HELP with thesis. HR/Psychology expert requested

Need for leadership as a moderator of the relationships

between leadership and individual outcomes

Reinout E. de Vries a,*, Robert A. Roe

b , Tharsi C.B. Taillieu

c

a Department of Organisational and Labour Studies, University of Western Australia,

Nedlands, Perth, WA 6907, Australia b Work and Organization Research Centre, Tilburg University, Tilburg, The Netherlands c Work and Organization Psychology, Catholic University Leuven, Louvain, Belgium

Received 4 October 2000; received in revised form 18 January 2001; accepted 8 March 2001

Abstract

Earlier research on situational leadership theories has produced little and partly contradictory

evidence about the role of situational moderator variables in explaining the relationship between

leadership and outcomes. In this article, we propose to concentrate on need for leadership as a singular

moderator of the relationships between leadership and employee outcomes. Using a sample of 958

Dutch employees from various organizations, the moderator hypothesis was tested. Need for leadership

was paired with three leadership factors and five outcome variables, generating 15 possible moderating

effects. Five of these were significant. Although the findings compare favorably with other studies using

leadership moderators, the effects are weak, and there is not much evidence that leadership–outcome

relations are reversed by need for leadership. D 2002 Published by Elsevier Science Inc.

1. Introduction

In 1948, Knickerbocker proposed a leadership theory based on the needs of subordinates.

He postulated, for example, that when ‘‘the objectives of the group require greater diversity of

effort and greater coordination, the need for a leader will increase’’ (Knickerbocker, 1948,

p. 29). Furthermore, he believed that the subordinates’ leadership needs, in their turn,

determined whether a leader could be effective or not. After a short flurry of studies on

1048-9843/02/$ – see front matter D 2002 Published by Elsevier Science Inc.

PII: S1048-9843(02)00097-8

* Corresponding author. Tel.: +61-8-9380-1425.

E-mail address: [email protected] (R.E. de Vries).

The Leadership Quarterly 13 (2002) 121–137

concepts related to need for leadership such as subordinates’ leadership expectations and

leadership preferences (Foa, 1957; Hunt & Liebscher, 1973; Mannheim, Rim, & Grinberg,

1967; Yukl, 1971), Hunt and Liebscher (1973) concluded that ‘‘a more refined way of

classifying situational differences would seem a logical next step’’ (p. 76). Around the same

time, several leadership theories sprang to life proposing moderating effects of a range of

subordinate, task, and organizational factors on the relationships between leadership

characteristics and individual or organizational outcome variables. Examples of these theories

include Fiedler’s (1967) contingency theory, Hersey and Blanchard’s (1969, 1982) situational

leadership theory, House’s (1971) path–goal theory, and Kerr and Jermier’s (1978) sub-

stitutes for leadership theory.

Many studies have been conducted to test the validity of situational theories of leadership,

but few have shown support of the moderator hypotheses postulated. There is some positive

evidence on moderators in the contingency theory (Schriesheim, Tepper, & Tetrault, 1994)

and path–goal theory (Wofford & Liska, 1993). Yet, the effects found are few and of modest

strength. The same applies to the substitutes theory (Howell & Dorfman, 1986; Podsakoff,

MacKenzie, & Fetter, 1993; Williams et al., 1988). Summarizing the findings on the

substitutes for leadership theory, De Vries (1997) noted a general absence of significant

moderating effects. Only 9.3% of the published hypothesized interactions were significant

and most of these interactions were not in the predicted direction.

In spite of these disappointing findings, it would be wrong to conclude that research on

situational moderators is not appropriate. The lack of findings may actually be due to the

often too refined moderators used in the situational leadership theories. In this article, we

argue for a more parsimonious situational leadership model, which uses, following the

example set by earlier literature (e.g., Knickerbocker, 1948; Seers & Graen, 1984), ‘‘need for

leadership’’ as the single most important moderator of the effect of leadership. It is in line

with the basic idea of Kerr and Jermier’s (1978) substitute theory, i.e., that under certain

conditions, leadership may have little or no added value.

However, we assume that employees are normally aware of what the leader can and cannot

contribute. Therefore, the situational factors known from the moderator literature, such as task

ambiguity or employee level of expertise, probably have their effect on leadership effective-

ness through the subordinate’s assessment of what they imply for the role of the leader. Thus,

according to our conception, need for leadership is a ‘‘catch-all’’ variable, which mediates the

effects of other situational variables on the relationships between leadership and personal and

work outcomes.

The aim of this article is to introduce the concept of need for leadership and to examine its

role as a moderator of the relationships between leadership characteristics and subordinate

outcomes. Our hypothesis is: Need for leadership has a moderating effect on the relation

between leadership characteristics and individual outcomes. With low need for leadership, the

relation between leadership and outcomes is weaker compared with high need for leadership.

Adopting the perspective of path–goal theory (House, 1971), we define need for

leadership as the extent to which an employee wishes the leader to facilitate the paths

towards individual, group, and/or organizational goals. Need for leadership should be

understood as an acquired need, or rather a quasi-need in the sense of Lewin (1951), which

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137122

can change with circumstances. The need comes about through socialization processes and is

provoked in social settings that entail expectations regarding hierarchical relations and

demands for action. The need is contextual in the sense that it depends on the person’s

assessment of the particular setting. If the setting is such that the person lacks the needed

competence or feels insecure, s/he will want the leader to act and help. When a change of

setting occurs, the person may feel more confident in his/her skills, feel more secure, and lack

the desire for an intervention by a leader.

