11
® Academy of Management Journal 2002, Vol. 45, No. 4, 735-744. IMPACT OF TRANSFORMATIONAL LEADERSHIP ON FOLLOWER DEVELOPMENT AND PERFORMANCE: A FIELD EXPERIMENT TALY DVIR DOV EDEN Tel Aviv University BRUCE J. AVOLIO University of Nebraska BOAS SHAMIR Hebrew University In a longitudinal, randomized field experiment, we tested the impact of transforma- tional leadership, enhanced by training, on follower development and performance. Experimental group leaders received transformational leadership training, and con- trol group leaders, eclectic leadership training. The sample included 54 military leaders, their 90 direct followers, and 724 indirect followers. Results indicated the leaders in the experimental group had a more positive impact on direct followers' development and on indirect followers' performance than did the leaders in the control group. Transformational leadership theory is a promi- nent representative of the new theories that have occupied center stage in leadership research in the last two decades. Follower development and fol- lower performance are the targeted outcomes of such leadership (Bass & Avolio, 1990). However, there has been no conceptual framework, or sys- tematic research, for examining the impact of trans- formational leadership on follov^rer development (House & Aditya, 1997). Transformational leader- ship has been shown to have a positive relationship with performance (Lowe, Kroeck, & Sivasubrama- niam, 1996). Yet a causal relationship between transformational leadership and follower perfor- mance has only rarely been demonstrated, because most prior studies have had static, correlational, or nonexperimental designs (Kirkpatrick & Locke, 1996). The present experiment focused on explor- The authors thank the Israel Defence Forces (IDF) School for Leadership Development and its consultants, and especially Yitzhak Gonen, Miki Rosenstein, Eliav Zakai, and Sigal Shavit-Traub, for their collaboration in conducting this experiment. Taly Dvir acknowledges the financial support of the Maya Fisher-David Fund, the Josef Buchmann Fund, the Israel Foundation Trustees, and the Faculty of Management at Tel Aviv University. The Lilly and Alejandro Saltiel Chair in Corporate Lead- ership and Social Responsibility supported Dov Eden's work on this experiment. ing the impact of transformational leadership on follower development and on examining its lasting causal impact on followers' performance. THEORY AND HYPOTHESES Transformational Leadership and Follower Development The past two decades have heralded some con- vergence among organizational behavior scholars concerning a new genre of leadership theories, al- ternatively referred to as "transformational," "char- ismatic," and "visionary" leadership. Despite dif- ferent emphases in each theory. House and Shamir asserted that "it can be safely concluded that there is a strong convergence of the findings from studies with charismatic leadership and those concerned with transformational and visionary leadership" (1993: 84). The full-range leadership model (Bass & Avolio, 1990) differentiates between transactional and transformational leaders. Transactional leaders ex- ert influence by setting goals, clarifying desired outcomes, providing feedback, and exchanging rewards for accomplishments. Transformational leaders exert additional influence by broadening and elevating followers' goals and providing them with confidence to perform beyond the expecta- tions specified in the implicit or explicit exchange agreement. Transformational leaders exhibit char- 735

IMPACT OF TRANSFORMATIONAL LEADERSHIP ON ... · PDF fileExperimental group leaders received transformational leadership training, and con-trol group leaders, ... relationships between

Embed Size (px)

Citation preview

® Academy of Management Journal2002, Vol. 45, No. 4, 735-744.

IMPACT OF TRANSFORMATIONAL LEADERSHIP ONFOLLOWER DEVELOPMENT AND PERFORMANCE:

A FIELD EXPERIMENT

TALY DVIRDOV EDEN

Tel Aviv University

BRUCE J. AVOLIOUniversity of Nebraska

BOAS SHAMIRHebrew University

In a longitudinal, randomized field experiment, we tested the impact of transforma-tional leadership, enhanced by training, on follower development and performance.Experimental group leaders received transformational leadership training, and con-trol group leaders, eclectic leadership training. The sample included 54 militaryleaders, their 90 direct followers, and 724 indirect followers. Results indicated theleaders in the experimental group had a more positive impact on direct followers'development and on indirect followers' performance than did the leaders in thecontrol group.

Transformational leadership theory is a promi-nent representative of the new theories that haveoccupied center stage in leadership research in thelast two decades. Follower development and fol-lower performance are the targeted outcomes ofsuch leadership (Bass & Avolio, 1990). However,there has been no conceptual framework, or sys-tematic research, for examining the impact of trans-formational leadership on follov r̂er development(House & Aditya, 1997). Transformational leader-ship has been shown to have a positive relationshipwith performance (Lowe, Kroeck, & Sivasubrama-niam, 1996). Yet a causal relationship betweentransformational leadership and follower perfor-mance has only rarely been demonstrated, becausemost prior studies have had static, correlational, ornonexperimental designs (Kirkpatrick & Locke,1996). The present experiment focused on explor-

The authors thank the Israel Defence Forces (IDF)School for Leadership Development and its consultants,and especially Yitzhak Gonen, Miki Rosenstein, EliavZakai, and Sigal Shavit-Traub, for their collaboration inconducting this experiment. Taly Dvir acknowledges thefinancial support of the Maya Fisher-David Fund, theJosef Buchmann Fund, the Israel Foundation Trustees,and the Faculty of Management at Tel Aviv University.The Lilly and Alejandro Saltiel Chair in Corporate Lead-ership and Social Responsibility supported Dov Eden'swork on this experiment.

ing the impact of transformational leadership onfollower development and on examining its lastingcausal impact on followers' performance.

