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

net/publication/8881160

The Forgotten Ones? The Validity of

Consideration and Initiating Structure in
Leadership Research

Article in Journal of Applied Psychology · March 2004

DOI: 10.1037/0021-9010.89.1.36 · Source: PubMed

CITATIONS READS

521 5,990

3 authors, including:

Ronald F Piccolo Remus Ilies

University of Central Florida National University of Singapore
33 PUBLICATIONS 3,779 CITATIONS 93 PUBLICATIONS 7,404 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Remus Ilies on 10 January 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Journal of Applied Psychology Copyright 2004 by the American Psychological Association, Inc.
2004, Vol. 89, No. 1, 36 –51 0021-9010/04/$12.00 DOI: 10.1037/0021-9010.89.1.36

The Forgotten Ones? The Validity of Consideration and Initiating Structure

in Leadership Research

Timothy A. Judge, Ronald F. Piccolo, and Remus Ilies

University of Florida

This study provided a meta-analysis of the relationship of the Ohio State leadership behaviors—
Consideration and Initiating Structure—with leadership. Overall, 163 independent correlations for
Consideration and 159 correlations for Initiating Structure were analyzed. Results revealed that both
Consideration (.48) and Initiating Structure (.29) have moderately strong, nonzero relations with
leadership outcomes. Consideration was more strongly related to follower satisfaction (leader satisfac-
tion, job satisfaction), motivation, and leader effectiveness, and Initiating Structure was slightly more
strongly related to leader job performance and group– organization performance. Validities did vary by
leadership measure, but in most cases validities generalized regardless of the measure used. Overall, the
results provide important support for the validity of Initiating Structure and Consideration in leadership
research.

The 1940s was a period of great importance to the field of generality of the validities and the nature of the measures them-
leadership. Disappointed by the yield from studies investigating selves; many may feel that these questions were never answered
the trait theory of leadership, a group of researchers at Ohio State satisfactorily. Fleishman (1995), for example, argued that the
University—led by R. Stogdill, C. Shartle, and J. Hemphill— validities of Consideration and Initiating Structure are curvilinear,
sought to uncover the behavioral indicators of effective leadership such that there are diminishing returns to the increased use of
(see Stogdill, 1950). Although at various periods many behaviors consideration and structure on the part of the leader. However,
were studied, the Ohio State studies isolated two factors: Consid- with the exception of the highly cited study by Fleishman and
eration and Initiating Structure (or Structure). Consideration is the Harris (1962), this hypothesis has not been tested further (Fleish-
degree to which a leader shows concern and respect for followers, man, 1998). Similarly, research was often criticized for its reliance
looks out for their welfare, and expresses appreciation and support on common source data in which the leadership behavior ratings
(Bass, 1990). Initiating Structure is the degree to which a leader and criteria were collected from the same source (Kerr &
defines and organizes his role and the roles of followers, is Schriesheim, 1974), although in fairness, many subsequent studies
oriented toward goal attainment, and establishes well-defined pat- did use independent data sources (e.g., Ilgen & Fuji, 1976; Sheri-
terns and channels of communication (Fleishman, 1973). Until the dan & Vredenburgh, 1978b). Another controversy is how Consid-
advent of transformational leadership theory beginning in the late eration and Structure should be measured. Several measures of the
1970s (Bass, 1985; Burns, 1978; House, 1977), these two dimen- constructs have been developed, each has been criticized on var-
sions dominated leadership research. As Fleishman (1995) noted, ious grounds (Schriesheim & Kerr, 1974), and yet each seems to
“Consideration and Initiating Structure have proven to be among have its adherents. As we note later, whether these measures in fact
the most robust of leadership concepts” (p. 51).1 assess the same construct, and which measure is superior, is not
In the more than half century since the discovery of Consider-
ation and Initiating Structure, much has been learned about these
1
concepts. At the same time, upon reflecting on this literature, one Although we make primary reference to the work of the Ohio State
cannot help but be impressed by the mysteries surrounding Con- leadership studies, we note that researchers at the University of Michigan
carried out a relatively concurrent and in some ways comparable leadership
sideration and Structure, as well as how quickly they fell out of
research program (e.g., Likert, 1961). Although the contributions of the
favor in leadership research. Questions were raised about the Michigan studies should be acknowledged, the Michigan dimensions of
employee-centered leadership and production-centered leadership were
much less frequently studied in empirical research. Perhaps the most focal
Timothy A. Judge, Ronald F. Piccolo, and Remus Ilies, Department of contribution of the Michigan studies was the Blake and Mouton managerial
Management, Warrington College of Business, University of Florida. grid (Blake, Mouton, & Bidwell, 1962), which has been criticized in the
Remus Ilies is now at the Department of Management, Michigan State literature on various grounds (Bernardin & Alvares, 1976). Similarly, at
University. roughly the same time, Bales and associates at Harvard studied person- and
We thank Ed Fleishman and Henry Tosi for their assistance in locating task-oriented behaviors (Bales, 1954) although again, these contributions
some of the studies included in the meta-analysis. have been less recognized in the literature. As noted by Chemers (1997),
Correspondence concerning this article should be addressed to Timothy “The most extensive program of research on the relative effectiveness of
A. Judge, Department of Management, Warrington College of Business, leadership behavior patterns was carried out by the very productive group
University of Florida, Gainesville, Florida 32611. E-mail: timothy of researchers at Ohio State” (p. 23). In accordance, we focused on the
[email protected] Ohio State dimensions of Consideration and Initiating Structure.

36
 CONSIDERATION AND INITIATING STRUCTURE 37

universally accepted (see Bass, 1990, and Fleishman, 1995, for Structure] varies from situation to situation” (p. 157). Kerr and
reviews). Schriesheim (1974) documented the situational moderators that
On a conceptual level, House’s (1977) theory of charismatic had been studied at that time, which include subordinate factors
leadership shifted the focus of leadership research to higher end (e.g., follower job knowledge), supervisor factors (e.g., upward
constructs. Further, with the emergence of implicit leadership influence), and task factors (e.g., autonomy). Given the assumed
theories, researchers started to question the internal validity of lack of support for the main effects of Consideration and Structure,
behavioral leadership survey measures in general, and measures of in the 1970s many researchers turned to the investigation of
Consideration and Structure were no exception (Rush, Thomas, & situational factors that might moderate their effects. One of the
Lord, 1977). In addition, from an implicit theory perspective, the most popular theories of leadership in the 1970s and 1980s—path–
shift toward higher end leadership constructs perhaps has made goal theory (House, 1971)—was based on the premise that factors
researchers realize that constructs such as charisma and vision are moderate and mediate the effect of Consideration and Structure on
more central to their own implicit theories of leadership than outcomes.
constructs such as Consideration and Structure, and this realization Although we do not deny that situational moderators may exist
was then reflected in the leadership constructs and processes and may prove productive in future research, we would note two
investigated in their research. points. First, theories that were explicitly developed to test mod-
In sum, research on Consideration and Structure has been crit- erators of Consideration and Initiating Structure have not received
icized on both methodological and conceptual grounds. Perhaps as clear support in the literature. For example, support for path– goal
a result of these criticisms, Consideration and Initiating Structure theory (House, 1971) was labeled as “mixed” by the creator of the
are widely believed to be of limited validity and little direct utility theory himself (House & Podsakoff, 1994, p. 52); Wofford and
in contemporary leadership thinking and research. To be sure, Liska (1993), on the basis of a meta-analytic review, described the
these constructs are recognized as an important part of the lineage support as “inconsistent” and “marginal.” Similarly, support for
of present day leadership research (House & Podsakoff, 1994). the “neutralizers” in Kerr and Jermier’s (1978) substitutes for
However, in most comprehensive reviews of the leadership liter- leadership theory, which sought to explain why the leadership
ature, one is left with the impression that these constructs are behaviors are of modest validity, was described as “not . . . all that
archaic. The Ohio State studies, contemporary thinking argues, are more supportive than the tests of path– goal moderators” (House &
to be remembered only to better understand where we are today; Podsakoff, 1994, p. 53). As with path– goal theory, this is a
Consideration and Structure are historical concepts that, due to conclusion borne out by meta-analytic evidence in which the
various methodological and conceptual deficiencies and failed substitutes main effects appear to exert more substantial influences
efforts at validation, led to other advancements that are of greater than the moderating effects as predicted by the theory (Podsakoff,
current relevance. One can readily find quotations in the literature MacKenzie, & Bommer, 1996). Second, before one concludes that
that reflect this viewpoint: there are situational moderators based on the premise of inconsis-
tent results, it is important to first document that these inconsis-
The results show a predominance of low to moderate correlations . . . tencies do in fact exist.
there is as yet almost no evidence on the predictive validity of The purpose of the present study is to provide a comprehensive
“Consideration” and “Initiating Structure” (Korman, 1966, p. 360). examination of the validity of Consideration and Initiating Struc-
Researchers have not been able to establish a consistent link between ture in leadership research. We examine every available study and
task and relationship behaviors and outcomes such as morale, job perform a meta-analytic review that (a) estimates the overall
satisfaction, and productivity (Northouse, 1997, p. 45). validity of Consideration and Initiating Structure in predicting both
The results have been weak and inconsistent for most criteria of subjective (e.g., satisfaction) and objective (e.g., performance)
leadership effectiveness (Yukl, 1998, p. 49). leadership criteria, (b) estimates the correlation between Consid-
Overall, the research based on a two-factor conceptualization of eration and Initiating Structure, and (c) determines whether valid-
leadership behavior has added little to our knowledge about effective ities vary by measure and, if so, by how much. Given the limita-
leadership (Yukl & Van Fleet, 1992, p. 156). tions of narrative reviews (potentially misleading conclusions,
Unfortunately, there was no pattern of leader behavior which was over- or underweighting of particular studies, failure to consider
found to be consistently associated with subordinates’ satisfaction or study artifacts and measurement error; Hunter & Schmidt, 1990),
any criteria of supervisor or manager effectiveness (House & Aditya, it is worthwhile to consider the actual effect sizes once sampling
1997). and measurement error have been taken into account. At the very
least, this review will provide a proper historical perspective on
Although not everyone is as pessimistic about the contemporary these constructs. More fundamentally, it could alter understanding
relevance of the Consideration and Structure factors (notably, of these constructs and suggest new areas for research. In the next
Fleishman, 1998), the above quotations suggest that most research- section of the article, we present hypotheses regarding Consider-
ers consider these leadership behaviors to be outmoded. ation and Initiating Structure.
Given the inconsistencies in the validity of Consideration and
Structure that were concluded to exist, researchers have argued Review of the Literature and Hypothesized Relations
that a way to advance the literature is to investigate situational
moderators (Korman, 1966). Yukl and Van Fleet (1992) effec- Although there may be considerable pessimism regarding the
tively summarized this thinking by commenting, “The relative validity of Consideration and Initiating Structure, one conclusion
importance of specific forms of [Consideration and Initiating that has often been drawn from an examination of the empirical
38 JUDGE, PICCOLO, AND ILIES