Need for leadership seems to be of immediate relevance for what happens in the interaction

between the leader and the subordinate. The subordinate will welcome a leader’s intervention

when s/he considers it as instrumental to the achievement of a work goal. However, an

unwanted intervention, or one considered to be unnecessary, is likely to result in opposition or

neglect at the side of the subordinate. Other situational variables proposed in leadership

theory seem to be more remotely linked to the leader–subordinate interaction. For instance,

task ambiguity is not likely to have a direct impact because what happens in the interaction

between the leader and the subordinate would depend on the latter’s interpretation of the

situation. If s/he would feel able to resolve the ambiguity by finding out oneself, or asking

others, there might be little room for a positive effect of the leader’s intervention. Thus, need

for leadership seems to be a more proximal variable to the leader–subordinate interaction,

whereas most other variables known from the literature seem to be more distal, and, hence,

may have a less clear effect.

An advantage of the need for leadership concept is that it can be linked to a variety of

personal, task, and organizational factors at the same time. Thus, a specific constellation of such

factors — a high degree of personal competence, a job with limited variety and autonomy, little

task-provided feedback, and a heavy reliance on written procedures — may produce a certain

level of need for leadership. Instead of investigating the moderating effects of all proposed

situational factors separately as well as in their interaction, one might study how these factors

relate to need for leadership and concentrate on the moderating effect of this variable only. Such

an investigation would not only make sense theoretically, as one would like to understand how

the subordinate responds to his/her work setting as a whole, but also methodologically, since a

simultaneous test of multiple moderating effects is almost impossible to conduct.

Some studies have been conducted on constructs that resemble need for leadership. Most

of these have provided limited or no information on the definition and item content of these

constructs and some have used unreliable scales, small samples, and inadequate statistical

techniques. Furthermore, the subordinate leadership needs have usually been explored as part

of a wider study. Consequently, although these studies have been precursors of what may be

considered a promising area of research, they have not been integrated in the mainstream

leadership research.

A survey of the literature provided us with the following closely related concepts:

subordinates’ leadership expectations (Foa, 1957; House, Filley, & Gujarati, 1971; Man-

nheim et al., 1967), leadership preference (Hunt & Liebscher, 1973; Yukl, 1971), need for

closer supervision (Ashkanasy & Gallois, 1994), need for supervision (De Vries, Roe, &

Taillieu, 1998; Martin, 1983), and leadership need strength (Seers & Graen, 1984).

Leadership studies that used more generalized ‘‘needs’’ include studies on need for clarity

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 123

(Ivancevich & Donnelly, 1974; Keller, 1989), need for structure (Stoker & De Jong, 1996),

need for autonomy (Emans & Radstaak, 1990; Landeweerd & Boumans, 1994), and need for

independence (Kerr & Jermier, 1978).

Recent studies that explicitly posited and tested the moderator hypothesis using a variable

resembling need for leadership (De Vries et al., 1998; Emans & Radstaak, 1990; Keller, 1989;

Landeweerd & Boumans, 1994; Stoker & De Jong, 1996) have provided some empirical

support. In those cases in which a significant moderator effect was detected, i.e., whenever

employees needed clarity in their job (Keller, 1989), needed more structure (Stoker & De

Jong, 1996), or needed less autonomy (Emans & Radstaak, 1990; Landeweerd & Boumans,

1994), leadership was more strongly related to subordinate outcomes than in cases in which

employees did not need clarity or did need autonomy.

However, De Vries’s et al. (1998) study showed opposite effects. In a study using two

samples of insurance agents, they found higher need for supervision to be associated with a

weaker (positive) relation between task-oriented leadership and work stress. De Vries et al.

explain the findings by arguing that with little need for supervision and high task-oriented

leadership, the insurance agents may have felt under pressure and without the possibility of

turning the supervisor’s directions into work activity.

Thus, the studies described above, although providing a more follower-centered view of the

leadership process, are at best inconclusive. However, theoretically, advances have been made

by concentrating on characteristics of the subordinates and on how these characteristics may

influence leadership–outcomes relationships. A study using a large sample from diverse

settings with a reliable measurement instrument may provide a better test of the hypothesis that

need for leadership moderates the relationships between leadership and outcome variables.

2. Method

2.1. Sample

The data for testing the hypothesis were gathered in a crossfunctional and crossorganiza-

tional sample of 958 employees, i.e., the sample contained people working in different jobs

within different firms. It was drawn in the following way. First, a random sample of 4523

households in the Dutch region Middle-Brabant was contacted by telephone in order to find

out whether they included an employee who was willing to participate in the study. Second, a

questionnaire was sent to 2000 employees who orally agreed to participate. A total of 958

(47.9%) usable questionnaires was returned.

Of the respondents, 291 (30.4%) were female and 665 (69.4%) were male. The average

age was 39.2 years (S.D. = 9.6). As for the educational level, 3.4% completed junior high

school, 15.4% completed high school, 14.7% completed lower occupational training, 34.1%

completed middle occupational training, 24.3% completed higher occupational training, and

8.1% were university graduates. Although the service sector is somewhat oversampled

(51.6% in the sample vs. 32.7% in the Dutch workforce) and banks/insurance and trade/hotels

are somewhat undersampled (5.3% vs. 15.9% and 8.4% vs. 19.2%, respectively), generally,

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137124

the sample is representative of the Dutch (Centraal Bureau voor Statistiek, 1996) and the

Middle Brabant workforce (Samenwerkingsverband Midden Brabant, 1996) with regard to

the sectors farming (2.6%), mining (0.2%), industry (16.9%), public utilities (2.1%),

construction (9.3%), and transport and storage (3.6%).

2.2. Need for leadership

All variables reported in this study are scored on 1–5 Likert-like scales. Unless otherwise

reported, the answering categories range from completely disagree (1) to completely agree (5).

The need for leadership measurement instrument is presented in Appendix A. Its answering

categories range from not at all (1) to a lot (5). The original instrument is in the Dutch language;

here, it is presented in an English translation. This scale, which has 17 items with a common

stem but a different ending, describes a series of specific work goals for which the subordinate

would need a contribution from his/her leader. To clarify what is meant by ‘‘leader,’’ four terms

are used referring to people in leadership roles, i.e., the Dutch equivalents supervisor, boss,

manager, and leader. De Vries (1997) has shown that the scale is unidimensional and has high

reliabilities in studies among agents of an insurance company (Cronbach’s alpha=.91) and

employees of three municipalities (Cronbach’s alpha=.92). Cronbach’s alpha in the present

sample is .93, which confirms that the measure is homogeneous and unidimensional.