THEORY AND HYPOTHESES

Transformational Leadership and FollowerDevelopment

The past two decades have heralded some con-vergence among organizational behavior scholarsconcerning a new genre of leadership theories, al-ternatively referred to as "transformational," "char-ismatic," and "visionary" leadership. Despite dif-ferent emphases in each theory. House and Shamirasserted that "it can be safely concluded that thereis a strong convergence of the findings from studieswith charismatic leadership and those concernedwith transformational and visionary leadership"(1993: 84).

The full-range leadership model (Bass & Avolio,1990) differentiates between transactional andtransformational leaders. Transactional leaders ex-ert influence by setting goals, clarifying desiredoutcomes, providing feedback, and exchangingrewards for accomplishments. Transformationalleaders exert additional influence by broadeningand elevating followers' goals and providing themwith confidence to perform beyond the expecta-tions specified in the implicit or explicit exchangeagreement. Transformational leaders exhibit char-

735

736 Academy of Management Journal August

ismatic behaviors, arouse inspirational motivation,provide intellectual stimulation, and treat follow-ers with individualized consideration. These be-haviors transform their followers helping them toreach their full potential and generate the highestlevels of performance.

A principal aspect of transformational leadershipis its emphasis on follower development (Avolio &Gibbons, 1988). Transformational leaders evaluatethe potential of all followers in terms of their abilityto fulfill current commitments, while also envi-sioning expansion of their future responsibilities.In contrast, transactional leaders expect followersto achieve agreed-upon objectives but do not en-courage them to assume greater responsibility fordeveloping and leading themselves and others(Bass, 1985; Burns, 1978). Although transforma-tional leaders' developing followers to what Bassand Avolio (1990) called their full potential is cen-tral to the theory, very little is known about howsuch leaders so develop followers. This lack ofknowledge led House and Aditya to conclude,"There is little evidence that charismatic, transfor-mational, or visionary leadership does indeedtransform individuals, groups, large divisions oforganizations, or total organizations, despite claimsthat they do so. . . . There is no evidence demon-strating stable and long-term effects of leaders onfollower self-esteem, motives, desires, preferences,or values" (1997: 443). In the absence of a theoryoutlining the developmental aspects of transforma-tional leadership, we have integrated differentsources to begin building a conceptual frameworkencompassing three main domains of follower de-velopment: motivation, morality, and empowerment.

Motivation. Burns (1978), the originator of trans-formational leadership theory, referred to two de-velopmental continua. The first concerns followermotivation. Burns proposed that transformational,as compared to transactional, leaders motivate fol-lowers in such a way that their primary motive is tosatisfy self-actualization needs rather than thelower needs in Maslow's (1954) need hierarchy.Drawing on Burns, Bass (1985,1998) suggested thattransformational leaders expand their followers'"need portfolios" by raising them or Maslow's hi-erarchy. Unlike transactional leaders, who concen-trate on fulfilling current follower needs, transfor-mational leaders arouse dormant needs. Bass(1985) also posited that followers' extra effortshows how much a leader motivates them to per-form beyond contractual expectations. Thus, em-phasis on satisfying self-actualization needs reflectsthe type of need underlying followers' motivation,and extra effort results from generating higher levelsof motivation.

Hypothesis la. Transformational leadershiphas a positive impact on the development offollowers' motivation in terms of their self-actualization needs and extra effort.

Morality. Burns's (1978) second developmentalcontinuum, follower moral development, is basedon Kohlberg's (1973) theory. Bass (1998) agreedwith Burns that to be transformational, a leadermust be "morally uplifting." One ofthe difficultiesin investigating moral development is that, accord-ing to Kohlberg, moving from one moral stage to thenext may take years, a time span rarely encom-passed in leadership studies. Shamir, House, andArthur (1993) provided an alternative for examin-ing moral development, specifying that the creationof value congruence between the leader/organiza-tion and followers is one of the processes under-taken by charismatic/transformational leaders. Wetherefore studied follower internalization of theirorganization's moral values as a manifestation ofmoral development. Drawing on Kohlberg, Bass(1985) emphasized the collectivistic aspect ofmoral development and suggested that transforma-tional leaders get their followers to transcend theirself-interest for the sake ofthe team or organization.This concept is similar to Wagner's (1995) defini-tion of collectivistic orientation. Shamir (1991) alsosuggested that follower collectivistic orientation isa transformational effect of charismatic leaders.

Hypothesis lb. Transformational leadershiphas a positive impact on the development offollowers' morality in terms of their internal-ization of their organization's moral valuesand a collectivistic orientation.