data is that Consideration correlates more strongly with follower Hypothesis 4: Validities of (a) Consideration and (b) Initiat-
satisfaction and Initiating Structure correlates more strongly with ing Structure will vary by measure.
performance or effectiveness. Both Bass (1990) and Yukl (1998),
for example, noted that the clearest set of results regarding the The correlation between Consideration and Initiating Structure has
validity of the two behaviors is the correlation of Consideration been the subject of much debate. The concern with the indepen-
with satisfaction. This pattern of associations fits well with the dence of these dimensions can be traced to two sources. First,
conceptual nature of the constructs. As noted by Halpin (1957b), orthogonality of the dimensions was often claimed in the literature;
one would expect leaders high on Initiating Structure to be more orthogonality suggests that the dimensions are wholly indepen-
effective at meeting role expectations, whereas one would expect dent. Second, perhaps the most popular practical application of the
followers to prefer (and thus be more satisfied by) leaders who are leader behavior approach—the managerial grid—is based on the
considerate. Considerate leaders are empathetic (Fleishman & assumption of orthogonality. Weissenberg and Kavanagh (1972)
Salter, 1963), and thus should be skilled at sensing and subse- reviewed the literature on the relationship between measures of
quently satisfying the needs of their followers. Because the orien- Consideration and Structure and concluded that the two dimen-
tation of structuring leaders is toward the task (Bass, 1990), they sions “are not always empirically independent as stated and im-
should be more effective at producing performance outcomes. plied” (p. 127). Bass (1990) agreed, noting, “Initiation and Con-
Although the strongest support can be offered for the expectation sideration should be independent, but such is not the case” (p.
515). Weissenberg and Kavanagh (1972) further argued that the
that Consideration correlates more strongly with follower satisfac-
relationship between Consideration and Initiating Structure varied
tion, whereas Initiating Structure correlates more strongly with
depending on the measure used. Fleishman (1995) also noted that
performance and leader effectiveness (Hypothesis 3), in order to
the Consideration–Structure correlation could be expected to vary
test their effects across all criteria, we hypothesized main effects
by measure, with the LOQ and SBDQ displaying lower intercor-
for Consideration (Hypothesis 1) and Structure (Hypothesis 2) for
relations. Thus, overall, we believe that Consideration and Initiat-
all the criteria.
ing Structure will be positively related, but we also believe that the
Hypothesis 1: Consideration will be positively related to (a) specific measure used will explain variability in these correlations
follower satisfaction (leader satisfaction, job satisfaction) and across studies.
(b) leader performance or effectiveness (leader job perfor-
Hypothesis 5: There will be a significant (nonzero) correla-
mance, group– organization performance, leader effective-
tion between Consideration and Initiating Structure.
ness).
Hypothesis 6: The intercorrelation of Consideration and Ini-
Hypothesis 2: Initiating Structure will be positively related to
tiating Structure will vary by measure.
(a) follower satisfaction (leader satisfaction, job satisfaction)
and (b) leader performance or effectiveness (leader job per- Method
formance, group– organization performance, leader effective-
ness). Literature Search
To identify all possible studies of the relationships between Consider-
Hypothesis 3: Compared with Initiating Structure, Consider- ation, Initiating Structure, and relevant organizational criteria, we searched
ation will be more strongly related to follower satisfaction the PsycINFO database (1887–2001) for studies (articles, book chapters,
(leader satisfaction, job satisfaction), whereas, compared with dissertations, and unpublished reports) that referenced two general key-
Consideration, Initiating Structure will be more strongly related word categories. First, we searched for studies that referenced the terms
to leader performance or effectiveness (leader job performance, Consideration or Initiating Structure, as well as related terms such as
group– organization performance, leader effectiveness). employee orientation, production-centered, or the names of the measures
used to assess Consideration and Initiating Structure (e.g., LBDQ). We
In the literature, four measures of Consideration and Initiating combined these keyword searches with search results for leadership crite-
Structure have been widely used: The Leader Behavior Descrip- ria, including keywords such as leader effectiveness, job satisfaction, group
performance, and motivation. In addition to the electronic searches, we
tion Questionnaire (LBDQ; Halpin, 1957a), the LBDQ, Form XII
examined the reference lists of comprehensive reviews of the literature
(LBDQ-XII; Stogdill, 1963), the Supervisory Behavior Descrip- (e.g., Bass, 1990; Fleishman, 1998; Fleishman, Mumford, Zaccaro, &
tion Questionnaire (SBDQ; Fleishman, 1989b), and the Leader Levin, 1991; Kerr & Schriesheim, 1974; Schreisheim, House, & Kerr,
Opinion Questionnaire (LOQ; Fleishman, 1989a). The LOQ is the 1976; Schriesheim & Kerr, 1974) and included articles that appeared
most unique of these measures in that it asks leaders to indicate relevant.
how often they believe they should (vs. actually do) engage in Our search efforts resulted in the identification of 18 articles referenced
considerate and structuring behaviors. A common theme in the in literature reviews or meta-analyses on relevant topics, and 1,180 ab-
literature is that the specific measures correlate differently with stracts identified by means of electronic searches (878 journal articles and
302 dissertations). In reviewing the abstracts, we eliminated studies that
outcomes (House & Aditya, 1997). Schriesheim and Kerr (1974)
clearly did not include primary data (such as qualitative studies or reviews)
concluded that the LBDQ-XII is the best measure of Consideration
and studies that did not appear to measure leadership. Further, we elimi-
and Initiating Structure. Fleishman (1995) disagreed, arguing that nated studies that did not appear to measure a relevant criterion such as
the SBDQ and LOQ were better measures. Irrespective of which leader job performance or motivation.
measure is superior, in light of past research we expect validities For the remaining 165 articles and 36 doctoral dissertations, we exam-
to vary by measure. ined each study to determine whether it contained the information needed
 CONSIDERATION AND INITIATING STRUCTURE 39

to calculate validities. Several studies were excluded because they reported correlations across studies: an 80% credibility interval excluding zero
percentages or proportions or means with no standard deviations, or be- indicates that 90% of the individual correlations in the meta-analysis
cause they reported other measures of association that could not be con- excluded zero (for positive correlations, less than 10% are zero or negative,
verted to correlations. In total, 130 studies met the criteria for inclusion in and 10% lie at or beyond the upper bound of the interval). Thus, confidence
the database (117 journal articles and 13 dissertations). These studies intervals estimate variability in the mean correlation, whereas credibility
reported a total of 593 correlations computed from 457 independent sam- intervals estimate variability in the individual correlations across the
ples. We formed three distinct databases in order to estimate the meta- studies.
analytic correlations between (1) Consideration and organizational criteria,
(2) Initiating Structure and organizational criteria, (3) Consideration and
Moderator Analyses
Initiating Structure. Table 1 shows the number of primary correlations
included in each meta-analytical database together with the number of We divided the primary estimates into categories according to the
studies that reported these correlations and the total number of independent expected moderator variables. We conducted separate meta-analyses for
samples that were used to compute the correlations. each of the categories to estimate the true correlations for the categories
delimited by moderator variables. Meta-analytical evidence for the pres-
Meta-Analytic Procedures ence of moderators requires that (a) true estimates are different in the
categories formed by the potential moderator variable, and (b) the mean
Using the methods of Hunter and Schmidt (1990), we conducted three corrected standard deviation within categories is smaller than the corrected
main meta-analyses to estimate the correlations among Consideration, standard deviation computed for combined categories. To test for the
Initiating Structure, and organizational criteria. We corrected each primary presence of moderator effects, as recommended by Sagie and Koslowsky
correlation for attenuation because of unreliability in both the predictor and (1993), we report the Q statistic (Hunter & Schmidt, 1990, p. 151), which
the criterion, and then we computed the sample-weighted mean of these tests for homogeneity in the true correlations across studies. A significant
corrected correlations. To estimate parameters describing the variability of Q statistic (which is approximately distributed as a chi-square, ␹2), indi-
the meta-analytical estimates and the confidence in these estimates, the cates the likelihood that moderators explain variability in the correlations
variance of the observed individual estimates was corrected for the effects across studies. If a significant Q statistic across moderator categories
of both sampling and differential measurement error. When authors of becomes nonsignificant within a moderator category, it suggests that the
original studies reported the internal consistency reliability for the mea- moderator explains a significant amount of the variability in the correla-
sures used to compute the primary correlations—as was the case for the tions across the moderator categories.
majority of correlations—we used this value to correct the observed
correlation for attenuation. When reliabilities for predictor and criteria
measures were not reported in the original studies, we averaged the
Results
reliabilities reported in the studies that did provide such estimates for Table 2 provides the results of the overall meta-analyses, which
similar measures and used these mean reliability values to correct the estimate the validity of Consideration and Initiating Structure,
primary correlations.2
collapsed across all criteria and measures. The meta-analysis re-
In addition to reporting point estimates for corrected correlations, it is
sults, on the basis of 163 correlations for Consideration and 159
also important to describe variability in these estimates. In accordance, we
report 80% credibility intervals and 90% confidence intervals around the correlations for Initiating Structure, revealed that both Consider-
estimated population correlations. We believe it is important to report both ation and Structure evince nonzero correlations across the leader-
confidence and credibility intervals because each tells us different things ship criteria. Specifically, because the 90% confidence intervals
about the nature of the estimates. Confidence intervals provide an estimate excluded zero, we can be confident that the mean Consideration
of the variability around the estimated mean corrected correlation that is correlation and the mean Initiating Structure correlation are non-
due to sampling error: a 90% confidence interval around a positive point zero. Furthermore, because the 80% credibility intervals exclude
estimate that excludes zero indicates that if the estimation procedures were zero, more than 90% of the individual correlations for Consider-
repeated a large number of times, the point estimate would be larger than ation and Structure are greater than zero (a maximum of 10% lie
zero in 95% of the cases (the other 5% would be zero or negative).
at or beyond the upper bound of the interval). Despite their
Credibility intervals provide an estimate of the variability of individual
significance, the average correlation for Consideration (␳ˆ ⫽ .48) is
appreciably larger than the average correlation for Initiating Struc-
Table 1 ture (␳ˆ ⫽ .29).
Number of Studies, Samples, and Primary Correlations Included
in the Meta-Analyses 2
As it was pointed out by an anonymous reviewer, compared with
Relationship Studies Samples Correlations Hunter and Schmidt (1990), the procedure developed by Raju, Burke,
Normand, and Langlois (1991) estimates the standard deviation of the
Consideration–Criteria 103 154 209 true-score correlations more accurately (because it takes into account the
Initiating Structure–Criteria 99 151 203 sampling errors associated with sample-specific estimates of the reliabili-
Consideration–Initiating Structure 78 166 181 ties of the scores on the predictor and criterion measures when computing
the sampling variance of the corrected correlations). However, for our
Note. In some cases, the number of correlations appearing in this table is specific analyses, the choice of method (Raju et al., 1991, vs. Hunter &
greater than the number of correlations appearing in subsequent tables. To Schmidt, 1990) made little difference in the analyses. For the overall
avoid violating the independence assumption, whereby only one correla-
analyses of the correlation of Consideration and Initiating Structure with
tion can be analyzed from each sample (Hunter & Schmidt, 1990), if a
study reported correlations of Consideration with motivation and with criteria, for example, the standard deviations of the true score correlations
leader satisfaction, only one of these correlations could be used in the differed by only .0099 (.2549 for Raju et al., 1991, vs. .2648 for Hunter &
overall analysis reported in Table 2. In such a case, the criterion was Schmidt, 1990) and .0112 (.2062 for Raju et al., 1991, vs. .2175 for Hunter
selected at random for the overall analysis. & Schmidt, 1990), respectively.
40 JUDGE, PICCOLO, AND ILIES

Table 2
Relationship of Consideration and Initiating Structure to Leadership

80% CV 80% CV 90% CI 90% CI

Leader behavior k N r៮ ␳ˆ SD␳ˆ lower upper lower upper Q

Consideration 163 20,963 .41 .48 .26 .14 .82 .44 .52 1,500.83**
Initiating Structure 159 20,431 .24 .29 .21 .02 .56 .26 .33 775.09**

Note. k ⫽ number of correlations; N ⫽ combined sample size; r៮ ⫽ mean observed correlation; ␳ˆ ⫽ estimated
true score correlation; SD␳ˆ ⫽ standard deviation of true score correlation; CV ⫽ credibility interval; CI ⫽
confidence interval; Q ⫽ test for homogeneity in the true correlations across studies.
** p ⬍ .01.