Need for leadership’s convergent and discriminant validity has been tested using Martin’s

(1983) ‘‘effects of changes in supervision’’ and De Vries et al. (1998) ‘‘need for supervision.’’

All relations were in the predicted direction. Need for leadership has a significant positive

relation with the following items of Martin: ‘‘Results of my work performance would be

better when my leader would provide more leadership’’ and ‘‘I would be more satisfied when

my leader would provide more leadership.’’ It has a significant negative relation with

Martin’s items ‘‘Result of my work performance would be better when my leader would

provide less leadership,’’ ‘‘Results of my work performance would be better when there

would not be a leader anymore,’’ ‘‘I would be more satisfied when my leader would provide

less leadership,’’ and ‘‘I would be more satisfied when there would not be a leader anymore.’’

As predicted, there was no relation between need for leadership and effects of changes in

supervision for the following two items of Martin: ‘‘Results of my work performance would

be better when there would be another leader’’ and ‘‘I would be more satisfied when there

would be another leader.’’ Furthermore, need for leadership has a significant positive relation

with De Vries et al. (1998) need for supervision.

The mean and standard deviation of the need for leadership scale are included in Table 1,

which contains descriptives and intercorrelations for all variables in this study. Compared to

the midpoint (i.e., 3), on the average, employees do not seem to have a strong need for

leadership (m = 2.63).

2.3. Leadership

The five leadership scales included in this study are human-oriented leadership, task-

oriented leadership, charismatic leadership, leader expertise, and leader’s encouragement of

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 125

self-management. The scales for human-oriented leadership and task-oriented leadership

were derived from a Dutch version (Syroit, 1979) of Fleishman’s (1953) consideration and

initiating structure scales of the Supervisory Behavior Description Questionnaire (SBDQ).

The autocratic and punitive elements in initiating structure have been reduced markedly

during translation, resulting in an instrument that does not have the negative connotations of

the original SBDQ, but which does retain some of its better properties, such as the almost

nonexistent correlation between consideration and initiating structure (in the Dutch version

respectively called human-oriented and task-oriented leadership). As a contrast, the LBDQ

and LBDQ-XII show substantial correlations between these two scales (Levanoni & Knoop,

1985; Schriesheim, House, & Kerr, 1976; Valenzi & Dessler, 1978). In the Dutch version of

the SBDQ, human-oriented leadership is measured using 14 items, and task-oriented

Table 1

Descriptives, reliabilities (on diagonal), and correlations (pairwise correlations; minimum = 671, maximum = 934)

of variables a

No.

respondents

No.

items Mean S.D. NL HL TL CL LE ES

1. NL 871 17 2.63 0.80 .93

2. HL 797 14 3.45 0.77 .12** .92

3. TL 805 10 3.06 0.69 .21** .10** .84

4. CL 783 11 2.95 0.84 .24** .73** .43** .93

5. LE 809 5 2.83 0.91 .23** .55** .40** .81** .87

6. ES 813 6 2.94 0.75 .29** .57** .52** .75** .55** .79

7. JS 900 11 3.87 0.58 � .04 .49** .09* .48** .34** .34** 8. OC 928 6 3.67 0.88 .01 .45** .11** .46** .34** .35**

9. WS 934 7 3.02 0.73 .07* � .21** .19** � .09* � .14** .03 10. RC 928 8 2.44 0.68 .09** � .26** .13** � .15** � .21** � .00 11. P 908 8 3.49 0.58 � .03 � .02 .08* � .02 � .15** .07* 12. SU 750 – 0.00 1.00 .09* .95** � .04 .63** .36** .57** 13. IS 750 – 0.00 1.00 .18** .25** .31** .64** .87** .27**

14. ST 750 – 0.00 1.00 .20** .01 .90** .36** .19** .64**

15. N � SU 717 – 0.00 1.00 � .00 � .03 � .03 � .04 .00 � .07 16. N � IS 717 – 0.00 1.00 .07 .03 � .05 .02 .00 � .02 17. N � ST 717 – 0.00 1.00 .01 � .07 � .12** � .12** � .10** � .12** 18. CMF 755 – 0.00 1.00 .09* .28** .25** .32** .18** .32**

a All items are scored on a Likert (1–5) scale, except for the leadership and common method factors, which

are based on factor loadings (variables 12–14 and 18) and the interaction terms (variables 15–17), which are

(standardized) multiplications of (standardized) leadership and need for leadership scales; NL= need for

leadership, HL= human-oriented leadership, TL= Task-oriented leadership, CL= Charismatic leadership,

LE = Leader expertise, ES = Encouragement of self-management, JS = Job satisfaction, OC = Organizational

commitment, WS = Work stress, RC = Role conflict, P= Performance, SU = Leader’s support, IS = Leader’s

inspirational skills, ST = Leader’s structure, N � SU = Need for leadership � Leader’s support, N � IS = Need for leadership � Leader’s inspirational skills, N � ST = Need for leadership � Leader’s structure, CMF = Common method factor.

* p < .05.

** p < .01.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137126

leadership using 10 items. Cronbach’s alphas in this study are .92 (human-oriented leadership)

and .84 (task-oriented leadership).

The scale measuring charismatic leadership is an 11-item version of the inspirational

leadership scale of Den Hartog, Van Muijen and Koopman’s (1997), which is a Dutch version

of Bass’ (1985) transformational leadership item pool. We renamed the scale ‘‘charismatic

leadership,’’ because the items strongly resemble those derived from the first factor in Bass’

(1985, pp. 207–210) original factor analyses of the transformational leadership scales, which

was originally named ‘‘charismatic leadership.’’ In our study, this scale has a Cronbach’s

alpha of .93.

