Transf SchoolLeader

School Effectiveness andSchool Improvement: AnInternational Journal ofResearch, Policy andPracticePublication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/nses20

The Relationship Between Transformational and Transactional Leadership and School Improvement OutcomesHalia C. Silins a

a School of Education, The Flinders University of SouthAustralia

Available online: 09 Jul 2006

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms- and-conditionsThis article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden.The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be

This article was downloaded by: [UNIVERSITY OF ADELAIDE LIBRARY]On: 26 March 2012, At: 19:19 Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH,UK

To cite this article: Halia C. Silins (1994): The Relationship Between Transformational and Transactional Leadership and School Improvement Outcomes, School Effectiveness and School Improvement: An International Journal of Research, Policy and Practice, 5:3, 272-298

To link to this article: http://dx.doi.org/10.1080/0924345940050305

independently verified with primary sources. The publisher shall not be liablefor any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material

LEADERSHIP AND SCHOOL OUTCOMES 3

.

School Effectiveness and SchoolImprovement 1994, Vol. 5, No. 3, pp. 272-298

The Relationship Between Transformational and Transactional Leadership and School Improvement Outcomes

Halia C. SilinsSchool of Education, The Flinders University of South Australia

ABSTRACT

This study examined the nature of the relationship between transformational andtransactional leadership and the nature of the relationships between specified schooloutcomes and the constructs of transformational and transactional leadership. Twoforms of analysis with different underlying structural assumptions, canonicalanalysis and partial least squares path analysis, were applied to an empiricallyderived data set to yield two path models. It was argued that the comparison ofthese two models provided evidence for the positive, relational nature of the linkbetween transformational and transactional leadership. The influence of leadershipon school improvement outcomes was discussed in the light of these findings.

Educational change has only recently begun to involve changing schools asorganizations (Barth, 1990; Joyce, 1990; Schlechty, 1990). The focus of attention hasshifted from individual to system change, from student achievement measures tobroader school outcomes, from teachers as agents of change to principals asorchestrators of change. Improvement of an organization involves restructuring, andrestructuring involves the acceptance of new ideas and new ways of acting (Heckman,1990). Greater demands are being placed on schools as organizations and units ofchange, than were placed on schools as a collection of classrooms. In Fullans words,more powerful strategies are needed for more powerful changes (1990, p. 21). Withthe shift from classroom to school as the focus of educational change, a correspondingshift in perception of the role of the principal from manager to leader has occurred.Some earlier studies (Blumberg & Greenfield, 1980; Wolcott, 1973) viewed principalsas systems maintenance managers rather than change agents or leaders. More recentnotions of the princi

Correspondence: Halia C. Silins, School of Education, The Flinders University of South

Downloaded by [UNIVERSITYOFADELAIDE LIBRARY] at 19:19 26 March 2012

0924-3453/94/0503-0272$6.00 Swets & Zeitlinger

4 HALIA C. SILINS

Australia, GPO Box 2100, Adelaide SA 5001, Australia.

Manuscript submitted: June 18, 1993 Accepted for publication: February 18, 1994palas an instructional leader have revived the perception of principals as key participantsin the leadership of the school (De Bevoise, 1984; Leithwood & Montgomery, 1982;Sweeney, 1982). Reports reveal, however, an oscillating focus from principals to staffas the crucial source of leadership for effective schools (Beare, Caldwell & Milliken,1989; Little & Bird, 1984; Stedman, 1987). Whenever this happens in any area ofinquiry, closer inspection usually reveals a false dichotomy. Principals do not have amonopoly on leadership, but they do have a position of privilege in terms of status,power and mechanisms readily available to them that facilitate the operationalizationof leadership into process strategies which can lead to school improvement. A principalcan demonstrate leadership by sharing leadership with others in the school. Moreover,it can be argued that a principal can achieve more status and more power byempowering others (Blase, 1987). The purpose of this study was to examine the impactof one leadership model (transformational leadership) on a broad array oforganizational outcomes in the context of major Canadian educational restructuringeffort. The model was selected because of its promise for the empowerment of organi-zational members.

CONCEPTUAL FRAMEWORK:TRANSFORMATIONAL AND TRANSACTIONAL LEADERSHIP

A number of recent studies by Leithwood and his associates (1990, 1991, 1992, 1993)have provided evidence that the form of leadership perceived as most helpful byteachers involved in educational change and restructuring is transformationalleadership. Transformational leadership has been identified by a number of writers asthe kind of educational leadership necessary to take schools into the 21st Century(Fullan, 1991, Leithwood, 1992; Sergiovanni, 1990; Schlechty, 1990). As a concept,transformational leadership attracted attention through the work of Burns (1978). Hisanalysis of the leadership manifested by world-renowned leaders provided foundationconcepts of ordinary and exemplary leadership identified by the terms transactionaland transformational.

Burns (1978) described the most common form of relationship found betweenleaders and followers as transactional, a term previously used by Downton (1973),when he contrasted transformational with transactional rebel leaders. Bass (1985),Burns (1978), and Hollander (1978) agreed that reinforcement theory formed the basisof transactional leadership which involved a social exchange between leader andfollower. Transactional leaders have been characterized as focusing on basic needs andextrinsic rewards as a source of motivation and basis for management. The leaderapproaches the followers with some transaction in mind and obtains compliance


(effort, productivity, loyalty) in exchange for expected rewards (economic, political orpsychological). Transactional leaders recognize what followers need and want andrecognize and clarify the roles and tasks required for followers to achieve desiredoutcomes. This form of leadership may produce an efficient and productive workplacebut is limited when compared with transformational leadership.

Transformational leaders not only recognize followers needs, but attempt to raisethose needs to higher levels of motivation and maturity while striving to fulfil humanpotential. Such total engagement (emotional, intellectual and moral) of both leadersand followers encourages followers to develop beyond expectations (Bass, 1985;Burns, 1978; Sergiovanni, 1991; Tichy & Devanna, 1986). Transformationalleadership bonds leader and followers within a collaborative change process thatimpacts on the performance of the whole organization resulting in a responsive andinnovative environment. In contrast, transactional leadership does not bind leaders andfollowers in any enduring way and promotes a routinized, non-creative but stableenvironment.