Because the overall correlations reported in Table 2 are fairly ples. Overall, the moderators explained 24.7% of the variance in
strong and the credibility intervals exclude zero, the overall valid- the Structure correlations across samples. Thus, Hypothesis 4 was
ities for Consideration and Initiating Structure are interpretable. supported by the results—validities for both Consideration and
However, for both Consideration and Structure, the Q statistic is Structure varied significantly by measure.
significant, meaning that there is significant variability in the Given the potential importance of the moderators, especially
correlations even after taking measurement and sampling error into Criteria and Measure, we conducted separate meta-analyses within
account. Thus, moderators of the overall relationships likely exist. each moderator variable category. Table 3 provides the meta-
In accordance, we specified analysis of variance (ANOVA) mod- analysis results for Consideration and Initiating Structure, broken
els in which the corrected correlation (␳ˆ ) was the dependent down by the six criteria: follower job satisfaction, follower satis-
variable and the study-level moderators were the explanatory vari- faction with the leader, follower motivation, leader job perfor-
ables. The study-level moderators were as follows: Criteria (which mance, group– organization performance, and leader effectiveness.
of the six criteria was used in the study), Measure (which of the As the table shows, both leadership behaviors have nonzero rela-
five measures [LBDQ, LBDQ-XII, LOQ, SBDQ, other] was used tions with the criteria. The only exceptions are the correlation of
in the study), Research Design (cross-sectional or longitudinal), Consideration with leader job performance and the correlation of
Independence of Data Sources (same source or different source), Initiating Structure with follower job satisfaction. In both of these
Study Setting (business, college, military, public sector), and Level cases, the average correlation is distinguishable from zero (the
of the Leader (supervisory or mid- to upper-level leaders). For 90% confidence interval excludes zero), but the 80% credibility
Consideration, the ANOVA results revealed that two variables— interval includes zero, indicating that more than 10% of the cor-
Criteria and Measure— explained a significant amount of variance relations are zero or negative. In general, however, the behaviors
in the correlations across studies, with Criteria accounting for are significantly related to the criteria, supporting Hypotheses 1
14.7% unique variance ( p ⬍ .01) and Measure accounting for and 2.
6.2% unique variance ( p ⬍ .05). As a set, the moderators ex- In Hypothesis 3, we predicted that leader Consideration will be
plained 32.1% of the variance in the Consideration correlations more strongly related to criteria that reflect follower satisfaction
across samples. For Initiating Structure, only Measure accounted (leader satisfaction, job satisfaction) whereas Initiating Structure
for a significant amount of variance. It explained 18.8% ( p ⬍ .01) will be more strongly related to criteria that reflect leader perfor-
unique variance in the Initiating Structure correlations across sam- mance or effectiveness (leader job performance, group–

Table 3
Relationship of Consideration and Initiating Structure to Leadership Criteria

Consideration Initiating Structure

Criterion k N r៮ ␳ˆ k N r៮ ␳ˆ Z
a,b a
Follower job satisfaction 76 11,374 .40 .46 72 10,317 .19 .22 20.49*
Follower satisfaction with leader 49 7,871 .68 .78a,b 49 8,070 .27 .33a,b 44.15*
Follower motivation 11 1,067 .36 .50a,b 12 1,041 .26 .40a,b 2.94*
Leader job performance 25 2,330 .18 .25a 22 2,085 .19 .24a,b .36
Group–organization performance 27 2,008 .23 .28a,b 27 2,079 .23 .30a,b .73
Leader effectiveness 20 1,605 .39 .52a,b 20 1,960 .28 .39a,b 4.11*

Note. Because of space limitations, the Q statistic is not reported but was significant in all cases except for the correlations of Consideration and Initiating
Structure with follower motivation. k ⫽ number of correlations; N ⫽ combined sample size; r៮ ⫽ mean observed correlation; ␳ˆ ⫽ estimated true score
correlation; Z ⫽ Steiger (1980) statistic for differences in correlations of Consideration and Initiating Structure with the criteria.
a
90% confidence interval excluded zero.
b
80% credibility interval excluded zero.
* p ⬍ .05.
 CONSIDERATION AND INITIATING STRUCTURE 41

organization performance, leader effectiveness). To test this hy- analyses, there was little reason to believe that validities were
pothesis, we conducted a significance test on the correlations to interactive. In the ANOVA models explaining variance in Con-
determine if differences were statistically significant. Because the sideration and Initiating Structure, a Criteria ⫻ Measure interac-
correlations reported in this particular moderator analysis are not tion was specified. In neither case did this interaction explain a
independent, our tests for equality must take the dependency of the significant amount of variance in the validities. Furthermore, when
correlations into account. For example, the correlation between separate meta-analyses were conducted by Measure within each
Consideration and follower job satisfaction is not independent of criterion group, the results did not appear to vary any more than
the correlation between Initiating Structure and follower job sat- they did in the separate moderator analyses.
isfaction because these correlations are computed from the same One hierarchical moderator analysis we performed was to in-
population. Thus, to test for equality of dependent correlations, we vestigate whether research design (cross-sectional vs. longitudinal)
used the technique described by Steiger (1980) to test the hypoth- or method variance (common methods vs. no common methods)
eses, H0: ␳ˆ y⫺c ⫽ ␳ˆ y⫺is. The Steiger test takes dependency into explained differences in the validities by measure (e.g., whether
account and allows us to calculate a test statistic with a Student’s the average validity of LOQ measures of Consideration is lower
t distribution. than the other validities because studies using this measure em-
Consistent with Hypothesis 3, Consideration is more strongly ployed more longitudinal designs that minimized common method
related to follower satisfaction (and motivation as well), whereas variance). Specifically, we performed a hierarchical moderator
Initiating Structure is slightly more strongly related to criteria that analysis that computed the validity of each measure within each of
reflect leader performance. Contrary to Hypothesis 3, Consider- the four moderator categories (cross-sectional, longitudinal, com-
ation was more strongly related to leader effectiveness, although mon methods, and no common methods). The results of this
both correlations were moderately strong in magnitude. Although hierarchical moderator analysis are provided in Table 5. In general,
most of the Q statistics are still statistically significant, their size is the results show that the effects of research design and method
greatly reduced, suggesting the criterion does reduce much of the variance do not vary substantially by measure. With few excep-
unexplained variability in the correlations. tions, when a research design effect or method effect was found for
Table 4 provides the meta-analysis results for both behaviors, one measure, it was found for the others. For example, for all four
broken down by measure. As revealed in the table, with a sole measures (LBDQ, LBDQ-XII, LOQ, SBDQ), the Consideration
exception, the validities of the specific measures generalize across validities were significantly higher when leadership behaviors and
the population of primary studies. Specifically, in all cases except outcomes were measured with common methods than when there
the SBDQ measure of Initiating Structure, both the 90% confi- were no common methods. Similarly, for Initiating Structure, none
dence intervals and the 80% credibility intervals exclude zero. of the cross-sectional versus longitudinal correlations were signif-
There were two cases in which the validities did vary appreciably. icantly different for any of the measures. Thus, overall, it does not
The LOQ-Consideration measure and the SBDQ-Initiating Struc- seem that research design or method variance substantially ex-
ture measure both displayed lower correlations than the other plains the different validities obtained by the measures of Consid-
measures. In the other cases, validities did not appear to vary much eration and Structure.
by measure (excepting these two cases, the average deviation in Although less powerful than criterion type and measure, the
validity was only .01 for Consideration and .04 for Initiating results for the other moderators are provided in Table 6. With a
Structure). single exception (Initiating Structure with longitudinal designs),
One might wonder whether validities should be nested hierar- the validities generalized across studies, and in all cases, the
chically, specifically Measure within Criteria. Though there were average validities were distinguishable from zero. To determine
sufficient sample sizes, in many cases, to perform these nested whether validities varied significantly across moderator categories,

Table 4
Validities of Consideration and Initiating Structure by Measure

Consideration Initiating Structure

Measure k N r៮ ␳ˆ Q k N r៮ ␳ˆ Q
a,b a,b
LBDQ 44 4,401 .43 .51 277.79** 43 4,360 .29 .37 211.21**
LBDQ-XII 86 13,110 .47 .54a,b 1,168.84** 86 12,945 .27 .32a,b 372.38**
LOQ 14 772 .24 .34a,b 31.04** 14 772 .27 .40a,b 41.26**
SBDQ 38 5,349 .46 .54a,b 536.50** 38 5,202 .05 .07 181.03**
Other–unidentified measure 27 2,655 .43 .52a,b 191.06** 22 2,305 .29 .37a,b 111.47**

Note. LBDQ ⫽ Leader Behavior Description Questionnaire; LBDQ-XII ⫽ Leader Behavior Description
Questionnaire, Form XII; LOQ ⫽ Leader Opinion Questionnaire; SBDQ ⫽ Supervisory Behavior Description
Questionnaire; k ⫽ number of correlations; N ⫽ combined sample size; r៮ ⫽ mean observed correlation; ␳ˆ ⫽
estimated true score correlation; Q ⫽ test for homogeneity in the true correlations across studies.
a
90% confidence interval excluded zero.
b
80% credibility interval excluded zero.
** p ⬍ .01.
42 JUDGE, PICCOLO, AND ILIES

Table 5
Effects of Research Design and Common Methods on Validities by Measure

Common methods (CM)

Cross-sectional (CS) vs. no common methods
vs. longitudinal (L) (NCM)

Measure ␳ˆ CS ␳ˆ L ␳ˆ CM ␳ˆ NCM

Consideration
LBDQ .55 .36 .57 .37
LBDQ-XII .49 .39 .50 .31
LOQ .44 .26 .50 .20
SBDQ .62 .53 .63 .31
Other .58 .16 .60 .35
Structure
LBDQ .38 .35 .41 .30
LBDQ-XII .34 .23 .34 .20
LOQ .45 .33 .52 .27
SBDQ .01 .17 .04 .14
Other .39 .34 .25 .51

Note. ␳ˆ ⫽ estimated true score correlation; LBDQ ⫽ Leader Behavior Description Questionnaire; LBDQ-
XII ⫽ Leader Behavior Description Questionnaire, Form XII; LOQ ⫽ Leader Opinion Questionnaire; SBDQ ⫽
Supervisory Behavior Description Questionnaire.