The leader expertise scale is a five-item version of the three-item scale with the same name

by Podsakoff, Todor, and Schuler (1983), which, in their study, had an internal consistency

JS OC WS RC P SU IS ST N � SU N � IS N � ST

.81

.71** .78

� .23** � .25** .77 � .30** � .37** .57** .75 .05 � .05 .30** .32** .77 .49** .41** � .19** � .21** .04 – .19** .26** � .13** � .19** � .22** .00 – .09* .08* .25** .22** .18** .00 .00 –

.01 � .00 � .08* � .06 � .12** � .05 .03 � .06 –

.10 * .08* .03 � .04 .06 .03 .01 � .06 � .02 – � .03 .04 .02 � .02 � .02 � .06 � .06 � .12** .02 .08* – .46** .34** � .06 � .03 .23** .28** .05 .26** � .03 .08* .04

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 127

reliability of .80. The Cronbach’s alpha of the five leader expertise items is .87 in our study.

An example of an item measuring leader expertise is: ‘‘Because of my supervisor’s

competence, I usually go along with his/her advice on how I should do my job.’’

Leader’s encouragement of self-management is a shortened version of the 22-items

instrument of Manz and Sims (1987), which, in their study, contained six scales. Because

the factor analysis conducted by Manz and Sims showed one strong first factor, which

explained 48.7% of the variance in the 22 items (seven times the variance explained in the

second factor), we decided to include from each of the original scales one semantically

most representative item in our study. The Cronbach’s alpha of the six-item scale in this

study is .79. The items describe encouraging behaviors of a supervisor towards a sub-

ordinate in order to make the subordinate go over activities before attempting them, de-

fine own goals, be self-critical, reinforce own behavior, expect high performance, and evaluate

own performance.

Correlations between the scales (Table 1) reveal strong relations between some of the

leadership scales. For instance, charismatic leadership correlates .81 with leader expertise, .75

with leader’s encouragement of self-management, and .73 with human-oriented leadership

(all p’s < .001). To ensure independence of the predictors in the regression analysis, we

conducted a Principal Component Analysis (PCA) on the leadership behavior items. Based on

the scree plot and interpretability of the factors, three factors with eigenvalues > 2 were

retained for further analysis. 1 The factors explained 51% of the variance in the 46 leadership

items. The first 17 high loading (>.50) items on the first factor contained all 14 human-

oriented leadership items, 2 charismatic leadership items, and 1 leader’s encouragement of

self-management item. We decided to rename this factor ‘‘leader’s support.’’

Of the second factor, the 10 highest loading items (>.50) contained five leader

expertise items and five charismatic leadership items. We renamed this factor ‘‘leader’s

inspirational skills’’ to reflect this mixture of expertise and charisma. All of the task-

oriented leadership items had high loadings (>.50) on the third factor, together with three

leader’s encouragement of self-management items. We renamed this factor ‘‘leader’s

structure.’’ The factor scores were retained for further analysis. As may be seen in

Table 1, the leadership factors correlate strongly with the respective leadership scales

through which they obtained their names. Leader’s support correlates .95 with human-

oriented leadership, leader’s inspirational skills correlates .87 with leader expertise and .64

with charismatic leadership, and leader’s structure correlates .90 (all p’s < .001) with task-

oriented leadership. Note that the original charismatic leadership scale correlates strongly

with both leader’s support and leader’s inspirational skills, and that the leader’s

encouragement of self-management scale correlates strongly with leader’s support and

leader’s structure.

1 Information about the factor analyses on the (Dutch) leadership items and ‘‘common method variance’’ items

(see below) can be obtained from the first author.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137128

2.4. Criteria

The dependent variables used in this study are: job satisfaction, organizational commit-

ment, work stress, role conflict, and performance. The scales for measuring job satisfaction,

organizational commitment, and work stress are adapted from Taillieu (1987). The job

satisfaction scale contains 11 items, which denote the degree of satisfaction derived from the

amount of variation, responsibility, autonomy, etc. in the job. In the present study, it has a

Cronbach’s alpha of .81. The scale for organizational commitment consists of six items in this

study, such as: ‘‘I would change company, if another were to offer me a higher salary

(reversed scored)’’ and ‘‘I praise my organization when talking to acquaintances.’’ Cronba-

ch’s alpha is .78. The scale measuring work stress consists of seven items. Examples include:

‘‘I have to hurry to finish my work in time’’ and ‘‘I often cannot cope with the amount of

work.’’ Cronbach’s alpha is .77.

Role conflict is viewed in terms of incompatibility of demands in the form of conflict

between organizational demands and own values, problems of personal resource allocation,

conflict between obligations to several other people, and conflict between excessively

numerous or difficult tasks. It was measured using the instrument of Rizzo, House, and

Lirtzman (1970) and consists of eight items. Alpha coefficients reported range from .56 to .82

(Cook, Hepworth, Wall, & Warr, 1981). In our study, it has a Cronbach’s alpha of .75.

A self-report measure of performance was used containing eight items. We used a self-

report measure, because the nature of the ‘‘crossorganizational’’ research prevented us from

obtaining objective and comparable performance data. The scale is a composite of a task- and

role-performance measure of Roe, Zinovieva, Dienes, and Ten Horn (2000). It is an indirect

performance measure that captures a person’s appraisal of the comparison of his/her

performance with the performance of others with similar tasks and roles.

The task-performance component measures the perception of an employee of his own

performance according to the supervisor and compared to others in the team. An example of a

task-performance item is: ‘‘People know me as someone who performs better than my

colleagues.’’ Role-performance measures the function of an employee compared to other

team members in terms of the amount of workload taken and the number of times colleagues

ask for advice. An example of a role-performance item is: ‘‘When facing difficulties, my

colleagues often ask me for advice.’’ In Roe et al.’s (2000) research, the composite measure of

performance was found to have adequate reliability (alpha’s between .72 and .80), and it was

positively related to job involvement and effort. In our research, the composite measure has a

Cronbach’s alpha of .77.