Empowerment. Transformational leadershiptheory, in contrast to early charismatic theories, hasconsistently emphasized followers' developmenttoward autonomy and empowerment over auto-matic followership (Graham, 1988). Still, researchhas not clarified whether or not charismatic ortransformational leaders are powerful because theirfollowers are weak (Klein & House, 1995). Scholarsconsider a critical-independent approach to be anessential empowerment-related process among fol-lowers of transformational leaders. For example,Bass and Avolio (1990) stated that transformationalleaders enhance followers' capacity to think ontheir own, develop new ideas, and question out-moded operating rules. Avolio and Gibbons (1988)posited that a major goal of transformational lead-ers is to develop follower self-management andself-development. Shamir (1991) similarly stressedthe transformational effects of charismatic leaderson follower independence. The view that a critical-

2002 Dvir, Eden, Avolio, and Shamir 737

independent approach is an outcome of transfor-mational leadership is also consistent with Kelley's(1992) conceptualization of styles of followership.Kelley's respondents described the best followersas those who "think for themselves," "give con-structive criticism," "are their own person," and"are innovative and creative." Kelley's (1992) re-view of the best, worst, and typical follower char-acteristics revealed a second dimension, namely,active engagement in the task. The best followers"take initiative," "participate actively," are "self-starters," and "go above and beyond the job." Wetherefore define active engagement as the energyinvested in the follower role as expressed by highlevels of activity, initiative, and responsibility. Ac-cording to Gonger and Kanungo (1988), charismaticleadership is tied to empowerment also throughself-efficacy. Shamir et al. (1993) and Avolio andGibbons (1988) specified increased follower self-efficacy as a developmental effect of transforma-tional leadership. We posited specific self-efficacyas a malleable developmental outcome (Eden,1990) enhanced among followers of transforma-tional leaders.

Hypothesis lc. Transformational leadershiphas a positive impact on the development offollowers' empowerment in terms of their crit-ical-independent approach, active engagementin the task, and specific self-efficacy.

Transformational Leadership and FollowerPerformance

Three types of studies have examined the rela-tionships between transformational and transac-tional leadership and performance. Many haveused ratings of leadership and outcomes collectedfrom a single source, leaving their results open tocommon-sourced-common-method bias (e.g., By-cio, Hackett, & Allen, 1995). Fewer studies haverelied on survey data on both leadership and out-comes collected from multiple sources (e.g., Keller,1992). The smallest number of studies have usedmultiple sources and multiple methods. Thesehave typically involved questionnaire ratings ofleadership and objective performance measures(e.g., Avolio, Waldman, & Einstein, 1988), or ma-nipulating leadership and measuring outcomes(e.g.. Barling, Weber, & Kelloway, 1996). Overall,there is evidence showing positive relationshipsbetween transformational leadership and perfor-mance; these relationships are stronger than therelationships between transactional leadership andperformance (Lowe et al., 1996). Yet there remainsa need for more rigorous field tests of the impact of

transformational leadership on objective perfor-mance to establish causality (Kirkpatrick & Locke,1996).

Hypothesis 2. Transformational leadership hasa positive impact on followers' performance.

Direct and Indirect Leadership

Direct leadership, or the relationships betweenfocal leaders and their immediate followers, hasbeen studied extensively. In contrast, knowledge ofindirect leadership, or the influence of focal leaderson individuals not reporting directly to them, ismuch more limited. The few attempts to under-stand indirect leadership have been limited toworld-class leaders or highly visible GEOs (Wald-man & Yammarino, 1999). It is assumed that trans-formational leadership at any level can impact bothdirect and indirect followers (Yammarino, 1994).There are, however, likely to be differences be-tween the processes that influence close and dis-tant followers (Shamir, 1995). The present experi-ment included direct and indirect followers. Giventhe dearth of theory and research, we formulated nohypothesis but took advantage ofthe opportunity toreveal whatever differences there may be betweendirect and indirect leadership.

METHODS

Design and Sample

This was a field experiment with random assign-ment of squads of leaders to conditions. The exper-imental leaders received transformational leader-ship training, and the control leaders went tliroughroutine eclectic leadership training. We predictedthat leaders assigned to the experimental trainingwould "enact" significantly more transformationalleadership than the control group leaders and thencorroborated this difference using a manipulationcheck. Having generated a higher level of transfor-mational leadership in the experimental conditionin phase 1, we then examined its causal impact onfollower development and performance in phase 2.

Phase 1. In phase 1, in the course of their officertraining, infantry cadets in the Israel DefenseForces (IDF) went through experimental and con-trol leadership workshops designed to enhancetheir leadership before they became platoon lead-ers. The phase 1 sample included 160 cadets in 12training squads.

Phase 2. After cadets have taken the officercourse, the IDF places the new officers in variousroles, mostly involving noncomparable perfor-

738 Academy of Management Journal August

mance contexts. On the basis of prior practice, weexpected 30 percent of the phase 1 cadets to heassigned as platoon leaders in hasic training, whereall followers are evaluated using the same mea-sures. To have comparable performance measuresfor all participating leaders, we tracked this groupin phase 2. Of the 160 phase 1 cadets, 54 (34%)were assigned to lead basic training platoons. Phase2 was conducted during a four-month infantry ba-sic training course, which began a month after theofficer course had ended and two months after theleadership workshops. The posttest measurementoccurred six months after the leadership work-shops. We assessed the impact ofthe platoon lead-ers' leadership, previously enhanced by the phase 1workshops, on their direct followers' (noncommis-sioned officers', or NCOs') development and theirindirect followers' (recruits') development and per-formance in phase 2. We collected leadership rat-ings and developmental data from NCOs and re-cruits at the beginning and at the end of basictraining. In the remainder of this report, these datacollections are designated "occasion 1" and "occa-sion 2," respectively. Performance grades were ren-dered at the end of basic training. The phase 2sample included 54 platoon leaders (32 who hadgone through the experimental workshops and 22who had been in the control workshops), 90 NCOs,and 724 recruits. All were men aged 18-22.