we used the Quiñones, Ford, and Teachout (1995) Z test.3 For zero, indicating that in more than 10% of the samples, the
Consideration, the average correlation from cross-sectional re- Consideration–Structure correlation was zero or negative. Indeed,
search designs was significantly higher than the average correla- an examination of the correlations in the database revealed that the
tion from longitudinal designs (Z ⫽ 2.23, p ⬍ .05), but for correlation was zero or negative in 35% of the samples. Thus,
Initiating Structure, the difference between the two designs was Hypothesis 5 (a significant Consideration–Structure correlation)
not significant (Z ⫽ 0.78, ns). When both Consideration and the received mixed support. Furthermore, the very large and signifi-
leadership criterion were measured by the same source, the aver- cant Q statistic suggests that there are factors that moderate this
age correlation was significantly higher than when measured by relationship. Hypothesis 6 predicted that the measure used would
different sources (Z ⫽ 3.73, p ⬍ .01). The independence of data moderate the Consideration–Structure relationship. In accordance,
sources, however, did not affect the average correlation for Initi- as with previous analyses, we specified an ANOVA model in
ating Structure. With the exception of a difference between the which measure was used to explain variability in the
average correlation for Consideration in business versus in public Consideration–Structure correlations across samples. The
sector settings (Z ⫽ ⫺2.16, p ⬍ .05), no other study setting ANOVA results revealed that the measure of Consideration and
moderated the average validity of Consideration or Initiating Structure explained a large and significant amount—52.7% ( p ⬍
Structure. Further, there did not exist any significant differences in .01)— of the variance in the correlations.
the average validity for Consideration or Initiating Structure based
Given that the measure used in studies explained much of the
on the level of the leader within the organization studied.
variability in the Consideration–Structure correlation across sam-
Finally, although not provided in the tables, we also investigated
ples, we estimated separate meta-analyses by the measure used in
whether validities for Consideration and Structure varied by
the study. The results of these meta-analyses broken down by
whether the correlation came from a published or unpublished
measure are provided in Table 8. As revealed in the table, the
(dissertation) source. Overall, the validities were as follows:
Consideration–Structure correlations varied substantially by mea-
Consideration-published, ␳ˆ p ⫽ .47 (k ⫽ 149); Consideration-
sure. In two cases (LBDQ and LBDQ-XII), the average correla-
unpublished, ␳ˆ u ⫽ .54 (k ⫽ 14); Structure-published, ␳ˆ p ⫽ .30
(k ⫽ 145); Structure-unpublished, ␳ˆ u ⫽ .30 (k ⫽ 14). Using the tions are moderately to strongly positive, in two cases (LOQ and
Quiñones et al. (1995) Z test, the published– unpublished differ- SBDQ), the average correlations are weakly negative, and for
ences in validity were not significantly different (Consideration: ␳ˆ p other– unidentified measures, the average correlation is moderately
vs. ␳ˆ u, Z ⫽ ⫺1.16, ns; Structure: ␳ˆ p vs. ␳ˆ u, Z ⫽ ⫺.01, ns), positive. Using the Quiñones et al. (1995) test, the average
suggesting that the overall validities do not differ between pub- Consideration–Structure correlation based on the LBDQ measure
lished and unpublished sources.
Table 7 provides results for the meta-analysis on the relationship 3
The difference between the Steiger (1980) test and the Quiñones et al.
between Consideration and Initiating Structure across the 166 (1995) Z test is that the former tests for differences in dependent correla-
samples where we were able to locate such a correlation. As is tions (a comparison of correlations from the same sample based on the
revealed in the table, the average corrected correlation is relatively same participants), whereas the latter tests for differences in independent
weak (␳ˆ ⫽ .17). Although this average corrected correlation is correlations (a comparison of correlations from two independent samples
distinguishable from zero, the 80% credibility interval included or two distinct moderator categories).
 CONSIDERATION AND INITIATING STRUCTURE 43

Table 6
Other Moderator Analysis Results

Consideration Initiating Structure

Moderator k N r៮ ␳ˆ k N r៮ ␳ˆ

Research design
Cross-sectional 104 13,419 .40 .46a,b 104 13,346 .25 .31a,b
Longitudinal 43 3,615 .28 .36a,b 38 3,135 .20 .26a
Independence of data sources
Same 98 13,260 .41 .47a,b 92 12,625 .24 .29a,b
Different 52 4,099 .25 .32a,b 53 4,181 .23 .30a,b
Study setting
Business 79 10,305 .37 .43a,b 78 10,062 .24 .29a,b
College 11 1,331 .46 .56a,b 10 1,297 .19 .25a,b
Military 11 649 .30 .40a,b 12 678 .31 .46a,b
Public sector 51 6,616 .46 .53a,b 47 6,311 .23 .28a,b
Level of leader
Supervisory 124 16,574 .40 .47a,b 121 16,215 .23 .28a,b
Mid or upper 25 2,244 .45 .54a,b 23 2,050 .25 .33a,b

Note. k ⫽ number of correlations; N ⫽ combined sample size; r៮ ⫽ mean observed correlations; ␳ˆ ⫽ estimated
true score correlation.
a
90% confidence interval excluded zero.
b
80% credibility interval excluded zero.

is significantly stronger than the average correlation based on the inconsistent” (p. 49). Northouse (1997) remarked that the “pre-
LOQ or SBDQ measures (Z ⫽ 10.33, p ⬍ .01, and Z ⫽ 7.22, p ⬍ ponderance of the research in this area was inconclusive” (p. 38).
.01, respectively). Similarly, the average correlation based on the Yukl and Van Fleet (1992) concluded, “The long fixation on
LBDQ-XII is significantly stronger than the average correlation consideration and initiating structure appears to have come to an
based on the LOQ or SBDQ measures (Z ⫽ 13.56, p ⬍ .01, and end” (p. 159). Even models that sought to delineate the conditions
Z ⫽ 8.26, p ⬍ .01, respectively). Thus, Hypothesis 6 is supported. under which Consideration and Initiating Structure were
The correlation between Consideration and Initiating Structure relevant—path– goal theory and substitutes for leadership— have
does vary depending on the measure used to assess these not received strong support in the literature (Dionne, Yammarino,
constructs. Atwater, & James, 2002; Wofford & Liska, 1993). Since 1980,
there have been only a handful of empirical journal articles on
Discussion Consideration or Initiating Structure, and there have been none
since 1987. A recent 500-page book on leadership revealed only a
Despite their predominance in the literature a generation ago, single, passing citation to the Ohio State studies (Mathieu, 2001, p.
Consideration and Initiating Structure have long been dismissed as 454). Similarly, Consideration and Structure were completely ab-
important influences on leadership effectiveness. House and Pod- sent from a recent 53-chapter edited book on leadership (Hickman,
sakoff (1994) labeled the body of findings as “relatively disap- 1998). These behaviors seem to be in danger of being viewed as
pointing” (p. 50). Yukl (1998) labeled the results as “weak and

Table 8
Table 7 Consideration–Initiating Structure Correlation by Measure
Meta-Analysis of Correlation Between Consideration and
Initiating Structure Measure k N r៮ ␳ˆ Q

Statistic Value LBDQ 37 5,138 .36 .44a,b 315.50**

LBDQ-XII 32 5,806 .37 .46a,b 153.27**
Number of correlations (k) 181 LOQ 78 10,051 ⫺.06 ⫺.08a 242.95**
Combined sample size (N) 26,295 SBDQ 18 2,858 ⫺.07 ⫺.08 120.02**
Average uncorrected correlation (r៮) .14 Other–unidentified measure 16 2,442 .22 .28a 130.66**
True score correlation (␳ˆ ) .17
Standard deviation of true score correlation (SD␳ˆ ) .33 Note. LBDQ ⫽ Leader Behavior Description Questionnaire; LBDQ-
80% credibility value lower ⫺.25 XII ⫽ Leader Behavior Description Questionnaire, Form XII; LOQ ⫽
80% credibility value upper .59 Leader Opinion Questionnaire; SBDQ ⫽ Supervisory Behavior Descrip-
90% confidence interval lower .12 tion Questionnaire; k ⫽ number of correlations; N ⫽ combined sample
90% confidence interval upper .22 size; r៮ ⫽ mean observed correlation; ␳ˆ ⫽ estimated true score correlation;
Q 1,826.28** Q ⫽ test for homogeneity in the true correlations across studies.
a
90% confidence interval excluded zero.
Note. Q ⫽ test for homogeneity in the true correlations across studies. b
80% credibility interval excluded zero.
** p ⬍ .01. ** p ⬍ .01.
44 JUDGE, PICCOLO, AND ILIES

historical artifacts in leadership research—important artifacts— That Consideration and Initiating Structure are far from obso-
but artifacts of little contemporary relevance nonetheless. lete—that they have significant main effects— does not deny the
This denouement for the Ohio State leadership behaviors, how- possibility that these effects are moderated by other variables.
ever, may be premature. The results of the present quantitative Indeed, our moderator analyses were successful in revealing sev-
review revealed that both Consideration and Initiating Structure eral factors that appear to affect the validity of Consideration and
have important main effects on numerous criteria that most would Initiating Structure. There may be other moderators. Furthermore,
argue are fundamental indicators of effective leadership. It is our results do not fully address concerns that have been expressed
striking how the validities for each behavior generalized—across about the measures of Consideration and Initiating Structure. For
criteria, across measures, and even over time and across sources. example, concerns have been raised about skew in responses to the
Of course, these behaviors are not all there is to solving the items, and there are concerns about leniency, social desirability,
mysteries of leadership effectiveness. However, just as surely, the and halo effects (see Schriesheim & Kerr, 1974). That the mea-
results do suggest that these behaviors—Consideration and Initi- sures correlate with “hard” performance and effectiveness criteria
ating Structure—are important pieces in the leadership puzzle. ameliorates some, but certainly not all, of these concerns. We
In interpreting the meaning of this meta-analytic review, one further note that measures of transformational leadership likely
should note that these results are not the result of “fantastic” suffer from many of the same limitations. Indeed, a meta-analysis
meta-analytic corrections. We simply corrected the leader behavior of the transformational leadership literature (Lowe, Kroeck, &
and criterion results on the basis of coefficient alpha reliability Sivasubramaniam, 1996) appears to be dominated by correlations
estimates, which, in most cases, were at acceptable levels (the suffering from common-source limitations.
average reliability, across all behaviors and criteria, was ␣ ⫽ .81). The measure used in leadership studies did moderate the validity
To be sure, there is an argument that interrater reliability estimates of both Consideration and Initiating Structure. Although
would be the more appropriate corrections (see Murphy & De- Schriesheim and Kerr (1974) favored the LBDQ-XII, and Fleish-
Shon, 2000, and Schmidt, Viswesvaran, & Ones, 2000). Further- man (1995) preferred the LOQ or SBDQ, the original LBDQ and
more, one could even argue that the correlations should be cor- the LBDQ-XXII have the highest validities averaged across Con-
sideration and Structure. That is, for Consideration, the LOQ was
rected for range restriction because it is likely that individuals are
less valid than the other three measures and, for Initiating Struc-
selected to leadership positions on the basis of the degree to which
ture, the SBDQ was less valid than the other three. On the other
they are considerate and structuring. It is likely that performing
hand, in only one of eight possible relationships involving the four
these two corrections (using interrater reliability and correcting
measures of the two behaviors was the validity not generalizable
estimates for range restriction) would have made the corrected
(the LOQ measure of Initiating Structure). Thus, the measure of
correlations considerably higher. The point here is that these
Consideration and Initiating Structure does matter, but only in one
results are not news because our corrections transformed weak
case does it matter so dramatically as to destroy the validity of the
correlations into moderately strong ones. The correlations were
behavior. Overall, the pattern of correlations is more consistent
always there, which makes it unfortunate that they were
than has been depicted in previous reviews (e.g., Yukl, 1998).
overlooked.
In general, Consideration exhibited stronger relationships with
Why were the correlations overlooked? How could the conclu- the criteria than did Initiating Structure. This was especially true
sions of reviewers of this literature have been so wrong? One with respect to follower satisfaction (follower job satisfaction,
means of reconciling the pessimistic conclusions of past research follower satisfaction with the leader). Initiating Structure did have
with the more positive results of this review is to examine the slightly stronger relations with group– organization performance.
history of other areas. In the area of personality and job perfor- Thus, to some extent, our hypothesis that followers prefer consid-
mance, prior to Barrick and Mount’s (1991) meta-analysis, re- erate leaders but will perform more effectively for structuring
searchers had cautioned against the use of personality tests (Mount leaders, was supported. On the other hand, Consideration was
& Barrick, 1998). Similarly, meta-analysis dramatically altered the linked to leader job performance and group– organization perfor-
conclusions in the area of intelligence and integrity tests (see mance, and Initiating Structure was linked to leader satisfaction.
Schmidt & Hunter, 1998, for a review). Although there is nothing Both behaviors also were linked to follower motivation and leader
wrong with qualitative reviews per se, they are subject to various effectiveness, with Consideration being somewhat more important.
errors, including the fact that subjective accounting of study results Whereas the validities of Consideration and Initiating Structure
often leads to inaccurate conclusions. As a specific case in point, (in correlating with the criteria) do not vary strongly across mea-
Kerr and Schriesheim (1974) concluded, “LOQ studies have typ- sures (as a rule), the intercorrelation between the two behaviors
ically yielded much less significant relationships between predic- does vary considerably across measures. Fleishman (1995) com-
tors and criteria than have those using the LBDQ or SBDQ” (p. mented that the LOQ and SBDQ Consideration and Initiating
561). Yet, our results show that although for Consideration this Structure intercorrelations are quite small, and our results confirm
conclusion is accurate, for Initiating Structure this conclusion this observation. The correlations between the LBDQ and LBDQ-
clearly is inaccurate. Perhaps the conclusions in past qualitative XII Consideration and Initiating Structure scales were appreciably
reviews were not tested meta-analytically because the literature larger but not so large as to render the two concepts redundant. For
had already been pronounced as dead before the advent of meta- example, the correlations among the transformational leadership
analysis in industrial– organizational psychology. Why meta- dimensions (average r ⫽ .78), and even between transformational
analyze a literature, researchers may have asked, that had long leadership and transactional leadership in the form of contingent
been obsolete? reward (average r ⫽ .68), are far higher than the LBDQ Consid-
 CONSIDERATION AND INITIATING STRUCTURE 45