Alternative Conceptions of the Relationship between Transformational andTransactional Leadership

Burns (1978) envisaged transformational and transactional leadership on opposite endsof a continuum implying a polarized, unidimensional relationship between the two.This differentiation of transactional versus transformational leaders paralleled anearlier distinction made by Zaleznik (1977) between managers and leaders. Zaleznik(1977) observed, managers and leaders are very different kinds of people. They differin motivation, personal history, and how they think and act (p. 70). Such earlierconceptions of managers and leaders implied a dichotomous relationship between thetwo kinds of leadership. Bennis and Nanus (1985) believed the two concepts werequalitatively different and they captured the essence of the distinction by their phrasemanagers do things right, and leaders do the right thing (p. 21). The linking of theconcepts of managers and transactional leaders to bureaucratic behaviors and theiridentification with ordinary leadership has imputed a value judgement favouringtransformational over transactional behaviors. In recent years, there has been a shiftaway from considering these two concepts as at the opposite ends of a bipolarcontinuum. The alternative conceptions which have been advanced are considered inthe sections that follow.

The need to distinguish between managers and leaders has arisen out of a recentparadigm shift in the area of leadership (Louis & Miles, 1990). Louis and Miles (1990)have described this shift as the need to move from bureaucratic models of leadership tomore adaptive models. They characterized adaptive models as vision-driven

6 HALIA C. SILINS

emphasizing shared decision making and collective problem solving. Adaptive modelsrequire more skills attributed to leadership to provide for constant learning andevolution. Bureaucratic models were described as goal directed, emphasizing controland accountability through standardized operating procedures. They require moreskills attributed to management to provide for maintenance and continuity. Bydifferentiating between leadership and management, Louis and Miles (1990) did notintend to imply that one subsumes or assumes the other. Those in positions of authoritycan choose to employ strategies perceived as falling into a leader or managercategory enabling adaptive and bureaucratic models to be distinguished in terms ofleader behaviors. For educational restructuring, Louis and Miles (1990) advocatedadaptive models of organizational functioning where both more leadership and bettermanagement are required. This relationship between leadership and management canbe described as orthogonal: management and leadership are independent variables thathave no correlational relationship to each other.

Kotter (1988, 1990) subscribed to this view when he pointed out that managementand leadership are both characterized by important processes that sometimes can becarried out effectively by the same person. Some people can be both effective managersand effective leaders. For Kotter, the leader/manager difference and thetransformational/transactional leadership difference are very similar. However, herejects the current widely held belief that leadership is good and management isbad. The essential function of leadership for Kotter is to produce appropriatechange, whereas management is used to maintain the operations of the currentorganization.

There has been some controversy over the degree of the differences betweenmanagement and leadership (Yukl, 1989) which has implications for the concepts oftransformational and transactional leadership. Some have argued that theconstructs are qualitatively different (Bennis & Nanus, 1985), even mutuallyexclusive (Zaleznik, 1977), while others perceive them to be interdependent(Duignan, 1988; Hunt, 1991) stating that transformational leadership finds ex-pression through the management/transactional processes. There is an untestedbelief that management and leadership are separate but correlated concepts. Somewould contend that good managers are more likely to be better leaders and goodleaders are more likely to be good managers. This differs markedly from Basss(1985) view of transformational and transactional leadership. A transformationalleader may use selected transactional behaviors effectively, but it does not followthat a transactional leader will be able to draw on transformational skills.Although transformational leaders can be transactional when appropriate, trans-actional leadership is often a prescription for lower levels of performance ornonsignificant change (Bass & Avolio, 1990, p. 17). Transformational behavior is


conceived as higher-order behavior in a developmental sense and, presumably, itsemergence depends on experience and learning.Transformational/TransactionalLeadership Model

Bass (1985) developed a transformational/transactional leadership model for use byorganizational leadership researchers. Unlike Burns, Bass (1985) conceived oftransformational leadership and transactional leadership as separate and related.Management is not only leadership nor is leadership only management (Bass 1985,p. xiii). Bass (1985) contended that most leaders exhibit both transformational andtransactional leadership, in varying degrees. The effective leader integratestransactional and transformational leadership behaviors (Bass & Avolio, 1990).Transformational leadership augments transactional leadership by focusing on thedevelopment of followers as well as addressing the goals of the leader, follower, group,and organization (Bass & Avolio, 1990). For Bass, the success of a transformationalleader is demonstrated both by increased performance outcomes and the degree towhich followers have developed their own leadership potential and skills (Bass &Avolio, 1990). However, Basss reference to Zalezniks work on managers and leadersto support the structure of transformational and transactional leadership assumes adichotomous relationship between the two in Zalezniks conceptions (Bass, 1985, pp.229-230). Overall the nature of the relationship between transformational andtransactional leadership appears unclear. The problem lies in the fact that Basss modelof leadership overlays a developmental notion of leadership on a dichotomousdefinition of the two constructs.

Orthogonal Versus Correlational Structure

Basss (1985) conceptualization of transformational and transactional leadership findssupport in Zalezniks (1977) view of a dichotomous relationship between managersand leaders. The leader/manager concepts are paralleled by the trans-formational/transactional concepts. Managers and leaders are very different kinds ofpeople and they develop different kinds of cultures: managerial and entrepreneurial(Zaleznik, 1977, p. 70). Bass and Avolio (1990) have also suggested thattransformational and transactional leaders are different kinds of people and thattransformational leaders are more likely to emerge as leaders in times of growth,change, and crisis. Transactional leaders work to preserve the status quo and findsupport within a more mechanistic bureaucratic organization. Although this view oftransformational and transactional leadership appears to suggest an orthogonalrelationship between the two constructs of leadership, both kinds of leadership skillscan be found in one leader and a dependent relationship seems postulated between the

8 HALIA C. SILINS

different concepts when Bass and Avolio argued that the optimal leader is one whointegrates both transactional and transformational leadership approaches; effectivetransactional leadership forms a broad base upon which transformational leadershipcan build to achieve optimal performance (1990, p. 7).

On the other hand, Yukl (1989) seemed to assume an orthogonal relationship whenhe pointed out that it was stating the obvious to say that a person can be both amanager and a leader and also a leader without being a manager, and a managerwithout being a leader. Clearly, a transformational leader is not a transactional leader,since the terms are used to describe predominant behaviors. What is more,transformational leadership has been found to enhance significantly satisfaction withand perceived effectiveness of leadership beyond levels achieved with transactionalleadership when the two constructs are assumed to be independent (Bass & Avolio,1990; Hoover, 1988).