2.5. Analysis

The use of a large cross-sectional sample of the workforce makes it almost impossible to

obtain independent and objective sets of predictor and outcome data. The downside of the

strategy used in this research is that same-source data may result in inflated relationships

between the constructs used. For instance, Doty and Glick’s (1998) investigation of common

methods bias shows that most observed relationships in monomethod research are approx-

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 129

imately 26% more positive than the true relationships. To correct for same-source bias, we

used a procedure suggested by De Vries et al. (1998). The procedure relies on the availability

of other same-source data in the study.

The present study contains a number of variables that are not reported in this article, such

as control (Zijlstra, Den Hoedt, & De Vries, 2000) and several individual, team, and

organizational characteristics, such as authoritarianism, self-efficacy, cohesion, and organ-

izational inflexibility (see footnote 1). The items are first screened for correlation with

‘‘objective’’ variables, such as age, gender, gender of the supervisor, and number of years of

supervision. Items that are significantly related to these objective variables were excluded

from further analysis to prevent the common method factor from extracting systematic

variance that might be attributable to objective factors. The remaining 22 items (not more

than two per construct) were subjected to an unrotated PCA. In line with Hartman’s one-

factor procedure (Podsakoff & Organ, 1986), the first factor is assumed to represent the

common methods factor. In the PCA, it explained 10% of the variance. The factor scores were

saved and retained for further analysis.

Consecutively, moderating effects were analysed using a modified version of the

Hierarchical or Moderated Multiple Regression (MMR) procedure, suggested by Howell,

Dorfman, and Kerr (1986). In the first step, the common method factor is entered in the

regression equation. In the second step, the leadership factors and need for leadership are

entered. In the third step, all three interactions between the leadership factors and need for

leadership are entered. Jaccard, Turrisi, and Wan (1990) have warned against using more than

one product term in a regression equation, because larger samples would be needed to provide

the same power. Since the regression results were obtained using at least 660 respondents, we

felt it appropriate to use all interactions at once.

MMR is generally biased against finding moderating effects in field studies (Fisicaro &

Tisak, 1994). The probability of finding interactions in field studies is especially low because of

the following: (1) In contrast with experimental studies, field studies most often deal with ‘‘fan-

shaped’’ or ‘‘nonsymmetrical’’ interactions (i.e., interactions that differ in steepness but not in

the direction of the regression lines) that have smaller moderating effects than ‘‘crossover’’ or

‘‘symmetrical’’ interactions (i.e., interactions that are positive at one level and negative at the

other level of the moderating variable). Fan-shaped interactions are characterized by smaller

moderating effects and consequently are harder to find (McClelland & Judd, 1993). (2)

Because experimental studies oversample extreme observations, standard errors of the estimate

of the coefficient for the interaction are much lower than in field studies. With higher standard

errors in field studies, the chances of committing a Type II error of rejecting hypotheses are

much greater (McClelland & Judd, 1993). (3) Furthermore, power problems in field studies are

aggravated due to multiplication of unreliability in the interaction terms (Arnold, 1982).

The above considerations have led some to recommend the use of a less conservative alpha

level of .10 or even .20 (Pedhauzer, 1982). Although with a larger number of interactions, it

becomes more important to control for experiment-wise errors, in studies using new

instruments, it may be important to prevent throwing away the baby with the bathwater.

As a result, we decided to report the findings using both the common alpha = .05 and the less

conservative .10 level.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137130

3. Results

As shown in the correlation matrix (Table 1), the common method factor is positively

related to two of the three leadership factors, and to the criteria job satisfaction, organizational

commitment, and self-rated performance. It has a slight positive relation with need for

leadership and a negligible relation with leader’s inspirational skills, work stress, role conflict,

and the three interaction terms. The leadership factors are positively related to job satisfaction

and organizational commitment. Leader’s support and leader’s inspirational skills are

negatively related to work stress and role conflict, while leader’s structure is positively

related to these criteria. Leader’s inspirational skills is negatively related to self-rated

performance and leader’s structure positively; leader’s support is unrelated to self-rated

performance. As for need for leadership: It is positively related to all of the leadership factors,

and it has small positive correlations with work stress and role conflict.

Table 2

Stepwise moderated multiple regressions of criteria on leadership characteristics, need for leadership, and

their interactions

Criteria

Job satisfaction

(n = 662)

Organizational

commitment

(n = 666)

Work stress

(n = 662)

Role conflict

(n = 662)

Self-rated

performance

(n = 660)

Step1: Common method factor

Common method factor .47*** .35*** � .07y � .04 .23*** DR

2 .22*** .12*** .01

y .00 .05***

R 2 (Radj

2 ) .22 (.22) .12 (.12) .01 (.00) .00 (.00) .05 (.05)

Step 2: Predictors

Leader’s support .38*** .33*** � .20*** � .20*** � .03 Leader’s inspirational skills .19*** .28*** � .16*** � .22*** � .24*** Leader’s structure .01 .03 .23*** .15*** .10*

Need for leadership � .08** � .04 .02 .05 � .09* DR

2 .17*** .17*** .12*** .11*** .08***

R 2 (Radj

2 ) .39 (.38) .29 (.29) .13 (.12) .12 (.11) .13 (.12)

Step 3: Interactions

Leader’s support � Need for leadership

.02 .02 � .08* � .06 � .11**

Leader’s inspirational skills � Need for leadership

.06 y

.04 .03 � .03 .08*

Leader’s structure � Need for leadership

.00 .08* .03 � .01 � .04

DR 2

.00 .01* .01 y

.00 .02**

R 2 (Radj

2 ) .39 (.38) .30 (.29) .14 (.13) .12 (.11) .15 (.14)

* p < .05.