eration and Structure correlation (see Lowe et al., 1996, p. 421). Structure, on the one hand, structuring leaders should foster pro-
The implications of the Consideration–Initiating Structure corre- cedural justice in that they clearly communicate expectations and
lation, whether the LOQ and SBDQ are to be preferred due to their set clear standards across people. On the other hand, their directive
lower Consideration–Structure intercorrelation, and implications nature may cause followers to feel they are without a voice in how
of this association for the incremental validity of the scales, are the rules and procedures are established. Clearly, the integration of
issues that need further analysis in future research. the Ohio State factors with justice theory is an important area for
The validity of Consideration was significantly stronger in future research.
cross-sectional designs than in longitudinal designs and when Further mediating mechanisms and integration with other lit-
variables were measured by the same source. Perhaps there is a eratures. Although formulations of leader–member exchange
higher social desirably component to Consideration, which would (LMX) theory suggest that leaders behave differently toward dif-
explain why same-source and cross-sectional validities were sub- ferent followers, affect or liking is closely linked to LMX opera-
stantially higher and, perhaps in part, why Consideration correlates tionalizations (Liden, Sparrowe, & Wayne, 1997). Are considerate
so strongly with leader satisfaction. It is interesting to note that leaders more effective, in part, because they are better liked?
these methodological features did not affect the validity of Initi- Liking has been speculated to be a contaminant in the measure-
ating Structure. ment of the behavioral dimensions (Schriesheim & Kerr, 1974).
Perhaps it is, instead, a mediating variable that explains why
Future Research considerate leaders are more satisfying to followers. As for medi-
ators of Initiating Structure effects, perhaps structuring leaders are
Given the validity estimates reported here, further study of more effective because they are more likely to set hard and specific
Consideration and Initiating Structure in leadership research is goals for their followers. The motivational effects of goal setting
warranted. There are two general directions for future research. are well documented (Locke, 1997), and Kirkpatrick, Locke, and
First, research should address some questions that were unan- Latham (1996) integrated the goal-setting and leadership litera-
swered (perhaps due to the virtual abandonment of the topic) in tures. This effort did not, however, include Initiating Structure.
past research. Second, Consideration and Structure have virtually Thus, it seems possible that integrating Consideration and Struc-
disappeared in contemporary research; these concepts should be ture with the concepts in LMX and goal-setting theories, respec-
integrated with more recent theorizing in leadership research spe- tively, might prove worthwhile. Another mechanism that might
cifically and organizational behavior research more generally. mediate the relationship between Consideration and Structure and
Thus, beyond the areas already suggested, we provide below an follower performance concerns followers’ beliefs about their per-
agenda for future research. formance abilities. That is, because goals are thought to influence
Mediators of consideration and structure effects. Future re- self-efficacy (e.g., Locke, 1997), Initiating Structure may contrib-
search should explore mediators that explain why Consideration ute to follower feelings of increased efficacy which, in turn, lead
and Structure have their effects on the outcomes. As noted by to performance (Bandura, 1986). In addition, supportive leader
Northouse (1997), past research “has not adequately shown how behavior is linked to follower self-confidence in House’s (1996)
leaders’ styles are associated with performance outcomes” (p. 45). revised theory of path– goal leadership, which suggests that Con-
Although many potential mediators could be investigated, an es- sideration should also lead to increased follower self-efficacy and
pecially promising mediating framework is organizational justice. performance. Future research should investigate such mediating
Justice research has suggested that there are at least three organi- mechanisms.
zational justice dimensions: distributive, procedural, and interac- Integration with transformational leadership. Transformational–
tional (Cropanzano, Byrne, Bobocel, & Rupp, 2001). It seems charismatic leadership is the most popular contemporary leader-
likely that leaders high on Initiating Structure are likely to foster ship theory, and one that also is supported by meta-analytic evi-
distributive justice. As discussed by Greenberg and Lind (2000), dence (Lowe et al., 1996). Although transformational leadership is
the equity principle is that outcomes should be distributed accord- seen as an advance over the Consideration and Initiating Structure
ing to contributions. Leaders high on Initiating Structure define dimensions, there has been surprisingly little discussion of the
rules regarding work duties, maintain standards, and determine the relationship of the Ohio State leader behaviors to transformational
consequences of goal attainment. It seems likely that followers of leadership. Bass (1999) pointedly argued that a facet of transfor-
such leaders should feel that their rewards are distributed accord- mational leadership—individualized consideration—was to be dis-
ing to the equity rule. Interactional justice is the degree to which tinguished from the Ohio State Consideration factor. However,
individuals are treated with dignity, respect, and sensitivity by Seltzer and Bass’s (1990) study revealed moderately strong cor-
authority figures. Because Consideration is “the extent to which a relations of transformational leadership with Consideration and
leader exhibits concern for the welfare of the other members of the Initiating Structure. Mindful of Bass’s (1999) argument that Con-
group” (Bass, 1990, p. 511), leaders high on Consideration should sideration and individualized consideration are distinct, we are
be better at fostering interactional justice. Finally, it appears that aware of no empirical research that has fully addressed this issue.
both Consideration and Structure are relevant to procedural justice, How the two frameworks can be integrated, given their validity, is
although in different ways. Considerate leaders should provide deserving of attention.
higher procedural justice in that they appear to follow several of Causal ordering. As Bass (1990) noted, a limitation of past
Leventhal’s (1980) rules in that they “put subordinates’ sugges- research is the inability to ascertain whether “leadership is a cause,
tions into operation” and “obtain subordinates’ approval on im- a consequence, or a coincidence of group effectiveness, satisfac-
portant matters before going ahead” (Bass, 1990, p. 511). As for tion, or other valued outcomes” (p. 542). This is a long-standing
46 JUDGE, PICCOLO, AND ILIES

criticism of this literature (Korman, 1966), yet with few exceptions existence or importance of nonlinearities. Linear and nonlinear
there has been little effort to study the causal relationship between effects can coexist.
Consideration, Structure, and outcomes. Moreover, because the Additional outcomes. It seems possible that the validity of
characteristics understanding, caring, and concerned, as well as Consideration and Structure might be even more impressive if
decisive, directive, and organized are endorsed by individuals as additional criteria were studied. Perhaps past researchers have not
attributes of leaders (Lord, Foti, & De Vader, 1984), it seems studied the criteria that these measures are most likely to predict.
possible that implicit theories of leadership may explain the va- Specifically, it would seem that structuring leaders might result in
lidities of Consideration and Structure. Specifically, individuals greater team efficacy and performance, and considerate leaders
may attribute effective leadership by perceiving such leaders as might produce greater team cohesiveness. Perhaps considerate
considerate and structuring, irrespective of whether those behav- leaders, by facilitating trust, generate more integrative conflict
iors actually led to effective leadership. resolution strategies and greater intra- and interorganizational co-
Construct validity of measures. Do different measures of Con- ordination. Undoubtedly, there are other criteria that are concep-
sideration and, in particular, Initiating Structure, reflect the same tually close to Consideration and Structure to which these behav-
construct and, if not, what differing aspects of leadership do they iors could be related.
tap? Is one more construct-valid than the others? There appears to Levels of analysis. Finally, although nearly all of the studies
be a trade-off in that the most valid measures across both factors included in our meta-analysis were at the individual leader level of
(LBDQ, LBDQ-XII) also have the strongest intercorrelations be- analysis, in future research this may not be the case. Most studies
tween the two factors. It seems that the higher overall validity of were at the individual level of analysis where a follower evaluated
the LBDQ and LBDQ-XII measures of Consideration and Struc- the leadership of his or her leader, and these evaluations were
ture is paid in the coin of higher intercorrelations between the related to a criterion corresponding to that leader, such as follower
measures. The measurement of the dimensions was a prominent satisfaction with the leader or an independent rating of leader job
theme in the literature (e.g., Schriesheim & Kerr, 1974) but one performance (e.g., House, Filley, & Kerr, 1971). In a few studies
that never reached a resolution. Given the results observed in this (e.g., Parker, 1963), Consideration and Structure were measured
review, more construct validity work is needed. with follower ratings averaged across followers. Such designs
Causes of Consideration and Structure. What causes a leader implicitly assume that the leader effects are equivalent across
to display considerate and structuring behaviors? It seems possible followers, which may or may not be a valid assumption. Given that
that agreeable individuals are more considerate, whereas consci- followers are nested within leaders and that there may be multiple
entious (and perhaps less open) individuals are more structuring. levels of leaders, in the future it would benefit researchers to use
Fleishman (1957) found that leaders who had high scores on a multilevel framework in investigating the effects of Consider-
Consideration also had high scores on benevolence, a trait that ation and Structure.
seems closely aligned to agreeableness. Similarly, Bass (1990, p.
522) reported the results of a study showing that charm was related Limitations
to Consideration; charming is a characteristic that could easily be
associated with extraverts, along with related traits such as witty, Although the findings reported here make a contribution to the
flamboyant, and vivacious (Goldberg, 1990). Bass also reported leadership literature, several limitations must be acknowledged.
that ascendancy, a trait associated with conscientiousness, is re- First, from a conceptual perspective, our moderator analyses were
lated to Initiating Structure (p. 523). Although these studies pre- limited by the nature of the primary studies in that we could not
date the popularity of the five-factor model, they do suggest that investigate the effects of theoretically relevant moderators such as
future research linking personality to Consideration and Structure those proposed in House’s (1996) revised theory of path– goal
would be worthwhile. leadership (e.g., uncertainty, predictability, interdependence, etc.).
Nonlinear relations. Fleishman (1995, 1998) has repeatedly Second, a methodological limitation stems from the fact that the
noted that the field has practically ignored potential nonlinearities number of correlations for some of the moderator analyses is
in the validity of Consideration and Structure. In 1998, for exam- relatively small. However, because meta-analytic estimates based
ple, Fleishman commented, “It is difficult to find other work on relatively small number of studies are not biased, and only their
stimulated by our findings of curvilinearity in leadership–work variability is affected (Schmidt, Hunter, Pearlman, & Hirsh, 1985),
group performance relationships” (p. 831). Nonlinearities involv- such meta-analytic results remain the best available estimates in
ing the Ohio State factors could take several forms. First, there the literature. Finally, our results, because they are a summary of
may be interactions between Consideration and Structure. Fleish- the extant literature, reflect more correlations relating the behav-
man and Harris (1962) found that relatively considerate leaders iors to some criteria (satisfaction) than to others (commitment,
could increase structure with little or no negative consequences, performance, etc.). However, we should note that, in general, the
whereas the same was not true for less considerate leaders. Second, behaviors do appear to be related to performance and other objec-
the effects of Consideration and Structure may be curvilinear, as tive criteria, even though the number of correlations is relatively
Fleishman and Harris (1962) also found in their study. Finally, small.
although moderating effects (e.g., pressure for production, climate)
have been hypothesized and tested (see Fleishman, 1995), the Conclusion
results are piecemeal and replication attempts have been rare. We
note that because we found linear relations of Consideration and In discussing the results of a meta-analytic study, the goal
Structure with the outcomes in this review does not deny the should be to answer the question, “‘Where are we now that this
 CONSIDERATION AND INITIATING STRUCTURE 47