Silins (1992) tested Basss model of transformational and transaction leadership inschools undergoing reform using a broader set of outcomes such as teachersperceptions of school culture, teacher development, impact on program andinstruction, and impact on students. Application of canonical analysis to the surveydata found support for a two factor model in which transformational leadershipinfluenced teachers perceptions of all school outcomes except teacher effects andtransactional leadership influenced teacher effects favorably and school effectsunfavorably. Transformational leadership was found to be the dominant dynamic andaccounted for a significant incremental effect above that of transactional leadership inpredicting these outcomes (Silins, in press). These analyses assumed an orthogonalrelationship between the two concepts of transactional and transformational leadership.

Other studies of transformational and transactional leadership have employedanalyses which assume orthogonality of the two kinds of leadership (Bass, 1985;Hoover, 1988; King, 1989; Leithwood, Jantzi, Silins &Dart, 1993). Yet transfor-mational and transactional concepts are described as related, one building on the other,and an optimal leader is described as using both behaviors effectively (Bass & Avolio,1990). It would appear that there is some confusion over the nature of the relationshipbetween the two types of leadership, and whether transformational and transactionalleadership satisfy an orthogonal relationship or are positively or negatively correlatedor are interdependent. Empirical study of the nature of the relationship between thetwo types of leadership is needed to clarify the conceptualization of thetransformational and transactional constructs. Such a study must be carried out by anexamination of the relationships of these constructs with specified outcomes. Theauthentic nature of the relationships between the two constructs cannot be properlyinferred from evidence that does not involve outcome measures.

Specific Leadership Factors


Basss (1985) model of transformational and transactional leadership includes threefactors that identify the behavioral components of transformational leadership anddefine it (Bass, 1985; Waldman, Bass & Einstein, 1987).

Charisma/inspiration. This factor is the degree to which the leader creates enthusiasmin followers, sees what is really important, and transmits a sense of mission to theorganization. The leader inspires loyalty and devotion, instills pride and faith, andcommands respect. Followers place a great deal of trust and confidence in the leadersvision and values, develop intense feelings about the leader, perceive the leader as arole model, and want to identify with him or her.

Intellectual stimulation. This factor is the degree to which the leader providesintellectual and problem-oriented guidance. The leader arouses followers to think innew ways (Bass, Waldman, Avolio & Bebb, 1987). Followers are encouraged toquestion their own and others assumptions, beliefs and values, and developindependent problem-solving capabilities.

Individualized consideration. This factor is the degree to which the leader is concernedwith the individual needs of followers (mentoring). The leader responds to individualdifferences in followers needs for growth and development, elevating needs andabilities to higher levels when appropriate, and delegating projects to stimulateindividual learning experiences.

The transactional leader motivates followers to perform at their levels of expectationand to achieve satisfaction of basic needs. Transactions are at the heart of theinterchange between leader and followers. There are two factors that identify thebehavioral components of transactional leadership and define it (Bass, 1985; Waldman,Bass & Einstein, 1987).

Contingent reward. This factor is the degree to which the leader makes clear what thefollower must accomplish in order to be rewarded. The leader provides rewards iffollowers perform in accordance with contracts or expend the necessary effort to meetperformance standards. Clarification of goals, work standards, and assignments areemphasized. Leaders recognize what subordinates need and, through extrinsic rewards,energize followers to reach objectives.

Management-by-exception. In its active form, this factor involves the degree to whichthe leader provides negative feedback for failure to meet agreed-upon standards. Theleader avoids giving directions if the old ways are working and allows followers tocontinue doing their jobs as always if performance goals are met. In the less activeform of transactional leadership, the leader intervenes only when standards are notbeing met.

Avolio and Bass (1987) asserted that, unlike transactional leaders who work within theexisting organizational system, transformational leaders change the system to recreate

10 HALIA C. SILINS

their environment. Transformational leaders come into their own in times of growth,crisis, and reform. Burns (1978) described reform lead-ership as exacting and by itsvery nature, transformational. Transformational leaders tend to separate from the systemand create change (Zaleznik, 1977), whereas transactional leaders tend to work withinthe system to preserve it. Reform efforts that accept the political and social structureswithin which they act are inevitably compromised, and often inhibited, because of theinertia of existing institutions (Burns, 1978). If this is the case, then transactional leader-ship, as defined by Bass, cannot be as effective in bringing about change in times ofreform as transformational leadership.

This study set out to examine the degree of overlap between operational definitionsof transformational and transactional leadership and the nature of the relationshipsbetween specified outcomes and the constructs of transformational and transactionalleadership. Two forms of analysis with different underlying structural assumptions(canonical analysis and partial least squares path analysis) were applied to anempirically derived data set. It was hypothesized that the contrasting underlyingstructure of the relationships between transformational and transactional leadershipassociated with applying canonical analysis and partial least squares path analysiswould result in alternative path models, the relative fidelity of which could becompared. Cronbach (1984) referred to the fidelity of a model as an indicator of itscorrespondence to the real world.

METHOD

Source of Data

The British Columbia Ministry of Education introduced in 1989 a wide ranging schoolreform policy that focussed on the first three years of schooling. The policy entitledThe Primary Program, was developed in response to recommendations of the SullivanRoyal Commission report, 1989. This policy represented the first step of a sequence ofthree closely related policies planned for implementation by the Year 2000. Survey datameasuring perceptions of school personnel on a number of aspects of the PrimaryProject were collected in 1991 by the Center for Leadership Development, OntarioInstitute for Studies in Education, University of Toronto. A random sample, stratified bysize, was drawn from half the school districts in British Columbia. A total of 679individual teacher responses to the survey were averaged to provide 256 school scores.The school was the unit of analysis since this study employed teachers perceptions ofselected school outcomes as the criteria against which the predictors were tested. Accessto this data base enabled the author to carry out these analyses.


Variables in the StudyIn accordance with Basss model (1985), transformational leadership was represented by threevariables: charisma/inspiration, intellectual stimulation, and individual consideration. Transactionalleadership was represented by two variables: contingent reward and management-by-exception. Thevariables associated with transformational leadership and transactional leadership have been definedas theoretical factors above.

Four variables were chosen to represent the school improvement outcomes: schooleffects, teacher effects, program and instruction effects, and student effects.

School effects. This variable encompasses teacher-perceived changes resulting fromthe school improvement process impacting on the school as a whole and related to thefunctioning, climate and direction of the school.

Teacher effects. This variable encompasses teacher-perceived changes resultingfrom the school improvement process impacting on teachers.

Program and instruction effects. This variable encompasses teacher-perceivedchanges resulting from the school improvement process impacting on school programsand instruction.