** p < .01.

*** p < .001. y p < .10.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 131

The results of the stepwise MMRs are shown in Table 2. Of the three leadership factors,

leader’s support and leader’s inspirational skills contribute significantly in explaining job

satisfaction. Need for leadership is negatively related to job satisfaction in the regression

equation. The third step in the MMR reveals that the interactions do not explain any

additional variance in the criteria. However, inspection of the beta’s of the interactions shows

one of the interactions to be significant at p < .10. There is a small moderating effect of need

for leadership on the relation between leader’s inspirational skills and job satisfaction. The

interaction effect may be interpreted as follows: high need for leadership is associated with a

slightly stronger relation between leader’s inspirational skills and job satisfaction than low

need for leadership.

Leader’s support and leader’s inspirational skills maintain a significant (positive) relation

with organizational commitment in the regression equation after the common method

variance is partialled out. There is a positive moderating effect of need for leadership on

the leader’s structure–organizational commitment relation. High need for leadership is

associated with a positive relation between leader’s structure and organizational commitment,

while low need for leadership is associated with no relation or even a negative relation

between leader’s structure and organizational commitment.

One moderating effect (at p < .05) is found in the relation between leader’s support and

work stress and none in the case of role conflict. Need for leadership moderates the relation

between leader’s support and work stress in the predicted direction, i.e., higher need for

leadership is associated with a stronger (negative) relation between leader’s support and

work stress. Two of the three potential moderating effects are found in the relations

between leadership factors and self-rated performance. Need for leadership is a pure

moderator of the relation between leader’s support and self-rated performance.

High need for leadership is associated with a negative relation between leader’s

support and self-rated performance, while low need for leadership is associated with no

or a positive relation between the two variables. Need for leadership weakens the re-

lation between leader’s inspirational skills and self-rated performance. High need for

leadership is associated with a weak negative or almost no relation between leader’s

inspirational skills and self-rated performance, while low need for leadership is as-

sociated with a strong negative relation between leader’s inspirational skills and self-

rated performance.

4. Discussion

The number of significant interactions found in this study represents some improvement

over the number found in the path–goal and substitutes for leadership research (Podsakoff,

MacKenzie, Ahearne, & Bommer, 1995; Podsakoff, MacKenzie, & Bommer, 1996; Podsak-

off, MacKenzie, et al., 1993; Podsakoff, Niehoff, MacKenzie, & Williams, 1993; Wofford &

Liska, 1993) and the number of significant interactions in previous research resembling need

for leadership (De Vries et al., 1998; Emans & Radstaak, 1990; Keller, 1989; Landeweerd &

Boumans, 1994; Seers & Graen, 1984; Stoker & De Jong, 1996). In 5 out of the 15 possible

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137132

interactions (33%), a significant interaction effect was found. However, the moderating

effects are generally weak. In the case of the relation between leader’s inspirational skills and

job satisfaction, leader’s structure and organizational commitment, and leader’s support and

work stress, a high need for leadership is associated with a stronger relationship between

leadership and the outcome variable, a low need for leadership with a weaker or nonexistent

relationship, offering some support for our hypothesis.

When should the leader back off and allow the subordinates to carry on without him/

her? As Podsakoff et al. (1995) pointed out, the leader does not have to back off when

leader behavior is at no point harmful to the individual and organizational outcomes.

Leader’s support and leader’s inspirational skills (a mixture of expertise and charisma) do

not seem to be harmful to any of the outcomes in this study. One exception is the negative

relation between leader’s inspirational skills and self-rated performance. However, this

negative relationship may actually point to a comparison effect. Employees that think

highly of themselves, i.e., rate themselves high on performance, may have given relatively

lower skill ratings to the leader, and vice versa. Need for leadership does not appear to

(strongly) reverse the relations between leader’s support/skills and the outcomes in this

study, thus, a leader would normally be advised to be supportive and use their skills as

much as possible.

Leader’s structure should be more carefully used, since, when used too much or in an

inappropriate manner, it may cause (too much) stress and role conflict. On the other hand, it

may have a beneficial effect on subordinate’s performance, while not hurting job satisfaction

and organizational commitment. Need for leadership does not seem to make too much

difference on the advise given to the leader, although from our study, it appears organizational

commitment may be negatively affected if the leader uses to much structuring with

subordinates who have a low need for leadership.

An area that deserves further research is the relation between need for leadership and

leadership characteristics. In this study, need for leadership is positively related to all

leadership characteristics, but the nature of this relation is not clear. Do leaders show more

support and structure (or human-oriented and task-oriented leadership) when subordinates

need them or do leaders invoke subordinates’ leadership needs with this kind of leadership?

Do leaders react to subordinates’ need for leadership by encouraging the self-management of

subordinates by exhibiting greater expertise and charisma (or leader’s inspirational skills), or

do subordinates with strong need for leadership perceive their leaders as being more

charismatic, having greater expertise, and encouraging self-management to a greater extent?

Especially the relation between need for leadership and the original charismatic leadership

scale is interesting, since scholars have maintained that charismatic leaders may develop

subordinates to greater levels of autonomy (Bass & Avolio, 1990) or even make subordinates

leaders in their own right (Yammarino, 1994). If we consider seriously the positive correlation

obtained between need for leadership and charismatic leadership, there is not yet much

support for this contention.

Although we corrected for common method variance in the MMR, future studies should

try to generate independent and more objective outcome measures. Furthermore, longit-

udinal data, ideally, should be collected to look at the effects of intraindividual changes in

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 133

need for leadership on the effectiveness of leadership. Since field studies are generally

biased against finding statistical interactions, another way to uncover the effects of need

for leadership would be through experimental studies. An example of an experimental

study is provided by De Vries (2000), who found that the combined manipulation of need

for leadership and performance information has strong effects on leadership ratings.

Ratings of leadership effectiveness and goal-orientation were much higher when respond-

ents believed. Subordinates had a high need for leadership and were performing well than

when respondents believed need for leadership or performance, or both, were low.