Measures

We pretested measures in a pilot sample of 320infantry commanders and followers and deletedand revised items on the basis of the pretest. Wecalculated alpha coefficients for each measure atthe group level in line with the unit of randomiza-tion, treatment, and analysis.

Manipulation checks. Cadets' initial reactions tothe workshop were assessed by 14 items (a = .95)developed for this study.^ Cadet knowledge acqui-sition regarding transformational leadership wasevaluated by 11 items developed for this study.Transformational leadership was measured withthe 20 transformational leadership items in theshort version of Bass and Avolio's MultifactorLeadership Questionnaire 5X (see Avolio, Bass,and Jung [1999] for a discussion of this revisedsurvey). We compared the experimental and con-trol leaders on a global transformational factor, themean of all transformational items, because our

^ All measures are available from the first author, ex-cept the MLQ, which can be obtained from its publisher.

goal was to create a higher level of overall transfor-mational leadership among the participants in theexperimental condition. The alphas for the globaltransformational scale ranged between .87 and .92over two occasions and subsamples.

Development. Except where noted, these mea-sures were the same for recruits and NCOs. Self-actualization needs were evaluated by 10 itemsbased mainly on Hackman and Oldham's (1980)growth needs index. Bass's (1985) three self-reportitems gauged followers' extra effort. Intemalizationof organizational moral values was assessed using17 items completed by the NCOs, with 3 itemsadded for recruits. We based this measure on the11 values included in the "IDF Code of EthicalConduct" and on in-depth interviews with IDFpersonnel regarding incidents that reflect moral di-lemmas. The 11 values include perseverance, com-radeship, discipline, sanctity of human life, loy-alty, personal example, professionalism, purity ofarms (which concerns ethical behavior during com-bat), representation, responsibility, and trustwor-thiness. Collectivistic orientation was measuredusing a 7-item scale based on Wagner's (1995)individualism/collectivism questionnaire. Critical-independent approach was gauged with 16 itemsdeveloped for this study on the basis of Kelley's(1992) concept of critical-independent thinking.Followers were asked about their thinking and ac-tions regarding themselves, their peers, theirleader, and the organization. Active engagementwas measured with 12 items developed for thisstudy on the basis of Kelley's (1992) construct. Toevaluate self-efficacy, we asked the NCOs to assesstheir ability to instruct recruits on each of the fivesubjects taught in basic training, and the recruitsestimated their own ability to master each subject.The measure was developed for this experiment onthe basis of Dvir, Eden, and Banjo's (1995) index.All but two of the developmental coefficient alphaswere above .70, and most exceeded .80. Alphaswere .60 and .69 for self-efficacy and collectivisticorientation, respectively, on one occasion in onesubsample. However, because the other three coef-ficients for each of these measures were above .70,we viewed them as reliable.

Performance. Recruits' performance was as-sessed by five routine IDF objective tests: lightweapons (written test), light weapons (practical test),physical fitness, obstacle course, and marksmanship.Leadership in this setting extends beyond conveyingtechnical or physical skills; it also involves develop-ing the recruits' understanding of the meaning ofthese tasks for overall vmit performance.

2002 Dvir, Eden, Avolio, and Shamir 739

Procedures

Twelve trainers at the IDF School for LeadershipDevelopment participated. All were experienced indelivering the eclectic leadership workshop. Seventrainers were randomly assigned to a five-day pre-liminary workshop to train them in delivering thenew transformational leadership model to the ex-perimental cadets in phase 1. The five control train-ers received no preliminary workshop. To precludecompensatory rivalry, we promised the controltrainers that they would receive the new trainingafter the experiment.

We assigned squads of cadets, rather than indi-viduals, to conditions hecause most trainingactivities in officer courses are done in squads.We randomly assigned seven squads to the trans-formational leadership condition and five to thecontrol condition. Three-day leadership workshopswere given to the cadets in hoth conditions. Theexperimental workshop^ emhodied the major prop-ositions of transformational leadership theory. Itwas huilt around four core themes that were con-veyed to the cadets in every workshop activity: (1)Transformational and transactional leadership aredifferent lenses through which a leader can viewrelationships with followers. (2) Transformationalleadership is enacted through a set of hehaviors. (3)Transformational leadership can create higher lev-els of development and performance among follow-ers than can transactional leadership. (4) Followersof transformational leaders should be continuouslydeveloped to higher levels of motivation, morality,and empowerment. The eclectic leadership work-shop delivered to control leaders was hased ondiscussing "here and now" individual and groupprocesses with a psychodynamic focus. The trainerrelated processes that occurred in the workshop tovarious concepts, such as goal setting, self-fulfillingprophecy, crisis intervention, contingency theory,trust building, personal example, and group cohe-sion. Both the experimental and the control work-shops employed role playing, group discussions,simulations, presentations and examples, videocases, and peer and trainer feedback. A month anda half after the workshops, before the leaders begantheir first leadership role, the trainers worked withthe experimental leaders for three hours to rein-force the treatment. Because of budgetary con-straints, we could not conduct booster sessionswith the control leaders.

After the experiment, when it could no longer

^ A full description of the experimental and controlworkshops is available upon request from the firstauthor.

affect the results, we conducted a five-day work-shop for the control trainers. The purpose was toenrich them with the knowledge and skills pro-vided earlier to the experimental trainers and sat-isfy the need for equitable treatment.