Student effects. This variable encompasses teacher-perceived changes resultingfrom the school improvement process impacting on students.

The dependent and independent variables were operationally defined by a process ofclustering items from the survey to reflect the conceptual definitions of the variables.Items from the survey were categorized by Bass and the author into subscalesrepresenting the three hypothesized behavioral components of the transformationalscale and the two hypothesized behavioral components of the transactional scale.Ambiguous items were excluded. Decisions made on the basis of Basss modelresulted in the clustering of 13 items for charisma/inspiration, three items forintellectual stimulation, six items for individual consideration, three items forcontingent reward, and three items for management-by-exception, reported in Table 1.

The survey also provided items that operationally defined the four scales measuring outcomes of the improvement process (school effects, teacher effects, program and instruction effects, student effects). To verify the validity of the items forming these clusters, five experienced school personnel were given definitions of the four outcome variables and asked to categorize every item in this section into one of these four scales. Only thoseitems were used on which there was 80 percent agreement amonst the scales. Items which were difficult to cate-gorize were excluded. Five items were identified as representing school effects, eight items for teacher effects, five items for program and instruction effects, and ten items for student effects. The operational definitions of thevariables, represented by the items selected from the survey, and their conceptual definitions are listed in Table 1.Table 1. Operational and Conceptual Definitions of Variables.

Variables of Study and School Survey Items measuring variables

TRANSFORMATIONAL LEADERSHIPCharisma/inspiration- degree to which leader creates enthusiasm in followers, sees what is really important, andtransmits a sense of mission to the organization, inspires loyalty and devotion, instills pride andfaith, and commands respect. Followers place a great deal of trust and confidence in the leadersvision and values, develop intense feelings about the leader, perceive the leader as a role model,and want to identify with him or her.

Items4. Insists on only the best performance from us.

10. Has the capacity and judgement to overcome any obstacle.13. Leads by doing rather than simply by telling.18. Commands respect from everyone in the school.20. Symbolizes success and accomplishment within our profession.22. Excites us with visions of what we may be able to accomplish if we work together. 26. Shows us that there are high expectations for us as professionals.28. Will not settle for second best in performance of our work.29. Encourages us to be team players.32. Makes us less concerned about our own immediate needs and more concerned about our

school reaching its objectives.36. Makes us feel and act like leaders.37. Provides good models for us to follow.38. Gives us a sense of overall purpose.

Intellectual stimulation- degree to which the leader provides intellectual and problem-oriented guidance, arouses followersto think in new ways. Followers are encouraged to question their own and others assumptions,beliefs and values, and develop independent problem-solving capabilities.

Items7.Challenges us to reexamine some basic assumptions about our work. 17. Asks

questions that prompt us to think about what we are doing.33. Stimulates us to rethink the way we do things.

Individual consideration- degree to which the leader is concerned with the individual needs of followers (mentoring), responds to individual differences in followers needs for growth and development, elevating needs and abilities to higher levels when appropriate, and delegating projects to stimulate individual learning experiences. Followers are provided with coaching and teaching as needed

.Table 1. Continued.

Variables of Study and School Survey Items measuring variables TRANSFORMATIONAL

LEADERSHIP

Items

2. Facilitates assistance and support for us from external personnel, as required.9. Pays us personal compliments when we do outstanding work.15. Provides for extended training to develop knowledge and skills relevant to the Primary Program.16. Treats us as individuals with unique needs and expertise.25. Behaves in a manner thoughtful of our personal needs.40. Provides special recognition when our work is especially good.

Contingent reward- degree to which the leader makes clear what the follower must accomplish in order to berewarded, provides rewards if followers perform in accordance with contracts or expend thenecessary effort to meet performance standards. Clarification of goals, work standards, andassignments are emphasized. Leaders recognize what subordinates need and, through extrinsicrewards, energize followers to reach objectives.

Items

8. Assures us that we can get what we personally want in exchange for our efforts. 24. Provides information relevant to implementing the Primary Program.34. Helps us get what we decide we want.

Management-by-exception

- degree to which the leader provides negative feedback for failure to meet agreed-upon standards,avoids giving directions if the old ways are working and allows followers to continue doing theirjobs as always if performance goals are met.

Items

5. Asks no more of us than what is absolutely essential to get our work done.6. Is satisfied with our performance as long as the old ways work.35. Does not try to change anything as long as things are going all right.

SCHOOL IMPROVEMENT OUTCOMES

School effects

- perceived changes from the school improvement process impacting on the school as a whole andrelated to the functioning, climate and direction of the school.

Items

1. A clearer sense of purpose in the school about what we want our students to accomplish.7. Greater agreement about educational values and beliefs.11. Alteration of timetables/schedules to facilitate improvements.

Table 1. Continued.

Variables of Study and School Survey Items measuring variables SCHOOL

IMPROVEMENT OUTCOMES17. A more positive atmosphere with less emphasis on punishment or control and more emphasis

on praising and rewarding students.29. Increased confidence in the value of our work at school.

Teacher effects- perceived changes from the school improvement process impacting on teachers.

Items3. A higher degree of consensus among staff about our goals for students.6. More collaboration among teachers in making curriculum and instructional decisions. 13.

Increased skill in using my usual instructional strategies.19. A higher level of job satisfaction experienced by me and/or colleagues.21. Greater sensitivity by me and/or my colleagues to student needs.23. Larger repertoire of instructional strategies for my use.24. Increased expectations by me and/or my colleagues for most students.27. More opportunities for professional growth.

Program and Instruction effects- perceived changes impacting on school programs and instruction.

Items5. Assessment practices with greater sensitivity to student growth.9. Greater emphasis placed on academic than on social goals.

18. Alterations in how students are assigned to classes to reflect program implementation.22. Alterations in our program to better reflect Primary Program policy.30. Changes in school programs to better reflect student needs.

Student effects- perceived changes from the school improvement process impacting on students.

Items2. Greater ability of students to use mathematical concepts to understand real-world situations.4. Acquisition of more skill in use of oral, visual and written language.8. More positive attitudes by students toward school and learning.10. Acquisition of greater competency in individual artistic expression.

12. Greater ability to engage in mathematical thinking and reasoning.14. Growth in students positive self-concept.15. More cooperative attitude by students toward working with one another.16. Greater appreciation for reading and listening to a variety of books.20. Growth in students use of appropriate problem solving and decision making strategies in

everyday living.

