climate approach to work environment percep-
tions. Personnel Psychology, 32, 563-588.
*James, L. R., Hater, J. J., & Jones, A. (1981). Perceptions of
psychological influence: A cognitive information processing
approach for explaining moderated relationships. Personnel
Psychology, 34, 453-477.
Johnson, B. T., & Tbrco, R. M. (1992). The value of goodness-
of-fit indices in meta-analysis: A comment on Hall and Rosen-
thai. Communication Monographs, 59, 388-396.
* Jones, A. P., Glaman, J. M., & Johnson, D. S. (1993). Percep-
tions of a quality program and relationships with work percep-
tions and job attitudes. Psychological Reports, 72, 619-624.
*Katerberg, R., & Horn, P. W. (1981). Effects of within-group
and between-group variation in leadership. Journal of Applied
Psychology, 66, 218-223.
"Keller, T., & Dansereau, F. (1995). Leadership and empow-
erment: A social exchange perspective. Human Relations, 48,
127-146.
*Kim, K. I., & Organ, D. W. (1982). Determinants of leader-
subordinate exchange relationships. Group and Organization
Studies, 7, 77-89.
*Kinicki, A. J., & Vecchio, R. P. (1994). Influences on the qual-
ity of supervisorsubordinate relations: The role of time-
pressure, organizational commitment, and locus of control.
Journal of Organizational Behavior, 15, 75-82.
*K'Obonyo, P. O. (1989). A dyadic upward influence process:
A laboratory investigation of the effect of a subordinate's
ingratiation (praise & performance) on the superior-subordi-
nate exchange relationship (Doctoral dissertation, University
of South Carolina, 1988). Dissertation Abstracts Interna-
tional, 50, 1366.
Kozlowski, S. W. J., & Doherty, M. L. (1989). Integration of
climate and leadership: Examination of a neglected issue.
Journal of Applied Psychology, 74, 546-553.
Kuhnert, K. W. (1994). Transforming leadership: Developing
people through delegation. In B. M. Bass & B. J. Avolio
(Eds.), Improving organizational effectiveness through trans-
formational leadership (pp. 10-25). Thousand Oaks, CA:
Sage.
Kuhnert, K. W., & Lewis, P. (1987). Transactional and transfor-
mational leadership: A constructive/developmental analysis.
Academy of Management Review, 12, 648657.
*Lagace, R. R. (1988). An investigation into the impact of
interpersonal trust on the quality of the relationship and out-
come variables in the sales manager/salesperson dyad (Doc-
toral dissertation, University of Cincinnati, 1987). Disserta-
tion Abstracts International, 48, 2679.
*Lagace, R. R. (1990). Leader-member exchange: Antecedents
and consequences of the cadre and hired hand. Journal of
Personal Selling and Sales Management, 10, 11-19.
*Lagace, R. R., Castlebeny, S. B., & Ridnour, R. E. (1993).
An exploratory salesforce study of the relationship between
leader-member exchange and motivation, role stress, and
manager evaluation. Journal of Applied Business Research,
9, 110-119.
*Leana, C. R. (1986). Predictors and consequences of delega-
tion. Academy of Management Journal, 2, 754774.
*Liden, R., & Maslyn, J. M. (in press). Multidimensionality of
leader-member exchange: An empirical assessment through
scale development. Journal of Management.
Liden, R. C., Sparrowe, R. T, & Wayne, S. J. (1997). Leader-
member exchange theory: The past and potential for the fu-
ture. In G. Ferris (Ed.), Research in personnel and human
resource management (Vol. 15, pp. 47-119). Greenwich,
CT: JAI Press.
*Liden, R., Wayne, S. J., & Stilwell, D. (1993). A longitudinal
study on the early development of leader-member exchanges.
Journal of Applied Psychology, 78, 662-674.
Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N. (1996).
Effectiveness correlates of transformational and transactional
leadership: A meta-analytic review of the MLQ literature.
Leadership Quarterly, 7, 385-425.
Mabe, P. A., & West, S. G. (1982). Validity of self-evaluation
of ability: A review and meta-analysis. Journal of Applied
Psychology, 67, 280-296.
*Major, D. A., Kozlowski, S. W. J., Chao, G. T., & Gardner,
P. D. (1995). Newcomer expectations and early socialization
outcomes: The moderating effect of role development factors.
Journal of Applied Psychology, 80, 418-431.
Manogran, P., & Conlon, E. J. (1993). A leader-member ex-
change approach to explaining organizational citizenship be-
havior. Proceedings of the annual meeting of the Academy
of Management (pp. 249-253). Atlanta, GA: Academy of
Management.
*Manogran, P., Stauffer, J., & Conlon, E. J. (1994, August).
Leader-member exchange as a key mediating variable be-
tween employees' perceptions of fairness and organizational
citizenship behavior. Paper presented at the 54th Academy of
Management meeting, Dallas, TX.
Mintzberg, H. (1973). The nature of managerial work. New
York: Harper & Row.
Mowday, R. T., Steers, R. M., & Porter, L. W. (1979). The mea-
surement of organizational commitment. Journal of Voca-
tional Behavior, 14, 224-247.
*Murry, W. D. (1993). Leader-member exchange and work
value congruence: A multiple levels approach (Doctoral dis-
sertation, Virginia Polytechnic Institute and State University,
1993). Dissertation Abstracts International, 54, 2225.
*Murry, W. D., & Markham, S. E. (1994, April). Leader-mem-
 LEADER-MEMBER EXCHANGE META-ANALYSIS
her exchange relationships: A social exchange perspective.
Paper presented at the 9th annual meeting of the Society for
Industrial and Organizational Psychology, Nashville, TN.
*Nystrom, P. C. (1990). Vertical exchanges and organizational
commitments of American business managers. Group and
Organization Studies, IS, 296-312.
*Peck, M. (1989). Head nurse fit with staff, administration, and
situation: Impact on unit performance and staff satisfaction