Analysis

Multivariate analyses of variance (MANOVA)and covariance (MANCOVA) were used to testwhether the treatment affected development andperformance. To estimate the differential effects oneach developmental and performance variable, weused one-way analysis of variance (ANOVA) forvariables measured once after the treatment andrepeated-measures ANOVA for variables measuredtwice. We computed correlations to estimate effectsizes and used the binomial effect size display(BESD) to express the practical impact of the treat-ment (Rosenthal & Rubin, 1982). When the treat-ment-by-occasion interaction was significant, weconducted simple-effects tests of the changes overtime in each condition. To reduce the likelihood oftype 2 error, we followed Sauley and Bedeian's(1989) recommendation and interpreted results sig-nificant beyond the .10 level as trends in the data.In phase 1, squads were assigned to conditions, andthus the analyses were done at the squad level. Inphase 2, 54 platoon leaders from the 12 experimen-tal and control squads participated. Therefore, datacollected in phase 2 were aggregated to, and ana-lyzed at, the platoon level.

Pretest data are typically collected prior to atreatment. In the present experiment, the firstround of phase 2 data collection took place onlytwo weeks into basic training. We could not expectenhanced transformational leadership among ex-perimental leaders to be evident so early into basictraining, when followers had so little exposure totheir leaders. We expected the experimental lead-ers to become more transformational as they hadmore interaction with their followers over time.Therefore, we regarded the first round of phase 2leadership and development data as a pretest andused the occasion-treatment interactions to test dif-ferences between experimental and control condi-tions in the amount of before/after change for vari-ables measured twice.

RESULTS

Manipulation Checks

The first manipulation check examined how theleaders perceived the workshops. We did not ex-pect a difference in how favorably they regarded

740 Academy of Management Journal August

the workshops, nor did we find one (x = 5.27, s.d.= 0.46, and x = 5.39, s.d. = 0.42 for the experi-mental and control conditions, respectively;Fl 11 = 0.22, n.s.). Thus, participants in both con-ditions got equally positively regarded leadershiptraining. The second manipulation check revealedthat the experimental leaders acquired moreknowledge of transformational leadership theorythan the control leaders {x = 79.0, s.d. = 7.0, and x= 56.0, s.d. = 3.0, respectively; F^ „ = 43.49, p <.001). The third manipulation check tested whetherthe training produced more transformational lead-ership behavior among the experimental leaders.The interaction of treatment and occasion was sig-nificant for NCO ratings of the platoon leaders'transformational leadership (F^ 43 = 4.43, p < .05).This interaction was not significant among the re-cruits (Fl 52 = 0.07, n.s.).

Direct and Indirect Followers' Development

MANCOVA revealed a significant treatment ef-fect [Fy 30 = 2.44, p < .05) for the combination ofthe seven developmental variables among NCOsafter adjustment for pretest differences. Table 1 pre-sents the means and the summary of the ANOVA ofeach developmental variable among the NCOs. Thetreatment-occasion interaction was significant forself-efficacy, critical-independent approach, andextra effort, and nearly significant [p = .06) forcollectivistic orientation.

The self-efficacy means showed the interactionstemmed from an increase in ratings in the experi-mental group and a decrease in the control group.Simple-effects tests showed the experimental in-crease approached significance (Fi 43 = 3.42, p <.10), whereas the decline in the control conditionwas not significant. The BESD equivalent of theinteraction effect size (r = .31) is a success rate of65 percent for the experimental platoons versus 35percent for the control platoons. Mean critical-in-dependent approach declined in the control group(Fl 43 = 11.12, p < .01) and remained unchanged inthe experimental group. The BESD equivalent ofthe interaction effect size (r = .38) is a success rateof 69 percent versus 31 percent. Mean extra effortdecreased sharply over time in the control condi-tion (Fl 43 = 7.49, p < .05), whereas the change inthe experimental group was not significant. TheBESD equivalent of the interaction effect size (r =.30) is a success rate of 65 versus 35 percent. Meancollectivistic orientation increased in the experi-mental condition and decreased in the control con-dition. Simple-effects tests revealed that neitherchange was significant. The BESD equivalent of theinteraction effect size (r = .28, p = .06) is a successrate of 64 versus 36 percent. Because there were sofew degrees of freedom in these group-level analy-ses, we interpreted a significance level of .06 aspreliminary confirmation of the positive impact oftransformational leadership on direct followers'collectivistic orientation. Table 1 shows no signif-

TABLE 1Means, Standard Deviations, and Summary of Repeated-Measures Analyses of Variance of Development

among Direct Followers*

Variable

Self-efficacy

Collectivistic orientation

Critical-independent approach

Extra effort

Active engagement

Internalization of moral values

Self-actualization needs

Occasion

12121212121212

Experimental

Mean

4.544.706.016.173.963.963.743.714.103.923.973.904.444.36

s.d.

0.320.260.600.500.310.400.560.750.340.390.350.400.450.41

Control

Mean

4.644.515.825.524.023.773.733.254.023.673.973.744.514.26

s.d.