Greater awareness and appreciation of the relationship of science to their surround-ings.Sample

The sample employed was a probability proportional size random sample of schooldistricts with a sampling fraction of approximately one third. In a second stage,schools were sampled from within the chosen districts. Schools to be surveyed werestratified by size and randomly selected. Because a cluster sample of schools withindistricts was employed, and not a simple random sample of schools, the number ofschools responding cannot appropriately be used in the significance tests applied to theanalyses carried out. As a consequence, it was necessary to examine the F valuesobtained and their levels of significance with caution. The data obtained from theindividual teachers within schools were aggregated to the school level in analysis.

Data Analysis

Two different methods of analysis involving different underlying assumptions anddifferent structural models were used to analyze the data set. First, the relationshipbetween the set of dependent variables and the independent variables was examinedusing canonical analysis which forces orthogonality on the relationships between thepairs of canonical variables. Canonical analysis allows the construction of latentvariables and the calculation of the effects of the predictor (leadership) latent variableson the criterion (school improvement outcomes) latent variables but commonly limitsthe procedure to the analysis of only two sets of variables at a time, one set ofpredictor and one set of criterion variables. Inference regarding the number ofsignificant canonical variate pairs proceeds on the assumptions of multivariatenormality and independence of variables.

Canonical analysis provided information about the number and nature of mutuallyindependent relations between the two sets of variables and about the degree ofredundancy between the two sets (Keeves, 1986). This analysis allows recombinationsof the observed variables of both sets to achieve a maximally correlated linearcombination of dependent variables with a linear combination of independentvariables. The first pair of linear combinations yields the highest canonical correlation(Rc) possible in a given set of data. The second pair of canonical variables are thenbased on linear combinations of dependent and independent variables that are notcorrelated with the first pair and that yield the second largest canonical correlationpossible in the data to a maximum of four in this analysis (number of variables in thesmaller set). Usually only the first two or three combinations are significant and need

interpretation. The analysis was performed using SAS CANCORR (SAS Institute Inc.,1986).

The second method of analysis, applied to the same data, was partial least squarespath analysis performed using LVPLS (Lohmoeller, 1987) because this procedureallows the latent constructs of transformational and transactional leadership to berelated. Partial least squares path analysis allows the construction of latent variables aswell as the estimation of effects between the constructs. LV- PLS, however, permits thecalculation of causal links not only between the latent variables and the outcomemeasures but also between the explanatory constructs. Although partial least squarespath analysis has been little used in the analysis of data, it is a promising multivariatetechnique which is capable of being used flexibly with regard to assumptions ofdistribution and independence, handling many variables and composite scalessimultaneously, providing an appropriate form of information reduction, providingestimates of direct and mediated effects, and is computationally and technicallymanageable (Keeves, 1986).

The LVPLS procedure is based on the use of ordinary least squares estimationprocedures. It calculates an estimate for each latent variable which is derived from thecorresponding observed variables thus partitioning the hypothesized inner model into itscomponent constructs. Once the program has assigned these values for each latentvariable, the paths between the latent variables are computed by calculating least squaresestimates for all variables (observed and latent) in the model. Although path analysisaims at the examination of causal relationship between certain variables, it does notestablish causality between the variables. Rather, the causal relationships in path modelshave to be based on theory and logic. Thus, path analysis has to be seen as a means oftesting the theoretical assumptions and of estimating the strength of the causal links andnot as a tool from which theory can be derived (Freedman, 1987).

RESULTS

The results of the canonical analysis and partial least squares path analysis are reportedand the relative fidelity of the resultant path models are examined to identify thesuperior causal model. Presentation of data and tables are limited to key relationshipsonly (more detailed exposition of these analyses and their underlying principles can befound in the second edition of International Encyclopedia of Education edited by Husenand Postlethwaite).

Canonical Analysis

CanonicalR

R2 Approx F df PCanonical analysis was used to provide information regarding the number and nature of mutually independent (orthogonal) relations between the two sets of observed variables:the first set included the five predictor variables, charisma/ inspiration, intellectual stimulation, individual consideration, contingent reward and management-by-exception,and the second set included the four criterion variables, school effects, teacher effects, program and instruction effects, and student effects. Canonical analysis also explores the relationships between the leadership variables and between the teacher perceived outcomes.Table 2. Canonical Correlations Between Predictors and Criteria.

0 0.33 0.11 3.07 20

Contingent reward 0.41 0.95 0.87 0,32Management-by-exception -0.01 0.53 -0.41 0,64

Variance extracted 0.62 0.17

VI V2 VI V2

Criterion measuresSchool effects 0.35 -1.32 0.83 -.049Teacher effects 0.04 0.23 0.04 0,46Program and instruction 0.74 0.65 0.97 0.21Student effects -0.01 0.61 0.76 0.17

Variance extracted 0.55 0.13

Canonical R 0.33 0.28Canonical R2 0.11 0.08Redundancy 0.06 0.01

by the four latent variables of the canonical analysis, but only the two significant latentvariables identified as transformational and transactional are shown in the diagram.

This path model illustrates that when an orthogonal relationship is forced betweentransformational and transactional leadership, transformational leadership is mainlydefined by intellectual stimulation, individual consideration and contingent reward.Transactional leadership is defined by contingent reward and management-by-exception with a negative contribution from charisma/inspiration and intellectualstimulation (indicating negative behaviors for these two variables contributing to thedefinition). When the nature of the relationship between transformational andtransactional leadership is assumed to be independent, the path model indicates that inthis data set, transformational leadership influences school effects, program andinstruction effects and student effects directly, but not teacher effects, whereas,transactional leadership influences teacher effects positively and school effectsnegatively

^'Transformation weights > 0.20 are underlined **Structural coefficients > 0.30 are underlined

Fig. 1. Path Model for canonical analysis of leadership influence on school outcomes.

.

Partial Least Squares Path Analysis

Partial least squares path analysis was used to examine the causal relationship betweenthe predictor variables, represented by the two constructs of leadership (as defined inBasss model), and the four criterion variables of school, teacher, program andinstruction, and student outcomes when transformational and transactional leadershipwere allowed to be related. It would seem inappropriate merely to assume that the twoleadership constructs were correlated since discussions of the nature of the two typesof leadership have extended beyond a simple association. Causal relations have alsobeen considered. Discussions of the relationship between the latent variables oftransactional and transformational leadership have argued that a transformationalleader may employ transactional behavior but that a transactional leader does notnecessarily employ transformational behavior. As a consequence, it is assumed thatthere is a causal path from transformational leadership to transactional leadership andnot vice-versa.