(Doctoral dissertation, The University of Utah, 1987). Disser-
tation Abstracts International, 49, 2570.
*Phillips, A. S., & Bedeian, A. G. (1994). Leader-follower ex-
change quality: The role of personal and interpersonal attri-
butes. Academy of Management Journal, 37, 990-1001.
Rizzo, J. R., House, R. J., & Lirtzman, S. E. (1970). Role con-
flict and ambiguity in complex organizations. Administrative
Science Quarterly, 15, 150-163.
Rosenthal, R. (1991). Meta-analytic procedures for social re-
search. Newbury Park, CA: Sage.
*Rosse, J. G., & Kraut, A. I. (1983). Reconsidering the vertical
dyad linkage model of leadership. Journal of Occupational
Psychology, 56, 63-71.
Russell, C. J., & Kuhnert, K. W. (1992). Integrating skill acqui-
sition and perspective taking capacity in the development of
leaders. Leadership Quarterly, 3, 335-353.
Sackett, P. R., Harris, M. M., & Orr, J. M. (1986). On seeking
moderator variables in the meta-analysis of correlational data:
A Monte Carlo investigation of statistical power and resis-
tance to type I error. Journal of Applied Psychology, 71, 302-
310.
Scandura, T. A., & Graen, G. B. (1984). Moderating effects of
initial leader-member exchange status on the effects of a
leadership intervention. Journal of Applied Psychology, 69,
428-436.
*Scandura, T. A., Graen, G. B., & Novak, M. A. (1986). When
managers decide not to decide autocratically: An investigation
of leader-member exchange and decision influence. Journal
of Applied Psychology, 71, 579-584.
*Scandura, T. A., & Schriesheim, C. A. (1994). Leader-mem-
ber exchange and supervisor career mentoring as complemen-
tary constructs in leadership research. Academy of Manage-
ment Journal, 37, 1588-1602.
*Schiemann, W. A. (1977). The nature and prediction of organi-
zational communication: A review of the literature and an
empirical investigation (Doctoral dissertation, University of
Illinois at Urbana-Champaign, 1977), Dissertation Abstracts
International, 38, 2922.
*Schriesheim, J. F. (1980). The social context of leader-subor-
dinate relations: An investigation of the effects of group cohe-
siveness. Journal of Applied Psychology, 65, 183-194.
*Schriesheim, C. A., Neider, L. L., Scandura, T. A., & Tepper,
B. J. (1992). Development and preliminary validation of a
new scale (LMX-6) to measure leader-member exchange in
organizations. Educational and Psychological Measurement,
52, 135-147.
*Seers, A. (1989). Team-member exchange quality: A new
construct for role-making research. Organizational Behavior
and Human Decision Processes, 43, 118-135.
*Seers, A., & Graen, G. B. (1984). The dual attachment con-
cept: A longitudinal investigation of the combination of task
843
characteristics and leader-member exchange. Organizational
Behavior and Human Performance, 33, 283306.
*Sidhu, K. (1989). The role of a leader-member exchange and
work experience in the early career development of profes-
sionals (Doctoral dissertation, University of Cincinnati,
1988). Dissertation Abstracts International, 49, 2307.
*Snyder, R. A., & Bruning, N. S. (1985). Quality of vertical
dyad linkages: Congruence of supervisor and subordinate
competence and role stress as explanatory variables. Group
and Organization Studies, 10, 81-94.
*Snyder, R. A., Williams, R., & Cashman, J. F. (1984). Age,
tenure, and work perceptions as predictors of reactions to
performance feedback. Journal of Psychology, 116, 1121.
Sparrowe, R. T, & Liden, R. C. (1997). Process and structure
in leader-member exchange. Academy of Management Re-
view, 22, 522-552.
Spector, P. E., & Levine, E. L. (1987). Meta-analysis for inte-
grating study outcomes: A Monte Carlo study of its suscepti-
bility to type I and type II errors. Journal of Applied Psychol-
ogy, 72, 3-9.
*Stepina, L. P., Perrewe, P. L., Hassell, B. L., Harris, J. R., &
Mayfield, C.R.(1991).A comparative test of the independent
effects of interpersonal, task, and reward domains on personal
and organizational outcomes. Journal of Social Behavior and
Personality, 6, 93-104.
Stogdill, R. M. (1948). Personal factors associated with leader-
ship: A survey of the literature. Journal of Psychology, 25,
35-71.
"Tanner, J. F., Jr., & Castleberry, S. B. (1990). Vertical exchange
quality and performance: Studying the role of the sales man-
ager. Journal of Personal Selling and Sales Management, 10,
17-27.
"Tanner, J. F., Dunn, M. G., & Chonko, L. B. (1993). Vertical
exchange and salesperson stress. Journal of Personal Selling
and Sales Management, 13, 2736.
*Tansky, J. (1993). Justice and organizational citizenship be-
havior: What is the relationship? Employee Responsibilities &
Rights Journal, 6, 195-207.
Tierney, P. (1992). The contribution of leadership, supportive
environment, and individual attributes to creative perfor-
mance: A quantitative field study (Doctoral dissertation, Uni-
versity of Cincinnati, 1992). Dissertation Abstracts Interna-
tional, 54, 246.
Uhl-Bien, M., Graen, G. B., & Scandura, T. A. (1997). Rela-
tional leadership theory: A role-making approach. Unpub-
lished manuscript.
*Uhl-Bien, M., Tiemey, P. S., Graen, G. B., & Wakabayashi,
M. (1990). Company paternalism and the hidden-investment
process: Identification of the "right type" for line managers
in leading Japanese organizations. Group and Organization
Studies, 15, 414-430.
*Vecchio, R. P. (1985). Predicting employee turnover from
leader-member exchange: A failure to replicate. Academy of
Management Journal, 28, 478-485.
*Vecchio, R. P. (1987). Situational leadership theory: An exam-
ination of a prescriptive theory. Journal of Applied Psychol-
ogy, 72, 444-451.
*Vecchio, R. P., & Gobdel, B.C. (1984). The vertical dyad
844 GERSTNER AND DAY