0.310.570.681.030.340.470.490.910.390.750.460.510.320.74

F for theTreatment-by-

OccasionInteraction

4.55*

3.59''

7.07**

4.11*

1.46

1.94

0.81

n = 27 experimental platoons and 18 control platoons.p = .06.* p < .05

** p < .01

2002 Dvir, Eden, Avolio, and Shamir 741

icant treatment-by-occasion effects for the NCOs'active engagement, internalization of moral values,and self-actualization needs; nevertheless, themeans changed in the hypothesized direction. Forthese variables, the BESD equivalent of the meaninteraction effect size (r = .17) is a success rate of59 percent for the experimental platoons versus 41percent for the control platoons.

The MANCOVA of the developmental variablesamong the recruits showed no significant treatmenteffect, and repeated-measures ANOVA showed nosignificant treatment-by-occasion interactions.

Indirect followers' performance. MANOVA de-tected a significant treatment effect (Fg 26 = 3.45,p < .02) for the combination of the f'ive perfor-mance measures. The means in Table 2 show thatthe experimental platoons outperformed the con-trol platoons in every performance area. One-wayANOVA detected significant treatment effects onthe written light weapons test and on the obstaclecourse. The treatment effect approached signifi-cance (p < .08) for the practical light weapons testeven though the degrees of freedom in this analysiswere reduced from 41 to 32 owing to incompletedata for nine platoons. The BESD equivalent of theeffect size (r = .32) for written light weapons is asuccess rate of 66 percent for the experimental pla-toons versus 34 percent for the control platoons; forpractical light weapons (r = .31), 65 versus 35percent; and for the obstacle course (r = .52), 76versus 24 percent. The treatment effects for physi-cal fitness and marksmanship were not significant.

Post hoc correlational analyses. We computedcorrelations between the developmental variahlesand performance at the platoon level. Very fewsignificant relationships were found between the

recruits' or the NCOs' developmental variables andthe recruits' performance grades in both condi-tions. However, most of the correlations betweenthe NCO developmental variables and recruit de-velopmental variables were higher for the partici-pants who were in the experimental condition thanfor those who were in the control condition. In 36of 49 pairs of correlations, the experimental corre-lation was higher than the control correlation. Fish-er's r-to-Z transformation revealed that the differ-ence was marginally significant (p < .10) for 11pairs of correlations. For 9 pairs, the correlation inthe experimental platoons was significantly higherthan the correlation in the control platoons.

DISCUSSION

Theoretical and Practical Implications

The more positive impact of the transformationalleaders on direct follower development and on in-direct follower performance confirms core causalpropositions of transformational leadership theory.Moreover, the positive impact of the transforma-tional leaders on their indirect followers' perfor-mance experimentally strengthens conclusionsdrawn from previous studies, mostly conductedwith causally ambiguous designs.

The newly proposed conceptual framework fordevelopmental aspects of transformational leader-ship theory was partially confirmed. Transforma-tional leadership enhanced at least one measureeach of motivation, morality, and empowermentamong the direct followers. However, the impact oftransformational leadership was not confirmed fordirect followers' active engagement, internalization

TABLE 2Means, Standard Deviations, and Summary of Analyses of Variance of Performance among

Indirect Followers"

Test

Written light weapons''Practical light weapons'^Physical fitness''Obstacle course''Marksmanship''

Mean

84.6586.6478.64

476.6372.18

Experimental

s.d.

5.804.987.11

123.396.06

Mean

79.9883.5576.45

600.5370.29

Control

s.d.

8.414.715.17

59.976.91

F for theTreatment Effect

4.41*3.22''1.21

15.26**0.87

° The grades for all tests, except for the obstacle course, ranged from 0 to 100. Obstacle course results are presented in seconds in sucha way that a lower score represents better perfonnance.

'' n = 23 experimental platoons and 18 control platoons.•̂ 27 = 17 experimental platoons and 15 control platoons.' ' p < . 0 8 .

* p < .05**p < .01

742 Academy of Management Journal August

of moral values, and self-actualization needs. Be-cause the means of all of these variables were in thehypothesized direction, we conclude that moretesting is needed before revising the proposed de-velopmental framework. Transformational andcharismatic leadership theories are still at earlystages of specifying the developmental mediatingprocesses between leader behavior and perfor-mance (Shamir, 1991). The first step toward con-firming the hypothesis that follower developmentcan mediate effects on performance is to show thattransformational leadership affects development aswell as performance. The present results supportthese links among direct followers. The next step isto develop specific hypotheses linking specificleadership styles, developmental variables, andperformance measures in a broader range ofcontexts.

The more positive impact of the experimentalleaders on their direct followers' development ap-peared to prevent decline in some developmentalvariables. In the arduous and stressful context ofthis experiment, the positive impact of a leader onfollower development may he evidenced by avert-ing demoralization or regression in feelings aboutoneself. Given the dynamic nature of personnelassignments and work fiow, organizational restruc-turing, downsizing, and takeovers and mergers, apositive transformational leadership effect may beevidenced by halting motivational, moral, or em-powerment decline among followers. We suggestthat theoretical formulations should incorporatethe prevention of developmental regression as apositive outcome of transformational leadership.

Our experimental leaders affected their indirectfollowers' performance without seemingly affect-ing their self-perceived development. The presentfindings suggest that complex dynamics among thedirect and indirect followers may in turn affectperformance. The stronger relationships hetweenthe direct and indirect followers' ratings on themeasures of developmental variables in the exper-imental group may indicate that transformationalleaders create a stronger social bond among theirdirect and indirect followers, thus improving theindirect followers' performance. Other explana-tions are also possible. First, transformational lead-ership theory explains the effects of leadership onboth immediate and long-term processes and out-comes. Effects on indirect follower developmentmay take longer (Avolio & Bass, 1988). Perhaps thepresent experimental leaders planted developmen-tal seeds among their indirect followers, and moretime was needed for these seeds to germinate. Thisidea is especially relevant in the present context,because recruits (as opposed to NCOs) are typicallymore performance- than development-oriented.