meta-analysis has been conducted?’” (Rosenthal, 1995, p. 190). hard-core unemployed over a two-year period. Journal of Applied Psy-
Where we are now, in terms of the validity of Consideration and chology, 59, 38 – 42.
Initiating Structure, is at a considerably more optimistic place than *Begum, H. A., & Anwar, M. N. (1978). Consideration and structure
past reviews have placed us. We are not arguing that all of the dimensions of leadership behaviour and supervisory effectiveness as
criticisms of past research have been wrong or that the present determined by managers’ ratings and group productivity. Asian Journal
of Psychology & Education, 3, 40 – 47.
investigation answers all of the unanswered questions regarding
*Bernardin, H. J. (1976). The influence of reinforcement orientation on the
the Ohio State leadership behaviors. For example, our results
relationship between supervisory style and effectiveness criteria. Dis-
cannot speak to nonlinearities in the relations, a possibility that has
sertation Abstracts International, 37, 1018.
been repeatedly raised (Fleishman, 1995, 1998) but seldom inves- *Bernardin, H. J. (1987). Effect of reciprocal leniency on the relation
tigated. Furthermore, the paradox of differing results by measure, between consideration scores from the Leader Behavior Description
although being informed by this review, is essentially left unre- Questionnaire and performance ratings. Psychological Reports, 60,
solved. What we are arguing is that it is inadvisable, at this point, 479 – 487.
to abandon Consideration and Initiating Structure in leadership Bernardin, H. J., & Alvares, K. M. (1976). The managerial grid as a
research. Some of the more prominent organizational psycholo- predictor of conflict resolution method and managerial effectiveness.
gists in the history of the field—Stogdill, Shartle, and Fleishman, Administrative Science Quarterly, 21, 84 –92.
among others—spent the better part of their careers researching *Besco, R. O., & Lawshe, C. H. (1959). Foreman leadership as perceived
these concepts. It appears their investigations were more produc- by superiors and subordinates. Personnel Psychology, 12, 573–582.
tive than previously thought, and that the fruit born of these *Bird, A. M. (1977). Development of a model for predicting team perfor-
investigations could prove useful once again in leadership re- mance. Research Quarterly, 48, 24 –32.
search. Blake, R. R., Mouton, J. S., & Bidwell, A. C. (1962). Managerial grid.
Advanced Management–Office Executive, 1, 12–15.
*Bledsoe, J. C. (1978). Morale of curriculum directors as related to
References perceptions of leadership of superintendents. Psychological Reports, 43,
1283–1288.
References marked with an asterisk indicate studies included in the *Bowers, D. G., & Seashore, S. E. (1966). Predicting organizational
meta-analysis.
effectiveness with a four-factor theory of leadership. Administrative
*Abdel-Halim, A. A. (1981). Personality and task moderators of subordi- Science Quarterly, 11, 238 –263.
nate responses to perceived leader behavior. Human Relations, 34, *Brown, S. E., & Bledsoe, J. C. (1978). Job satisfaction of school super-
73– 88. intendents as related to perceptions of leader’s behavior. Psychological
*Aldag, R. J., & Brief, A. P. (1977). Relationships between leader behavior Reports, 42, 171–174.
variability indices and subordinate responses. Personnel Psychology, 30, *Bryman, A., Bresnen, M., Ford, J., & Beardsworth, A. (1987). Leader
419 – 426. orientation and organizational transience: An investigation using Fied-
*Anderson, L. R. (1966). Leader behavior, member attitudes, and task ler’s LPC Scale. Journal of Occupational Psychology, 60, 13–19.
performance of intercultural discussion groups. Journal of Social Psy- Burns, J. M. (1978). Leadership. New York: Harper & Row.
chology, 69, 305–319. Chemers, M. M. (1997). An integrative theory of leadership. Mahwah, NJ:
*Badin, I. J. (1974). Some moderator influences on relationships between Erlbaum.
consideration, initiating structure and organizational criteria. Journal of *Conger, J. A., & Kanungo, R. N. (1992). Perceived behavioral attributes
Applied Psychology, 59, 380 –382. of charismatic leadership. Canadian Journal of Behavioral Science, 24,
Bales, R. F. (1954). In conference. Harvard Business Review, 32, 44 –50. 86 –102.
Bandura, A. (1986). Social foundations of thought and action: A social
*Conger, J. A., & Kanungo, R. N. (1994). Charismatic leadership in
cognitive theory. Englewood Cliffs, NJ: Prentice Hall.
organizations: Perceived behavioral attributes and their measurement.
Barrick, M. R., & Mount, M. K. (1991). The big five personality dimen-
Journal of Organizational Behavior, 15, 439 – 452.
sions and job performance: A meta-analysis. Personnel Psychology, 44,
Cropanzano, R., Byrne, Z. S., Bobocel, D. R., & Rupp, D. E. (2001). Moral
1–26.
virtues, fairness heuristics, social entities, and other denizens of organi-
*Bartol, K. M., & Wortman, M. S. (1975). Male versus female leaders:
zational justice. Journal of Vocational Behavior, 58, 164 –209.
Effects on perceived leader behavior and satisfaction in a hospital.
*Cummins, R. C. (1971). Relationship of initiating structure and job
Personnel Psychology, 28, 533–547.
*Bartolo, K., & Furlonger, B. (2000). Leadership and job satisfaction performance as moderated by consideration. Journal of Applied Psy-
among aviation fire fighters in Australia. Journal of Managerial Psy- chology, 55, 489 – 490.
chology, 15, 87–97. *Cummins, R. C. (1972). Leader-member relations as a moderator of the
*Bass, B. M. (1956). Leadership opinions as forecasts of supervisory effects of leader behavior and attitude. Personnel Psychology, 25, 655–
success. Journal of Applied Psychology, 40, 345–346. 660.
*Bass, B. M. (1958). Leadership opinions as forecasts of supervisory *Dawson, J. E., Messe, L. A., & Phillips, J. L. (1972). Effect of instructor–
success: A replication. Personnel Psychology, 11, 515–518. leader behavior on student performance. Journal of Applied Psychology,
Bass, B. M. (1985). Leadership and performance beyond expectations. 56, 369 –376.
New York: Free Press. *Dessler, G., & Valenzi, E. R. (1977). Initiation of structure and subordi-
Bass, B. M. (1990). Bass and Stogdill’s handbook of leadership. New nate satisfaction: A path analysis of path– goal theory. Academy of
York: Free Press. Management Journal, 20, 251–259.
Bass, B. M. (1999). Two decades of research and development in trans- Dionne, S. D., Yammarino, F. J., Atwater, L. E., & James, L. R. (2002).
formational leadership. European Journal of Work and Organizational Neutralizing substitutes for leadership theory: Leadership effects and
Psychology, 8, 9 –32. common-source bias. Journal of Applied Psychology, 87, 454 – 464.
*Beatty, R. W. (1974). Supervisory behavior related to job success of *Distefano, M. K., & Pryer, M. W. (1973). Comparisons of leader and
48 JUDGE, PICCOLO, AND ILIES