Regarding common method variance, our research indicates that measures such as job

satisfaction are strongly affected by it, but measures such as work stress, role conflict, and

need for leadership are weakly or not related to the common method factor used in this

study. The findings lend support to Crampton and Wagner’s (1994) and Doty and Glick’s

(1998) assertion that some areas of research may be more susceptible to common method

or percept–percept bias than others. That need for leadership in this research and need for

supervision in previous research (De Vries et al., 1998) do not show a substantial relation

with a common method factor is promising for future research.

Appendix A.

Instruction: Would you please indicate on which of the following aspects you personally

need the contribution of your supervisor/boss/manager/leader?

I need my supervisor to. . .

1. . . .set goals. 2. . . .decide what work should be done. 3. . . .transfer knowledge. 4. . . .motivate me. 5. . . .coordinate, plan and organize my work. 6. . . .maintain external contacts. 7. . . .provide me with information. 8. . . .gear all activities of the team to one another. 9. . . .create a good team spirit. 10. . . .provide me with support. 11. . . .arrange things with higher-level management. 12. . . .handle conflicts. 13. . . .give work-related feedback. 14. . . .correct mistakes. 15. . . .help solve problems. 16. . . .recognize and reward contributions. 17. . . .inspire me.

Answering categories: 1 = not at all; 2 = not much; 3 = partly; 4 = mainly; 5 = a lot.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137134

References

Arnold, H. J. (1982). Moderator variables: a clarification of conceptual, analytic, and psychometric issues.

Organizational Behavior and Human Performance, 29, 143–174.

Ashkanasy, N. M., & Gallois, C. (1994). Leader attributions and evaluations: effects of locus of control, super-

visory control, and task control. Organizational Behavior and Human Decision Processes, 59, 27–50.

Bass, B. M. (1985). Leadership and performance beyond expectations. New York: Free Press.

Bass, B. M., & Avolio, B. J. (1990). The implications of transactional and transformational leadership for

individual, team, and organizational development. Research in Organizational Change and Development, 4,

231–272.

Cook, J. D., Hepworth, S. J., Wall, T. D., & Warr, P. B. (1981). The experience of work: a compendium and review

of 249 measures and their use. London: Academic Press.

Crampton, S. M., & Wagner III, J. A. (1994). Percept–percept inflation in microorganizational research: an

investigation of prevalence and effect. Journal of Applied Psychology, 79 (1), 67–76.

Den Hartog, D. N., Van Muijen, J. J., & Koopman, P. L. (1997). Transactional versus transformational leadership:

an analysis of the MLQ. Journal of Occupational and Organizational Psychology, 70 (1), 19–34.

De Vries, R. E. (1997). Need for leadership: a solution to empirical problems in situational theories of leadership.

Unpublished dissertation, Tilburg University.

De Vries, R. E. (2000). ‘‘When leaders have character’’: Need for leadership, performance, and the attribution of

leadership. Journal of Social Behavior and Personality, 15 (3/4), 413–430.

De Vries, R. E., Roe, R. A., & Taillieu, T. C. B. (1998). Need for supervision: its impact on leadership effective-

ness. Journal of Applied Behavioral Science, 34 (4), 486–501.

Doty, D. H., & Glick, W. H. (1998). Common methods bias: does common methods variance really bias results?

Organizational Research Methods, 1 (4), 374–406.

Emans, B., & Radstaak, G. (1990). De rol van leidinggevenden bij verschillende niveaus van taakverruiming:

Onderzoek op een aantal verpleegafdelingen van een ziekenhuis. Gedrag en Organisatie, 3 (5), 376–392.

Fiedler, F. E. (1967). A theory of leadership effectiveness. New York: McGraw-Hill.

Fisicaro, S. A., & Tisak, J. (1994). A theoretical note on the stochastics of moderated multiple regression.

Educational and Psychological Measurement, 54, 32–41.

Fleishman, E. A. (1953). The description of supervisory behavior. Journal of Applied Psychology, 37, 1–6.

Foa, U. G. (1957). Relation of workers expectation to satisfaction with supervisor. Personnel Psychology, 10,

161–168.

Hersey, P., & Blanchard, K. H. (1969). Life cycle theory of leadership. Training and Development Journal, 2,

6–34.

Hersey, P., & Blanchard, K. H. (1982). Management of organizational behavior: utilizing human resources.

Englewood Cliffs, NJ: Prentice-Hall.

House, R. J. (1971). A path–goal theory of leader effectiveness. Administrative Science Quarterly, 16, 321–338.

House, R. J., Filley, A. C., & Gujarati, D. N. (1971). Leadership style, hierarchical influence, and the satisfaction

of subordinate role expectations: a test of Likert’s influence proposition. Journal of Applied Psychology, 55 (5),

422–423.

Howell, J. P., & Dorfman, P. W. (1986). Leadership and substitutes for leadership among professional and

nonprofessional workers. Journal of Applied Behavioral Science, 22, 29–46.

Howell, J. P., Dorfman, P. W., & Kerr, S. (1986). Moderator variables in leadership research. Academy of

Management Review, 11, 88–102.

Hunt, J. G., & Liebscher, V. K. C. (1973). Leadership preference, leadership behavior, and employee satisfaction.

Organizational Behavior and Human Performance, 9, 59–77.

Ivancevich, J. M., & Donnelly, J. H. (1974). A study of role clarity and need for clarity for three occupational

groups. Academy of Management Journal, 17, 28–36.

Jaccard, J., Turrisi, R., & Wan, C. K. (1990). Interaction effects in multiple regression. Newbury Park:

Sage Publications.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 135

Keller, R. T. (1989). A test of the path–goal theory of leadership with need for clarity as a moderator in research

and development organizations. Journal of Applied Psychology, 74, 208–212.

Kerr, S., & Jermier, J. M. (1978). Substitutes for leadership: their meaning and measurement. Organizational

Behavior and Human Performance, 22, 375–403.