Perhaps the stressful context suppressed the pre-dicted developmental effects among the recruits,who hore the brunt of this stress. Second, platoonleaders spend much less time with their indirectfollowers than with their direct followers, therebyrestricting the leaders' impact on the former's de-velopment. Perhaps a critical level of interactionwith a transformational leader is indispensable forthe impact on follower development to emerge; thedirect followers' interaction may have been abovethis threshold level, and the indirect followers',below it. Finally, it is possible that other, unmea-sured, variables played a crucial role in enhancingthe indirect followers' performance. For example,group developmental processes (for instance, pla-toon cohesiveness, potency, or culture) may havemediated the impact of leadership on recruitperformance.

Limitations and Future Research

Examining the effects of glohal transformationalleadership renders it impossihle to pinpoint thespecific components of transformational leadershipthat contributed to the effects produced. Futureresearch should add treatment conditions and fo-cus on specific aspects of transformational leader-ship, as Kirkpatrick and Locke (1996) did in theirlaboratory experiment on visionary leadership.

The fact that a booster session was offered only tothe experimental group raises the possibility of aHawthorne effect (whereby the attention given toparticipants in an experiment may improve theirattitudes and performance). We could not ohtainpermission for a control booster. Faced with eitherforegoing the booster altogether and risking failureat enhancing transformational leadership, or com-promising internal validity hy giving the boosteronly to the experimental leaders, we chose the lat-ter. However, in a military context, where person-nel are accustomed to participating in various cur-ricula, going through different programs did notappear to be an issue for participants. We are un-able to rule out the possibility that the boostersession given only to the experimental leaders ac-counted for some of the effects, hut we helieve thatthis threat to internal validity in the present exper-iment was relatively low.

According to Klein and House (1995), in homo-geneous charismatic relationships the leader sharescharismatic relationships with all followers or withnone. In nonhomogeneous cases, the leader sharescharismatic relationships with a select few follow-ers. Thus, the theory allows for either homoge-neous or variahle charismatic effects on followers.

2002 Dvir, Eden, Avolio, and Shamir 743

Our analyses used the platoon as the unit of anal-ysis in line vvfith the treatment level. We do notclaim that the platoon means in the present exper-iment represent homogeneous group effects, andaggregating to the platoon level may have maskedsome of the variance within platoons. However, wecould not address this issue better because thenumber of NCOs in each platoon was both smalland variable.

Finally, the idiosyncrasy of a military organiza-tion limits the external validity of our experiment.Replication in civilian organizations with mixed-gender and older participants is needed. Yet manyorganizational features in our sample are notunique to the military. Adherence to hierarchy andprofessionalism, a salient organizational missionthat depends on strong individual commitment,demanding and stressful jobs in which leaders andfollowers spend most of their time, and the need towork with direct and indirect followers character-ize many organizations. Indeed, the positive effectsof transformational leadership have been con-firmed in civilian as well as in military samples(Bass, 1998]. We conclude that transformationalleadership, enhanced by training, can augment thedevelopment of human resources and their perfor-mance in a variety of organizational contexts.

REFERENCES

Avolio, B. J., & Bass, B. M. 1988. Transformational lead-ership, charisma and beyond. In J. G. Hunt, H. R.Baliga, H. P. Dachler, & C. A. Schriesheim (Eds.),Emerging leadership vistas: 29-49. Lexington,MA: Heath.

Avolio, B. J., Bass, B. M., & Jung, D. I. 1999. Reexaminingthe components of transformational and transac-tional leadership using the Multifactor LeadershipQuestionnaire. Joumal of Occupational and Organ-izational Psychology, 72: 441-462.

Avolio, B. J., & Gibbons, T. C. 1988. Developing transfor-mational leaders: A life span approach. In J. A. Gon-ger & R. N. Kanungo (Eds.), Charismatic leadership:The elusive factor in organizational effectiveness:276-308. San Francisco: Jossey-Bass.

Avolio, B. J., Waldman, D. A., & Einstein, W. O. 1988.Transformational leadership in a management sim-ulation: Impacting the bottom line. Group and Or-ganization Studies, 13: 59-80.

Barling, J., Weber, T., & Kelloway, K. E. 1996. Effects oftransformational leadership training on attitudinaland financial outcomes: A field experiment. Joumalof Applied Psychology, 81: 827-832.

Bass, B. M. 1985. Leadership and performance beyondexpectations. New York: Free Press.

Bass, B. M. 1998. Transformational leadership: Indus-

try, military, and educational impact. Mahwah, NJ:Erlbaum.

Bass, B. M., & Avolio, B. J. 1990. The implications oftransactional and transformational leadership for in-dividual, team, and organizational development. InR. W. Woodman & W. A. Pasmore (Eds.), Researchin organizational change and development, vol. 4:231-272. Greenwich, GT: JAI Press.

Biu-ns, J. M. 1978. Leadership. New York: Harper & Row.

Bycio, P., Hackett, R. D., & Allen, J. S. 1995. Furtherassessments of Bass's (1985) conceptualization oftransactional and transformational leadership. Jour-nal of Applied Psychology, 80: 468-478.