Table 4 presents the direct, indirect and total effects, latent variable correlationswith corresponding percentage of variance explained which were obtained throughLVPLS analysis. In the formation of a causal model of factors influencing the schoolimprovement outcomes of school effects, teacher effects, program and instructioneffects and student effects, outer and inner models were hypothesized. As a first step,the estimation mode (inward or outward) is defined indicating whether observed or

Transformational leadership abbreviated to TransF Transactional leadership abbreviated to TransA

manifest variables (MVs) are combined to form or to reflect a particular latentvariable. The inward mode increases the predictive power of the model. The outwardor reflective mode primarily aims at extracting the common characteristics of theobserved variables in a construct and increasing the internal consistency of the latentvariable. The outward mode, applied to transformational leadership, results in the MVsbeing assigned factor loadings as the reflected latent variable is estimated using aprinciple component or factor analytic approach. The inward or formative mode,applied to transactional leadership, results in what may be considered as standardizedregression weights (6) being assigned to the MVs since the latent variable formed isestimated using a least squares regression approach. Singleton variables are alwaysrepresented in the reflective mode. Thus school, teacher, program and student effectsare all in the reflective or outward mode.

The outer model relationships are first estimated by LVPLS and having estimatedthe outer model, the program goes on to estimate the inner model relationships, andcontinues to iterate until convergence is achieved. Once the outer

VariablesDirect Effects Total Effects Indirect Effects Correlations

Student OutcomesTransactional 0.22 0.22 * 0.26Transformational Variance explained 7%

(0.05) 0.22 0.17 0.22

Program and Instruction OutcomesTransactional (0.12) (0.12) * 0.28Transformational Variance explained 10%Teacher Outcomes

0.21 0.30 * 0.30

Transactional (0.07) (0.07) * 0.22Transformational 0.20 0.26 * 0.26Variance explained 7% School OutcomesTransactional (0.08) (0.08) * 0.26Transformational 0.25 0.30 * 0.30Variance explained 9% Transactional Transformational Variance explained 55%

0.74 0.74 * 0.74

model is stabilized, the path estimates for the inner model are computed. The innermodel comprises the relationships which are specified among the latent variables(LVs). Initially a fully recursive model is specified and the inner model is refinedsuccessively as paths which do not contribute significantly to explaining another LV

Table 4. Direct Indirect Total Effects and Latent Variable Correlations.

* Indirect path coefficient r

are deleted.It is widely recommended to exclude MVs from the outer model equation if their

weight falls below 0.10 (

communality coefficient (0.87) (Sellin & Keeves, in press). The variances of thecriterion variables explained by the model as indicated in Table 4 were school effects 9percent, teacher effects 7 percent, program and instruction effects 10 percent andstudent effects 7 percent. The average criterion variance explained by the predictorvariables using PLS is 8.2 percent.

This path model illustrates that when transformational and transactional leadershipare allowed to be related, transactional leadership is strongly dependent ontransformational leadership. In the outer model, transformational leadership ispositively reflected in the three defining variables of charisma/inspiration, intellectualstimulation and individual consideration. Transactional leadership is formed anddefined positively by contingent reward and negatively by management-by-exception.Focussing on the inner model, transformational leadership is shown directlyinfluencing teachers perceptions of school, teacher and program and instructionoutcomes, while indirectly (through transactional leadership) influencing studentoutcomes. Transactional leadership directly influences student outcomes.

DISCUSSION

Partial least squares analysis indicated a strong positive relationship betweentransformational and transactional leadership with transformational leadershipexplaining 55 percent of the variance in transactional leadership. This fits Basssdevelopmental notion of transformational and transactional leadership and the morerecent correlational conceptions of leadership. If optimal leaders are integratingtransformational and transactional behaviors then the negative relationship ofmanagement-by-exception to transactional leadership seen in the data also supportsBasss model and the emerging view that important leadership processes are identifiedby both concepts. Management-by-exception appears to be a negative or passivecharacteristic of leadership. Behaviors opposite to those encompassed by management-by-exception combine with contingent reward to form transactional leadership.

The path model based on the assumption of a correlated relationship betweenvariables (LVPLS) indicated clearly the integrated nature of the two concepts ofleadership in terms of their impact on school improvement outcomes. As noted earlier,transformational leadership has a significant direct influence on school, teacher andprogram outcomes, but the impact of transformational leadership on students isindirect and through transactional leadership, that is, the organizational (Kotter,1988) aspects of leadership that result in the establishment of clear roles andexpectations, planning and scheduling and managing the environment of the school.This has been expressed in the following way : Transactional leadership practiceshelp teachers recognize what needs to be done in order to reach a desired outcome . . . .increases teachers confidence and enhances motivation (Leithwood, Jantzi, Silins &Dart, 1992). However, the influence of such practices is on perceived student

outcomes and not on the perceived teacher outcomes.This PLS model supports recent notions of the way transformational leadership is

manifested in schools. Manasse (1986) noted that research has demonstrated thatleaders lead as they manage (p. 153), a view supported by others (Duignan, 1988;Kotter, 1988). Principals on-going daily tasks and interactions provide opportunitiesto keep a finger on the pulse of the school and the people in it, to plant their ideas,convey their vision and suggest interpretations of events. Dwyer et al. (1985), Manasse(1984, 1986), Martinko & Gardner (1983), Morris et al. (1981) have all demonstratedthat the actual observed activities of principals vary remarkably little between highlyeffective and relatively ineffective leaders. The difference in their impact on theirschool appears to come not from what the individuals do during the day, but fromhow they think about what they do, how they communicate how they think, and whatthey do while they are doing it (Manasse, 1986, pp. 153-154). Effective principalsuse activities of management to accomplish the goals of leadership.

By forcing an orthogonal relationship between transformational and transactionalleadership, canonical analysis gave rise to two latent leadership variables L, and L 2concerned with transformational and transactional leadership respectively.Charisma/inspiration and intellectual stimulation are negatively correlated to thesecond latent variable (transactional) together with a positively correlatedmanagement-by-exception and highly positively correlated contingent reward. At thesame time contingent reward also contributed positively to the formation of the firstlatent leadership variable (transformational). There exists a link between the twoconstructs of leadership through the contingent reward variable in this model. Bass in apersonal communication to the author (April 28, 1992) suggested that two independentforms of contingent reward operated in defining the transformational and transactionallatent variables. The psychic rewards exchanged concern development, praise, orrecognition ... fall into individualized consideration which in turn is highly correlatedwith other transformational factors. A purely transactional contingent reward is a payincrease or a bonus. In this same communication, Bass indicated that partial leastsquares path analysis rather than factor analysis resulted in the unambiguousseparation of the defining or manifest variables associated with the leadershipconstructs. The assumption of orthogonality does not appear to be theoreticallyspecified nor supported by these analyses. What is more, the identified orthogonaltransactional leadership explains only one percent of the variance in the criteriavariables (the transformational leadership variable explains 6 percent).