linkage model of leadership: Problems and prospects. Organi-
zational Behavior and Human Performance, 34, 5-20.
*Vecchio, R. P., Griffeth, R. W., & Horn, P. W. (1986). The
predictive utility of the vertical dyad linkage approach. Jour-
nal of Social Psychology, 126, 617-625.
Wakabayashi, M., Graen, G., Graen, M., & Graen, M. (1988).
Japanese management progress: Mobility into middle man-
agement. Journal of Applied Psychology, 73, 217-227.
*Wayne, S. J., & Ferris, G. R. (1990). Influence tactics, affect,
and exchange quality in supervisor-subordinate interactions:
A laboratory experiment and field study. Journal of Applied
Psychology, 75, 487-499.
*Wayne, S. J., & Green, S. A. (1993). The effects of leader-
member exchange on employee citizenship and impression
management behavior. Human Relations, 46, 1431-1440.
*Weitzel, J. R., & Graen, G. B. (1990), System development
project effectiveness: Problem-solving competence as a mod-
erator variable. Decision Sciences, 20, 507-531.
*Wilhelm C. C, Herd, A. M., & Steiner, D. D. (1993). Attribu-
tional conflict between managers and subordinates: An inves-
tigation of leader-member exchange effects. Journal of Or-
ganizational Behavior, 14, 531-544.
*Williams, L. J., Gavin, M. B., & Williams, M. L. (1996). Mea-
surement and nonmeasurement processes with negative af-
fectivity and employee attitudes. Journal of Applied Psychol-
ogy, 81, 88-101.
Yukl, G. A. (1989). Leadership in organizations. Englewood
Cliffs, NJ: Prentice Hall.
*Zalesny, M. D., & Kirsch, M. P. (1989). The effect of similar-
ity on performance ratings and interrater agreement. Human
Relations, 42, 81-96.
Received June 25, 1996
Revision received May 29, 1997
Accepted May 30, 1997