Knickerbocker, I. (1948). Leadership: a conception and some implications. Journal of Social Issues, 4 (3), 23–40.

Landeweerd, J. A., & Boumans, N. P. G. (1994). The effect of work dimensions and need for autonomy on nurses’

work satisfaction and health. Journal of Occupational and Organizational Psychology, 67, 207–217.

Levanoni, E., & Knoop, R. (1985). Does task structure moderate the relationship of leaders’ behavior and

employees’ satisfaction? Psychological Reports, 57, 611–623.

Lewin, K. (1951). Field theory in social science: selected theoretical papers. New York: Harper and Row.

Mannheim, B. F., Rim, Y., & Grinberg, G. (1967). Instrumental status of supervisors as related to workers’

perceptions and expectations. Human Relations, 20, 387–397.

Manz, C. C., & Sims, H. P. (1987). Leading workers to lead themselves: the external leadership of self-managing

work teams. Administrative Science Quarterly, 32, 106–129.

Martin, S. (1983). Managing without managers: alternative work arrangements in public organizations. Beverly

Hills: Sage Publications.

McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties of detecting interactions and moderator effects.

Psychological Bulletin, 114, 376–390.

Pedhauzer, E. J. (1982). Multiple regression in behavioral research. New York: Holt, Rinehart, and Winston.

Podsakoff, P. M., MacKenzie, S. B., Ahearne, M., & Bommer, W. H. (1995). Searching for a needle in a

haystack: trying to identify the illusive moderators of leadership behaviors. Journal of Management, 21 (3),

422–470.

Podsakoff, P. M., MacKenzie, S. B., & Bommer, W. H. (1996). Transformational leader behaviors and substitutes

for leadership as determinants of employee satisfaction, commitment, trust, and organizational citizenship

behaviors. Journal of Management, 22, 259–298.

Podsakoff, P. M., MacKenzie, S. B., & Fetter, R. (1993). Substitutes for leadership and the management of

professionals. Leadership Quarterly, 4, 1–44.

Podsakoff, P. M., Niehoff, B. P., MacKenzie, S. B., & Williams, M. L. (1993). Do substitutes for leadership really

substitute for leadership? An empirical examination of Kerr and Jermier’s situational leadership model.

Organizational Behavior and Human Decision Processes, 54, 1–44.

Podsakoff, P. M., & Organ, D. W. (1986). Self-reports in organizational research: problems and prospects. Journal

of Management, 12, 531–544.

Podsakoff, P. M., Todor, W. D., & Schuler, R. S. (1983). Leader expertise as a moderator of the effects of

instrumental and supportive leader behaviors. Journal of Management, 9, 173–185.

Rizzo, J., House, R. J., & Lirtzman, S. I. (1970). Role conflict and ambiguity in complex organizations. Admin-

istrative Science Quarterly, 15, 150–163.

Roe, R. A., Zinovieva, I. L., Dienes, E., & Ten Horn, L. A. (2000). A comparison of work motivation in Bulgaria,

Hungary and the Netherlands: Test of a model. Applied Psychology, an International Review, 49 (4), 658–687.

Samenwerkingsverband Midden Brabant (1996). Rapportage Werkgelegenheidsenquete 1995. Tilburg: SMB/

Bureau IAP.

Schriesheim, C. A., House, R. J., & Kerr, S. (1976). Leader initiating structure: a reconciliation of discrepant

research results and some empirical tests. Organizational Behavior and Human Performance, 15, 297–321.

Schriesheim, C. A., Tepper, B. J., & Tetrault, L. A. (1994). Least preferred co-worker score, situational control,

and leadership effectiveness: a meta-analysis of contingency model performance predictions. Journal of

Applied Psychology, 79, 561–573.

Seers, A., & Graen, G. B. (1984). The dual attachment concept: a longitudinal investigation of the combination of

task characteristics and leader–member exchange. Organizational Behavior and Human Performance, 33,

283–306.

Stoker, J., & De Jong, R. (1996). Leidinggeven aan zelfstandige taakgroepen. Gedrag en Organisatie, 9 (6),

401–415.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137136

Syroit, J. (1979). Mens- en taakgerichtheid: constructie en validering van een verkorte leiderschapsschaal.

Gedrag,Tijdschrift voor Psychologie, 3, 176–192.

Taillieu, T. C. B. (1987). Meaning of work: relationships between personal and contextual factors. Oisterwijk:

Research Meeting for the European Committee for Work and Pay.

Valenzi, E. R., & Dessler, G. (1978). Relationships of leader behavior, subordinate role ambiguity and subordinate

job satisfaction. Academy of Management Journal, 21, 671–678.

Williams, M. L., Podsakoff, P. M., Todor, W. D., Huber, V. L., Howell, J. P., & Dorfman, P. W. (1988). A

preliminary analysis of the construct validity of Kerr and Jermier’s ‘‘Substitutes for Leadership’’ Scales.

Journal of Occupational Psychology, 61, 307–333.

Wofford, J. C., & Liska, L. Z. (1993). Path–goal theories of leadership: a meta-analysis. Journal of Management,

19, 857–876.

Yammarino, F. J. (1994). Indirect leadership: transformational leadership at a distance. In B. M. Bass, & B. J.

Avolio (Eds.), Improving organizational effectiveness through transformational leadership (pp. 26–47).

Thousands Oaks: Sage Publications.

Yukl, G. A. (1971). Toward a behavioural theory of leadership. Organizational Behavior and Human Perform-

ance, 6, 414–440.

Zijlstra, F. R. H., Den Hoedt, M. C., & De Vries, R. E. (2000). ‘Job control’, hoe meer, hoe beter? Een studie naar

de relatie van job control met stress. Gedrag and Organisatie, 13 (1), 1–12.

R.E. de Vries et al. / The Leadership Quarterly 13 (2002) 121–137 137