Yammarino and Bass (1990) have pointed out that a problem with studies ofleadership is the typically small amount of variance explained by any particularinvestigation (p. 993). This study indicates that such disappointing past results maybe due to the choice of analysis. Most empirical studies have employed factor analysisand regression analysis assuming independence of variables (Bass, 1985; Bass &Avolio, 1990; Hoover, 1988; King, 1989; Silins, 1992). Partial least squares pathanalysis explained on average, 8.2 percent of the criterion variance, marginally more

than the 7 percent explained by canonical analysis. If a test of Cronbachs fidelity isthe amount of variance explained by the two models generated by the data under thediffering assumptions, then LVPLS is a marginal improvement.

While the comparative fidelity of the two resultant models is difficult to resolvestatistically, evaluation of the two models is possible by weighing up the nature of therelationship between leadership constructs and school improvement outcomes offeredfor interpretation by each model. Figure 1 and Figure 2 help to illustrate theidiosyncrasies of the two perspectives with their different explanatory powers. Bothpath models indicate that transformational leadership has the more generalized andstronger influence over school improvement outcomes whether the two forms ofleadership are viewed as orthogonal or correlated. This helps justify the currentemphasis on transformational forms of leadership for improving school performanceand for school reform (Leithwood & Jantzi, 1990; Leithwood, Jantzi, Silins & Dart,1993; Silins, 1992). The influence of transformational and transactional leadership onteachers perceptions of school improvement outcomes, however, differs radically inthe two models.

In Figure 1 (canonical analysis), transformational leadership impacts on school,program and instruction, and student outcomes directly and strongly. Perplexingly,transformational leadership (when viewed as independent of transactional leadership)has no direct effect on teacher outcomes. Transactional leadership influences teacheroutcomes positively and directly while influencing school effects (whole schoolfunctioning, climate, culture, direction) negatively. The author has attempted to explainthis apparent contradictory impact of transactional leadership more fully elsewhere(Silins, 1992). Briefly, it was hypothesized that, to the extent that transactionalleadership could be seen as a more

passive form of leadership than transformational, its presence in a school mayempower and develop teachers by providing them with more decision makingopportunities through the leaders default of responsibilities. This was also seen asexplaining the negative influence on school effects of transactional leadership whenviewed as the abrogation of the recognized and necessarily active role of the leader forimproved school performance (Beare, Caldwell & Milliken, 1989; Fullan, 1991;Sashkin, 1988). This contradictory impact of transactional leadership can provide thebasis for a more sinister interpretation of these results; that the kind of leadership thatis good for teachers is not necessarily good for the whole school; that teachersinterests are not compatible with the schools interests. Controversial results of thiskind cast doubt on the fidelity of this model. Such an interpretation can only begenerated from the path model for canonical analysis of leadership influence on schooloutcomes which requires that the relationship between transformational andtransactional leadership be orthogonal.

In Figure 2 (partial least squares path analysis), transformational leadershipinfluences school, teacher, and program and instruction outcomes directly and strongly.However, since this form of analysis allows transformational and transactional

leadership to be related, the data in this study indicate that transformational leadershipstrongly influences transactional leadership, which in turn impacts on studentoutcomes. It has been noted that over half of the variation in transactional leadership inthese data is explained by transformational leadership. Transformational leadership,therefore, has a strong indirect influence on student outcomes. This model indicatesrelationships between leadership and school outcomes that generate interpretations thatfit better with our understanding of the real world, and thus have greater fidelity.

CONCLUSION

This study presents evidence for the positive, relational nature of the link betweentransformational and transactional leadership. It casts strong doubts on themeaningfulness of using analyses which assume an orthogonal relationship betweentransformational and transactional leadership as independent variables. Comparison ofthe relative fidelity of each of the two path models presented has resulted in greatersupport for the relational view of the nature of the link between transformational andtransactional leadership than for the orthogonal view. This has been achieved byexamining (a) the structure of the two leadership concepts, (b) the assumptions and theprocedures underlying the two modes of analysis, and (c) the relationships betweenleadership and school improvement outcomes as presented in the two path models.

In addition, this study has demonstrated that applying analyses with alternativeunderlying assumptions to the same data for comparative purposes can haveilluminating consequences. Where contradictions exist in the understanding of thenature of theoretical constructs being used, some clarity can be achieved by analyzingthe data in more than one way.

This paper makes explicit the ambiguity with which the leadership constructs oftransformation and transaction are dealt with in the current literature. The findingsreported here illustrate how the assumed nature of the relationship betweentransformational and transactional leadership influences the kind of relationships foundbetween leadership and valued school outcomes that emerge from the analysis ofempirical studies. The challenge is to clarify whether these leadership constructs arerelated or orthogonal, and if related, as presently implied, to what extent? Researchersneed to collect evidence that can be examined in such a way that determines the specificnature of this relationship.

If transformational and transactional leadership are related, then transactionalleadership behaviors cannot be ignored when studying the impact of leadership onschool improvement outcomes. Transactional leadership behaviors may provide theindispensable bureaucratic linkages between transformational leadership and improvedschool outcomes. Behaviors that mediate transformational leadership need closer study.Transactional behaviors appear to be effective mediators for the effects oftransformational leadership on student outcomes and cannot be ignored in our attempts

to understand effective leadership for school reform.

REFERENCES

Avolio, B., & Bass, B. (1987). Transformational leadership, charisma, and beyond. In J.Hunt, B. Baliga, H.P. Dachler & C. Schriesheim (Eds.) Emerging leadershipvistas. Lexington, MA: Lexington Books.

Barth, R. S. (1990). Improving schools from within: Teachers, parents, and principalscan make the difference. San Francisco, CA: Jossey-Bass.

Bass, B. M., & Avolio, B. J. (1988). Manual for the Multifactor Leadership Question-naire. Binghamton, NY: Center for Leadership Studies, State University ofNew York at Binghamton.