subordinate descriptions of leadership behavior. Perceptual & Motor From conceptualization to implication to application. In C. L. Cooper &
Skills, 37, 714. E. A. Locke (Eds.), Industrial and organizational psychology: Linking
*Downey, H. K., Sheridan, J. E., & Slocum, J. W. (1975). Analysis of theory with practice (pp. 72–108). Oxford, United Kingdom: Blackwell.
relationships among leader behavior, subordinate job performance and *Greene, C. N. (1975). The reciprocal nature of influence between leader
satisfaction: A path– goal approach. Academy of Management Journal, and subordinate. Journal of Applied Psychology, 60, 187–193.
18, 253–262. *Gruenfeld, L. W., & Weissenberg, P. (1966). Supervisory characteristics
*Drory, A., & Ben-Porat, A. (1980). Leadership style and leniency bias in and attitudes toward performance appraisals. Personnel Psychology, 19,
evaluation of employees’ performance. Psychological Reports, 46, 735– 143–151.
739. *Halpin, A. W. (1954). The leadership behavior and combat performance
*Ehrlich, S. B., Meindl, J. R., & Viellieu, B. (1990). The charismatic of airplane commanders. Journal of Abnormal & Social Psychology, 49,
appeal of a transformational leader: An empirical case study of a small, 19 –22.
high-technology contractor. Leadership Quarterly, 1, 229 –247. *Halpin, A. W. (1955). The leader behavior and leadership ideology of
*Evans, M. G. (1970). The effects of supervisory behavior on the path– educational administrators and aircraft commanders. Harvard Educa-
goal relationship. Organizational Behavior & Human Decision Pro- tional Review, 25, 18 –32.
cesses, 5, 277–298. Halpin, A. W. (1957a). Manual for the Leader Behavior Description
*Evans, M. G. (1974). Extensions of a path– goal theory of motivation. Questionnaire. Columbus, OH: Bureau of Business Research, Ohio State
Journal of Applied Psychology, 59, 172–178. University.
*Fleishman, E. A. (1953). The description of supervisory behavior. Journal *Halpin, A. W. (1957b). The leader behavior and effectiveness of aircraft
of Applied Psychology, 37, 1– 6. commanders. In R. M. Stogdill & A. E. Coons (Eds.), Leader behavior:
*Fleishman, E. A. (1957). A leader behavior description for industry. In Its description and measurement (pp. 52– 64). Columbus, OH: Bureau of
R. M. Stogdill & A. E. Coons (Eds.), Leader behavior: Its description Business Research.
and measurement (pp. 103–119). Columbus, OH: Bureau of Business *Halpin, A. W. (1957c). The observed leader behavior and ideal leader
Research. behavior of aircraft commanders and school superintendents. In R. M.
*Fleishman, E. A. (1957). The Leader Opinion Questionnaire. In R. M. Stogdill & A. E. Coons (Eds.), Leader behavior: Its description and
Stogdill & A. E. Coons (Eds.), Leader behavior: Its description and measurement (pp. 65– 68). Columbus, OH: Bureau of Business Re-
measurement (pp. 120 –133). Columbus, OH: Bureau of Business Re- search.
search. *Halpin, A. W., & Winer, B. J. (1957). A factorial study of the leader
Fleishman, E. A. (1973). Twenty years of consideration and structure. In behavior descriptions. In R. M. Stogdill & A. E. Coons (Eds.), Leader
E. A. Fleishman & J. G. Hunt (Eds.), Current developments in the study behavior: Its description and measurement (pp. 39 –51). Columbus, OH:
of leadership (pp. 1– 40). Carbondale, IL: Southern Illinois University Bureau of Business Research.
Press. *Hand, H. H., & Slocum, J. W. (1972). A longitudinal study of the effects
*Fleishman, E. A. (1989a). Leadership Opinion Questionnaire (LOQ) of a human relations training program on managerial effectiveness.
examiner’s manual. Park Ridge, IL: Science Research Associates. Journal of Applied Psychology, 56, 412– 417.
*Fleishman, E. A. (1989b). Supervisory Behavior Description Question- *Hemphill, J. K. (1955). Leadership behavior associated with the admin-
naire (SBD) examiner’s manual. Park Ridge, IL: Science Research istrative reputation of college departments. Journal of Educational Psy-
Associates. chology, 46, 385– 401.
Fleishman, E. A. (1995). Consideration and structure: Another look at their *Hemphill, J. K. (1957). Leader behavior associated with the administra-
role in leadership research. In F. Dansereau & F. J. Yammarino (Eds.), tive reputations of college departments. In R. M. Stogdill & A. E. Coons
Leadership: The multiple-level approaches (pp. 51– 60). Stamford, CT: (Eds.), Leader behavior: Its description and measurement (pp. 74 – 85).
JAI Press. Columbus, OH: Bureau of Business Research.
Fleishman, E. A. (1998). Patterns of leadership behavior related to em- *Hemphill, J. K., & Coons, A. E. (1957). Development of the Leader
ployee grievances and turnover: Some post hoc reflections. Personnel Behavior Description Questionnaire. In R. M. Stogdill & A. E. Coons
Psychology, 51, 825– 834. (Eds.), Leader behavior: Its description and measurement (pp. 6 –38).
*Fleishman, E. A., & Harris, E. F. (1962). Patterns of leadership behavior Columbus, OH: Bureau of Business Research.
related to employee grievances and turnover. Personnel Psychology, 15, *Herold, D. M. (1974). Interaction of subordinate and leader characteristics
43–56. in moderating the consideration–satisfaction relationship. Journal of
Fleishman, E. A., Mumford, M. D., Zaccaro, S. J., & Levin, K. Y. (1991). Applied Psychology, 59, 649 – 651.
Taxonomic efforts in the description of leader behavior: A synthesis and Hickman, G. R. (Ed.). (1998). Leading organizations: Perspectives for a
functional interpretation. Leadership Quarterly, 2, 245–287. new era. Thousand Oaks, CA: Sage.
*Fleishman, E. A., & Peters, D. R. (1962). Interpersonal values, leadership *House, R. J. (1971). A path– goal theory of leader effectiveness. Admin-
attitudes and managerial “success.” Personnel Psychology, 15, 127–143. istrative Science Quarterly, 16, 321–338.
*Fleishman, E. A., & Salter, J. A. (1961). Humanizing relationships in a House, R. J. (1977). A 1976 theory of charismatic leadership. In J. G. Hunt
small business. Advanced Management Journal, 26, 18 –20. & L. L. Larsen (Eds.), Leadership: The cutting edge. Carbondale, IL:
*Fleishman, E. A., & Salter, J. A. (1963). Relationship between the Southern Illinois University Press.
leader’s behavior and his empathy toward subordinates. Journal of House, R. J. (1996). Path– goal theory of leadership: Lessons, legacy and
Industrial Psychology, 1, 79 – 84. a reformulated theory. Leadership Quarterly, 7, 323–352.
*Fleishman, E. A., & Simmons, J. (1970). Relationship between leadership House, R. J., & Aditya, R. N. (1997). The social scientific study of
patterns and effectiveness ratings among Israeli foremen. Personnel leadership: Quo vadis? Journal of Management, 23, 409 – 473.
Psychology, 23, 169 –172. *House, R. J., Filley, A. C., & Gujarati, D. N. (1971). Leadership style,
Goldberg, L. R. (1990). An alternative “description of personality”: The hierarchical influence, and the satisfaction of subordinate role expecta-
Big-Five factor structure. Journal of Personality and Social Psychology, tions: A test of Likert’s influence proposition. Journal of Applied Psy-
59, 1216 –1229. chology, 55, 422– 432.
Greenberg, J., & Lind, E. A. (2000). The pursuit of organizational justice: *House, R. J., Filley, A. C., & Kerr, S. (1971). Relation of leader consid-
 CONSIDERATION AND INITIATING STRUCTURE 49

eration and initiating structure to R and D subordinates’ satisfaction. Locke, E. A. (1997). The motivation to work: What we know. Advances in
Administrative Science Quarterly, 16, 19 –30. Motivation and Achievement, 10, 375– 412.
*House, R. J., & Kerr, S. (1973). Organizational independence, leader Lord, R. G., Foti, R. J., & De Vader, C. L. (1984). A test of leadership
behavior, and managerial practices: A replicated study. Journal of Ap- categorization theory: Internal structure, information processing, and
plied Psychology, 58, 173–180. leadership perceptions. Organizational Behavior and Human Perfor-
House, R. J., & Podsakoff, P. M. (1994). Leadership effectiveness. In J. mance, 34, 343–378.
Greenberg (Ed.), Organizational behavior: The state of the science (pp. Lowe, K. B., Kroeck, K. G., & Sivasubramaniam, N. (1996). Effectiveness
45– 82). Hillsdale, NJ: Erlbaum. correlates of transformation and transactional leadership: A meta-
*Hunt, J. G., & Liebscher, V. K. (1973). Leadership preference, leadership analytic review of the MLQ literature. Leadership Quarterly, 7, 385–
behavior, and employee satisfaction. Organizational Behavior & Human 425.
Decision Processes, 9, 59 –77. *Lowin, A., Hrapchak, W. J., & Kavanagh, M. J. (1969). Consideration
Hunter, J. E., & Schmidt, F. L. (1990). Methods of meta-analysis. Newbury and initiating structure: An experimental investigation of leadership
Park, CA: Sage. traits. Administrative Science Quarterly, 14, 238 –253.
*Ilgen, D. R., & Fuji, D. S. (1976). An investigation of the validity of Mathieu, J. E. (2001). “Into the box” thinking about leadership research. In
leader behavior descriptions obtained from subordinates. Journal of S. J. Zaccaro & R. J. Klimoski (Eds.), The nature of organizational
Applied Psychology, 61, 642– 651. leadership (pp. 437– 463). San Francisco: Jossey-Bass.
*Johns, G. (1978). Task moderators of the relationship between leadership *Matsui, T., Osawa, T., & Terai, T. (1975). Relations between supervisory
style and subordinate responses. Academy of Management Journal, 21, motivation and the consideration and structure aspects of supervisory
319 –325. behavior. Journal of Applied Psychology, 60, 451– 454.
*Johnson, P. O., & Bledsoe, J. C. (1973). Morale as related to perceptions *McCabe, J. B. (1986). Leadership behavior and job satisfaction of med-
of leader behavior. Personnel Psychology, 26, 581–592. ical technologists: A path– goal approach. Dissertation Abstracts Inter-
*Katerberg, R., & Hom, P. W. (1981). Effects of within-group and national, 46, 3789.
between-groups variation in leadership. Journal of Applied Psychology, *Michaelsen, L. K. (1973). Leader orientation, leader behavior, group
66, 218 –223. effectiveness and situational favorability: An empirical extension of the
*Katz, R. (1977). The influence of group conflict on leadership effective- contingency model. Organizational Behavior & Human Decision Pro-
cesses, 9, 226 –245.
ness. Organizational Behavior & Human Decision Processes, 20, 265–
*Miles, R. H., & Petty, M. M. (1977). Leader effectiveness in small
286.
bureaucracies. Academy of Management Journal, 20, 238 –250.
*Kenis, I. (1978). Leadership behavior, subordinate personality, and sat-
Mount, M. K., & Barrick, M. R. (1998). Five reasons why the “Big Five”
isfaction with supervision. Journal of Psychology, 98, 99 –107.
article has been frequently cited. Personnel Psychology, 51, 849 – 857.
Kerr, S. J., & Jermier, J. M. (1978). Substitutes for leadership: Their
Murphy, K. R., & DeShon, R. (2000). Interrater correlations do not
meaning and measurement. Organizational Behavior and Human Per-
estimate the reliability of job performance ratings. Personnel Psychol-
formance, 22, 375– 403.
ogy, 53, 873–900.
Kerr, S. J., & Schriesheim, C. (1974). Consideration, initiating structure,
*Nealy, S. M., & Blood, M. R. (1968). Leadership performance of nursing
and organizational criteria—An update of Korman’s 1966 review. Per-
supervisors at two organizational levels. Journal of Applied Psychology,
sonnel Psychology, 27, 555–568.
52, 414 – 422.
Kirkpatrick, S. A., Locke, E. A., & Latham, G. P. (1996). Implementing the
Northouse, P. G. (1997). Leadership: Theory and practice. Thousand
vision: How is it done? Polish Psychological Bulletin, 27, 93–106.
Oaks, CA: Sage.
*Kohli, A. K. (1989). Effects of supervisory behavior: The role of indi-
*Nystrom, P. C. (1978). Managers and the hi-hi leader myth. Academy of
vidual differences among salespeople. Journal of Marketing, 53, 40 –50. Management Journal, 21, 325–331.
Korman, A. K. (1966). “Consideration,” “initiating structure,” and orga- *Oaklander, H., & Fleishman, E. A. (1964). Patterns of leadership related
nizational criteria—A review. Personnel Psychology, 19, 349 –361. to organizational stress in hospital settings. Administrative Science
*Korsgaard, M. A., Schweiger, D. M., & Sapienza, H. J. (1995). Building Quarterly, 8, 520 –532.
commitment, attachment, and trust in strategic decision-making teams: *Oldham, G. R. (1976). The motivational strategies used by supervisors:
The role of procedural justice. Academy of Management Journal, 38, Relationships to effectiveness indicators. Organizational Behavior &
60 – 84. Human Decision Processes, 15, 66 – 86.
*Lahat-Mandelbaum, B.-S., & Kipnis, D. (1973). Leader behavior dimen- *Osborn, R. N., & Hunt, J. G. (1975). Relations between leadership, size,
sions related to students’ evaluation of teaching effectiveness. Journal of and subordinate satisfaction in a voluntary organization. Journal of
Applied Psychology, 58, 250 –253. Applied Psychology, 60, 730 –735.
*Larson, L. L., Hunt, J. G., & Osborn, R. N. (1976). The great hi-hi leader *Osborn, R. N., & Vicars, W. M. (1976). Sex stereotypes: An artifact in
behavior myth: A lesson from Occams razor. Academy of Management leader behavior and subordinate satisfaction analysis? Academy of Man-
Journal, 19, 628 – 641. agement Journal, 19, 439 – 444.
*Lee, C., & Schuler, R. S. (1980). Goal specificity and difficulty and leader *Parker, T. C. (1963). Relationships among measures of supervisory be-
initiating structure as strategies for managing role stress. Journal of havior, group behavior, and situational characteristics. Personnel Psy-
Management, 6, 177–187. chology, 16, 319 –334.
Leventhal, G. S. (1980). What should be done with equity theory? New *Pennington, P. W. (1998). Principal leadership and teacher motivation in
approaches to the study of fairness in social relationship. In K. J. Gergen, secondary schools. Dissertation Abstracts International Section A: Hu-
M. S. Greenberg, & R. H. Willis (Eds.), Social exchange: Advances in manities & Social Sciences, 58, 2943.
theory and research (pp. 27–55). New York: Plenum. *Pettman, P. J. (1977). Consideration, initiating structure, field dependence
Liden, R. C., Sparrowe, R. T., & Wayne, S. J. (1997). Leader–member and student evaluation of teacher behavior. Dissertation Abstracts In-
exchange theory: The past and potential for the future. Research in ternational, 37, 5418.
Personnel and Human Resources Management, 15, 47–119. *Petty, M. M., & Bruning, N. S. (1980). A comparison of the relationships
Likert, R. (1961). New patterns of management. New York: McGraw-Hill. between subordinates’ perceptions of supervisory behavior and mea-
50 JUDGE, PICCOLO, AND ILIES