AMERICAN PSYCHOLOGICAL ASSOCIATION

SUBSCRIPTION CLAIMS INFORMATION Today's Date:_

We provide this form to assist members, institutions, and nonmember individuals with any subscription problems. With the
appropriate information we can begin a resolution. If you use the services of an agent, please do NOT duplicate claims through
them and directly to us. PLEASE PRINT CLEARLY AND IN INK IF POSSIBLE.

RUNT FULL NAME OR KEY NAME OF INSTITUTION MEMBER OR CU5TOMERNUMBER (MAY BE FOUND ON ANY PAST ISSUE LABEL)

ADDRESS DATE YOUR ORDER WAS MAILED (OR PHONED)

PREPAID . _CHECK __CHAROE

CHECK/CARD CLEARED DATE:
CITY STATE/COUNTRY
(If possible, send a copy, front and back, of your cancelled check to help us in our research
of your claim.)
YOUR NAME AND PHONE NUMBER ISSUES: MISSING DAMAGED

TITLE VOLUME OR YEAK NUMBER OR MONTH

Thank you. Once a claim w received and resolved, delivery of replacement issues routinely takes 46 weeks.
(TO BE FILLED OUT BY APA STAFF) ^
DATE RECEIVED:. DATE OF ACTION: _
ACTION TAKEN: _ INV. NO. & DATE:
STAFF NAME: LABEL NO. & DATE:

Send this form to APA Subscription Claims, 750 First Street, NE, Washington, DC 20002-4242

PLEASE DO NOT REMOVE. A PHOTOCOPY MAY BE USED.