Gonger, J. A., & Kanungo, R. N. 1988. Behavioral dimen-sions of charismatic leadership. In J. A. Gonger &R. N. Kanungo (Eds.), Charismatic leadership: Theelusive factor in organizational effectiveness: 78-97. San Francisco: Jossey-Bass.

Dvir, T., Eden, D., & Banjo, M. L. 1995. Self-fulfillingprophecy and gender: Gan women be Pygmalion andGalatea? Joumal of Applied Psychology, 80: 253-270.

Eden, D. 1990. Pygmalion in management: Productivityas a self-fulfilling prophecy. Lexington, MA: Lex-ington Books.

Graham, J. W. 1988. Transformational leadership: Foster-ing follower autonomy, not automatic leadership. InJ. G. Hunt, R. B. Baliga, P. H. Dachler, & G. A.Schriesheim (Eds.), Emerging leadership vistas:73-79. Lexington, MA: Lexington Books.

Hackman, J. R., & Oldham, G. R. 1980. Work redesign.Reading, MA: Addison-Wesley.

House, R. J., & Aditya, R. M. 1997. The social scientificstudy of leadership: Quo vadis? Journal of Manage-ment, 23: 409-473.

House, R. J., & Shamir, B. 1993. Toward the integration oftransformational, charismatic, and visionary theo-ries. In M. M. Ghemers & R. Ayman (Eds.), Leader-ship theory and research: Perspectives and direc-tions: 81-107. San Diego: Academic Press.

Kelley, R. E. 1992. The power of followership. NewYork: Gurrency and Doubleday.

Keller, R. T. 1992. Transformational leadership and theperformance of research and development projectgroups. Journal of Management, 18: 489-501.

Kirkpatrick, S. A., & Locke, E. A. 1996. Direct and indi-rect effects of three core charismatic leadership com-ponents on performance and attitudes. Joumal ofApplied Psychology, 81: 36-51.

Klein, K. J., & House, R. J. 1995. On Bre: Gharismaticleaders and levels of analysis. Leadership Quar-terly, 6: 183-198.

Kohlberg, L. 1973. Stages and aging in moral develop-ment: Some speculations. Gerontologist, 13: 497-502.

744 Academy of Management Journal August

Lowe, K. B., Kroeck, K. G., Sivasubramaniam, N. 1996.Effectiveness correlates of transformational andtransactional leadership: A meta-analytic review ofthe MLQ literature. Leadership Quarterly, 7: 385-425.

Maslow, A. H. 1954. Motivation and personality. NewYork: Harper.

Rosenthal, R., & Rubin, D. B. 1982. A simple generalpiupose display of magnitude of experimental ef-fects. Journal of Educational Psychology, 74: 166-169.

Sauley, K. S., & Bedeian, A. G. 1989. .05: A case of the tailwagging the distribution. Journal of Management,15: 335-344.

Shamir, B. 1991. The charismatic relationship: Alterna-tive explanations and predictions. Leadership Quar-terly, 2: 81-104.

Shamir, B. 1995. Social distance and charisma: Theoret-ical notes and exploratory study. Leadership Quar-terly, 6: 19-47.

Shamir, B., House, R. J., & Arthur, M. B. 1993. The mo-tivational effects of charismatic leadership: A self-concept based theory. Organization Science, 4(2):1-17.

Wagner, J. A. 1995. Studies of individualism-collectiv-ism: Effects on cooperation in groups. Academy ofManagement Journal, 38: 152-172.

Waidman, D. E., & Yammarino, F. J. 1999. GEO charis-matic leadership: Levels-of-management and levels-of-analysis effects. Academy of Management Re-view, 24: 266-285.

Yammarino, F. J. 1994. Indirect leadership: Transforma-tional leadership at a distance. In B. M. Bass & B. J.Avolio (Eds.), Improving organizational effective-ness through transformational leadership: 26-47.Thousand Oaks, GA: Sage.

AA

Taly Dvir ([email protected]) is a lecturer in manage-ment and organizational behavior at the Faculty of Man-agement, Tel Aviv University. She earned her Ph.D. inmanagement from Tel Aviv University. Her current re-search interests include transformational and charis-matic leadership, followership, leadership and emotions,and leadership differences in high- and low-tech firms.

Dov Eden is the Lilly and Alejandro Saltiel Professor ofGorporate Leadership and Social Responsihility at theFaculty of Management, Tel Aviv University. He earnedhis Ph.D. in organizational psychology at the Universityof Michigan. His current research interests include lead-ership, motivation, self-fulfilling prophecy at work, prac-tical application of organizational behavior theorythrough management training, and job stress and vaca-tion relief.

Bruce J. Avolio is the Don and Shirley Glifton Ghair inLeadership in the Department of Management and Gol-lege of Business Administration at the University of Ne-braska. His current research interests include how lead-ership is affected by the introduction of new informationtechnology in organizations and how e-leadership differsfrom leadership that is not mediated by technology. He isalso examining how information technology can be usedto augment and accelerate leadership development overtime.

Boas Shamir is a professor in the Department of Sociol-ogy and Social Anthropology, Hebrew University ofJerusalem. He holds a Ph.D. in social psychology fromthe London School of Economics and Political Science.His current research interests include leadership, follow-ership, and trust in organizations.