sures of subordinates’ job satisfaction for male and female leaders. pirical tests. Organizational Behavior & Human Decision Processes, 15,
Academy of Management Journal, 23, 717–725. 297–321.
*Petty, M. M., & Lee, G. K. (1975). Moderating effects of sex of super- Schriesheim, C. A., & Kerr, S. (1974). Psychometric properties of the Ohio
visor and subordinate on relationships between supervisory behavior and State leadership scales. Psychological Bulletin, 81, 756 –765.
subordinate satisfaction. Journal of Applied Psychology, 60, 624 – 628. *Schriesheim, C. A., Kinicki, A. J., & Schriesheim, J. F. (1979). The effect
Podsakoff, P. M., MacKenzie, S. B., & Bommer, W. H. (1996). Meta- of leniency on leader behavior descriptions. Organizational Behavior &
analysis of the relationships between Kerr and Jermier’s substitutes for Human Decision Processes, 23, 1–29.
leadership and employee job attitudes, role perceptions, and perfor- *Schriesheim, C., & von Glinow, M. A. (1977). The path– goal theory of
mance. Journal of Applied Psychology, 81, 380 –399. leadership: A theoretical and empirical analysis. Academy of Manage-
*Pool, S. W. (1996). The path– goal theory of organizational leadership: ment Journal, 20, 398 – 405.
An empirical analysis of leader behavior by integrating leader substitute *Schuler, R. S. (1976). Participation with supervisor and subordinate
theory. Dissertation Abstracts International Section A: Humanities & authoritarianism: A path– goal theory reconciliation. Administrative Sci-
Social Sciences, 57, 1225. ence Quarterly, 21, 320 –325.
Quiñones, M. A., Ford, J. K., & Teachout, M. S. (1995). The relationship *Seeman, M. (1957). A comparison of general and specific leader behavior
between work experience and job performance: A conceptual and meta- descriptions. In R. M. Stogdill & A. E. Coons (Eds.), Leader behavior:
analytic review. Personnel Psychology, 48, 887–910. Its description and measurement (pp. 86 –102). Columbus, OH: Bureau
Raju, N. S., Burke, M. J., Normand, J., & Langlois, G. M. (1991). A new of Business Research.
meta-analytic approach. Journal of Applied Psychology, 76, 432– 446. *Seltzer, J., & Bass, B. M. (1990). Transformational leadership: Beyond
*Rambo, W. W. (1958). The construction and analysis of a leadership initiation and consideration. Journal of Management, 16, 693–703.
behavior rating form. Journal of Applied Psychology, 42, 409 – 415. *Shapiro, M. I. (1971). Initiating structure and consideration: A situation-
*Rider, L. H. (1975). Leader behavior, locus of control and consultation ist’s view of the efficacy of two styles of leadership. Dissertation
effectiveness of school psychologists. Dissertation Abstracts Interna- Abstracts International, 31, 4382– 4383.
tional, 35, 5134. *Sheih, C. S.-M. (1999). The effect of perceived leadership behavior on the
*Rooker, J. L. (1968). The relationship of need achievement and need job satisfaction of subordinates in Taiwan’s university libraries. Disser-
affiliation to leader behavior. Dissertation Abstracts International, 28, tation Abstracts International Section B: The Sciences & Engineering,
4426. 59, 3749.
Rosenthal, R. (1995). Writing meta-analytic reviews. Psychological Bul- *Sheridan, J. E., & Vredenburgh, D. J. (1978a). Predicting leadership
letin, 118, 183–192. behavior in a hospital organization. Academy of Management Journal,
*Rowland, K. M., & Scott, W. E. (1968). Psychological attributes of 21, 679 – 689.
effective leadership in a formal organization. Personnel Psychology, 21, *Sheridan, J. E., & Vredenburgh, D. J. (1978b). Usefulness of leadership
365–377. behavior and social power variables in predicting job tension, perfor-
Rush, M. C., Thomas, J. C., & Lord, R. G. (1977). Implicit leadership mance, and turnover of nursing employees. Journal of Applied Psychol-
theory: A potential threat to the internal validity of leader behavior ogy, 63, 89 –95.
questionnaires. Organizational Behavior and Human Decision Pro- *Sheridan, J. E., & Vredenburgh, D. J. (1979). Structural model of lead-
cesses, 20, 93–110. ership influence in a hospital organization. Academy of Management
Sagie, A., & Koslowsky, M. (1993). Detecting moderators with meta- Journal, 22, 6 –21.
analysis: An evaluation and comparison of techniques. Personnel Psy- *Sheridan, J. E., Kerr, J. L., & Abelson, M. A. (1982). Leadership activa-
chology, 46, 629 – 640. tion theory: An opponent process model of subordinate responses to
Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection leadership behavior. In J. G. Hunt, U. Sekaran, & C. A. Schriesheim
methods in personnel psychology: Practical and theoretical implications (Eds.), Leadership: Beyond establishment views (pp. 122–141). Carbon-
of 85 years of research findings. Psychological Bulletin, 124, 262–274. dale, IL: Southern Illinois University Press.
Schmidt, F. L., Hunter, J. E., Pearlman, K., & Hirsh, H. R. (1985). Forty *Sims, H. P., & Szilagyi, A. D. (1975). Leader structure and subordinate
questions about validity generalization and meta-analysis. Personnel satisfaction for two hospital administrative levels: A path analysis ap-
Psychology, 38, 697–798. proach. Journal of Applied Psychology, 60, 194 –197.
Schmidt, F. L., Viswesvaran, C., & Ones, D. S. (2000). Reliability is not *Skinner, E. W. (1969). Relationships between leadership behavior pat-
validity and validity is not reliability. Personnel Psychology, 53, 901– terns and organizational–situational variables. Personnel Psychology,
912. 22, 489 – 494.
*Schriesheim, C. A. (1979). The similarity of individual directed and group *Spitzer, M. E., & McNamara, W. J. (1964). A managerial selection study.
directed leader behavior descriptions. Academy of Management Journal, Personnel Psychology, 17, 19 – 40.
22, 345–355. *Stanley, D. M. (1975). The effects of consideration and initiating structure
*Schriesheim, C. A. (1982). The great high Consideration– high Initiating as moderated by incremental influence. Dissertation Abstracts Interna-
Structure leadership myth: Evidence on its generalizability. Journal of tional, 35, 3639.
Social Psychology, 116, 221–231. Steiger, J. H. (1980). Tests for comparing elements of a correlation matrix.
*Schriesheim, C. A., Cogliser, C. C., & Neider, L. L. (1995). Is it Psychological Bulletin, 87, 245–251.
“trustworthy?”: A multiple-levels-of-analysis reexamination of an Ohio *Stinson, J. E., & Johnson, T. W. (1975). The path– goal theory of
state leadership study, with implications for future research. Leadership leadership: A partial test and suggested refinement. Academy of Man-
Quarterly, 6, 111–145. agement Journal, 18, 242–252.
*Schriesheim, C. A., & DeNisi, A. S. (1981). Task dimensions as moder- Stogdill, R. M. (1950). Leadership, membership and organization. Psycho-
ators of the effects of instrumental leadership: A two-sample replicated logical Bulletin, 47, 1–14.
test of path– goal leadership theory. Journal of Applied Psychology, 66, Stogdill, R. M. (1963). Manual for the Leader Behavior Description
589 –597. Questionnaire, Form XII. Columbus, OH: Bureau of Business Research,
*Schriesheim, C. A., House, R. J., & Kerr, S. (1976). Leader initiating Ohio State University.
structure: A reconciliation of discrepant research results and some em- *Stogdill, R. M., & Shartle, C. L. (1956). Methods in the study of admin-
 CONSIDERATION AND INITIATING STRUCTURE 51

istrative leadership (Research Monograph No. 80). Columbus, OH: *Vecchio, R. P. (1980). Alternatives to the least preferred co-worker
Bureau of Business Research. construct. Journal of Social Psychology, 112, 261–269.
*Stogdill, R. M., Goode, O. S., & Day, D. R. (1963a). The leader behavior *Vecchio, R. P. (1981). Situational and behavioral moderators of subor-
of United States Senators. Journal of Psychology, 56, 3– 8. dinate satisfaction with supervision. Human Relations, 34, 947–963.
*Stogdill, R. M., Goode, O. S., & Day, D. R. (1963b). The leader behavior Weissenberg, P., & Kavanagh, M. J. (1972). The independence of initiating
of corporation presidents. Personnel Psychology, 16, 127–132. structure and consideration: A review of the evidence. Personnel Psy-
*Stogdill, R. M., Goode, O. S., & Day, D. R. (1964). The leader behavior chology, 25, 119 –130.
of presidents of labor unions. Personnel Psychology, 17, 49 –57. Wofford, J. C., & Liska, L. Z. (1993). Path– goal theory of leadership: A
*Swanson, R. G., & Johnson, D. A. (1975). Relation between peer per- meta-analysis. Journal of Management, 19, 857– 876.
ception of leader behavior and instructor-pilot performance. Journal of *Yoho, S. K. (1995). Follower-centered leadership: An investigation of
Applied Psychology, 60, 198 –200. leader behavior, leader power, follower competency, and follower job
performance in leader–follower relationships. Dissertation Abstracts
*Szilagyi, A. D., & Keller, R. (1976). A comparative investigation of the
International Section A: Humanities & Social Sciences, 56, 0629.
supervisory behavior description questionnaire (SBDQ) and the revised
Yukl, G. (1998). Leadership in organizations. Upper Saddle River, NJ:
leader behavior description questionnaire (LBDQ-Form XII). Academy
Prentice Hall.
of Management Journal, 19, 642– 650.
*Yukl, G., & Lepsinger, R. (1990). Preliminary report on validation of the
*Szilagyi, A. D., & Sims, H. P. (1974). An exploration of the path– goal
Managerial Practices Survey. In K. E. Clark & M. B. Clark (Eds.),
theory of leadership in a health care environment. Academy of Manage- Measures of leadership (pp. 223–237). West Orange, NJ: Leadership
ment Journal, 17, 622– 634. Library of America.
*Tscheulin, D. (1973). Leader behavior measurement in German industry. Yukl, G., & Van Fleet, D. D. (1992). Theory and research on leadership in
Journal of Applied Psychology, 57, 28 –31. organizations. In M. D. Dunnette & L. M. Hough (Eds.), Handbook of
*Tucker, J. H. (1977). Leadership in autonomous group environments, industrial and organizational psychology (Vol. 2, pp. 147–197). Palo
task/person orientation and interpersonal competence. Dissertation Ab- Alto, CA: Consulting Psychologists Press.
stracts International, 37, 4202– 4203. *Yunker, G. W., & Hunt, J. G. (1976). An empirical comparison of the
*Valenzi, E. R., & Dessler, G. (1978). Relationships of leader behavior, Michigan Four-Factor and Ohio State LBDQ leadership scales. Orga-
subordinate role ambiguity and subordinate job satisfaction. Academy of nizational Behavior & Human Decision Processes, 17, 45– 65.
Management Journal, 21, 671– 678.
*van Engen, M. L., van der Leeden, R., & Willemsen, T. (2001). Gender, Received September 3, 2002
context, leadership styles: A field study. Journal of Occupational & Revision received April 7, 2003
Organizational Psychology, 74, 581–598. Accepted May 2, 2003 䡲

View publication stats