Bass, B. M., & Avolio, B. J. (1990). Transformational leadership development: Manualfor the Multifactor Leadership Questionnaire. Palo Alto, CA.: Consulting Psy-chologists Press.

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

Bass, B. M., Waldman, D. A., Avolio, B. J., & Bebb, M. (1987). Transformationalleadership and the falling dominoes effect. Group & Organization Studies, Y2(1 ), 7387.

Beare, H., Caldwell B. J., & Millikan, R. H. (1989). Creating an excellent school: Somenew management techniques. New York: Routledge.

Bennis, W. G., & Nanus, B. (1985). Leaders: Strategies for taking charge. New York:Harper & Row.

Blase, J. (1987). Dimensions of effective school leadership: The teachers perspective.American Educational Research Journal, 24, 589-610.

Blumberg, W., & Greenfield, W. (1980). The effective principal: Perspectives onschool leadership. Boston: Allyn and Bacon.

Burns, J. M. (1978). Leadership. New York: Harper and Row.Cronbach, L. J. (1984). Essentials of psychological testing. New York: Harper & Row.De Bevoise, W. (1984). Synthesis of research on the principal as instructional leader.

Educational Leadership, 41(5), 14-20.Downton, J. V. (1973). Rebel leadership: Commitment and charisma in a

revolutionary process. New York: Free Press.Duignan, P. A. (1988). Reflective management: The key to quality leadership. Interna-

tional Journal of Educational Management, 2(2), 3-12.Dwyer, D. C., Lee, G. V., Barnett, B. G., Filby, N. N., Rowan, B., Alpert, B., &

Kojimoto, C. (1985). Understanding the principals contribution toinstruction: Seven principals, seven stories (Vols. 1-8). San Francisco: FarWest Laboratory.

Freedman, D. A. (1987). As others see us: A case study in path analysis. Journal ofEducational Statistics, 12(2) 101-128.

Fullan, M. G. (1990). Staff development, innovation, and institutional development. InB. Joyce (Ed.). Changing school culture through staff development. The 1990Yearbook of the Association for Supervision and Curriculum Development.Alexandria: ASCD.

Fullan, M. G. (1991). The new meaning of educational change (2nd ed.). New York:Teachers College Press.

Hollander, E. (1978). Leadership dynamics. New York: The Free Press.Hoover, N. R. (1988). Transformational and transactional leadership: A test of the

model (Doctoral dissertation, University of Louisville, 1987). DissertationAbstracts International, 48, 3020A.

Hunt, J. G. (1991). Leadership. A new synthesis. Newbury Park, CA: Sage Publications.Husen, T., & Postlethwaite, T. N. (Eds.), (in press). International encyclopedia of

education (2nd ed.). Oxford: Pergamon Press.Joyce, B. (Ed.). (1990). Changing school culture through staff development. The 1990

Yearbook of the Association for Supervision and Curriculum Development.Alexandria: ASCD.

Keeves, J. P. (1992). The IEA study of science III. Changes in science education andachievement, 1970-1984. Oxford: Pergamon Press.

Keeves, J. P. (Ed.). (1986). Aspiration, motivation and achievement: Different methodsof analysis and different results. International Journal of EducationalResearch. 10(2), 115-243.

King, M. I. (1989). Extraordinary leadership in education: Transformational andtransactional leadership as predictors of effectiveness, satisfaction, andorganizational climate in K-12 and higher education (Doctoral dissertation,University of New Orleans). Dissertation Abstracts International, 50, 08A.

Kotter, J. P. (1988). The leadership factor. New York: Free Press.Kotter, J. P. (1990). A force for change: How leadership differs from management.

New York: Free Press.Leithwood, K., & Jantzi, D. (1990). Transformational leadership: How principals can

help reform school cultures. School Effectiveness and School Improvement, 1,249-280.

Leithwood, K., & Montgomery, D. (1982). The role of the elementary school principalin program improvement. Review of Educational Research, 52(3), 309-339.

Leithwood, K. (1992). The move toward transformational leadership. EducationalLeadership, 49(5), 8-18.

Leithwood, K.A., Begley, P.T., & Cousins J.B. (1991). The nature, causes and conse-quences of principals practices: An agenda for future research. Journal ofEducational Administration, 28(4), 5-31.

Leithwood, K., Jantzi, D., Silins, H., & Dart, B. (1992). Transformational leadershipand school restructuring. Paper presented at the International Congress forSchool Effectiveness and Improvement, Victoria, B.C.

Leithwood, K., Jantzi, D., Silins, H., & Dart, B. (1993). Using the appraisal of schoolleaders as an instrument for school restructuring. Peabody Journal ofEducation, 68(2), 85-109.

Little, J., & Bird, T.(1984). Is there instructional leadership in high schools? First find-ings from a study of secondary school administrators and their influence onteachers professional norms. Boulder, Colorado: Center for Action Research,Inc. (ERIC Document Reproduction Service No. ED 263 690).

Lohmoeller, J.B. (1987). Latent variables path analysis with partial least squares. Ver-sion 1.8. Department of Political Science, Free University Berlin, Germany.

Louis K.S., & Miles M.B. (1990). Improving the urban high school: What works andwhy. New York: Teachers College Press.

Manasse, A.L. (1984). Principals as leaders of high-performing systems. EducationalLeadership, 41(5), 42-46.

Manasse, A.L. (1986). Vision and leadership: Paying attention to intention. PeabodyJournal of Education, 63, 150-173.

Martinko, W., & Gardner, W. (1983). The behavior of high performing educational

managers: An observational study. Tallahassee: Florida State University.Morris, V., Corwson, R., Hurwitz, E., Jr., & Porter-Gehrie, C. (1981). The urban

principal: Discretionary decision-making in a large educational organization.Chicago: University of Illinois at Chicago Circle.

Pedhazur, E.J. (1982). Multiple regression in behavioral research: Explanation andprediction (2nd. ed.). New York: Holt, Rinehart and Winston.

SAS Institute Inc. (1986). VMS SAS Production release 5.16. Cary, NC: SAS Institute Inc.Sashkin, M. (1988). The visionary principal: School leadership for the next century.

Education and Urban Society, 20, 239-249.Schlechty, P.C. (1990). Schools for the 21st Century: Leadership imperatives for

educational reform. San Francisco, CA: Jossey-Bass.Sellin, N., & Keeves, J.P. (in press). Path analysis with latent variables. In T. Husen &

T. N. Postlethwaite (Eds.), The international encyclopedia of education.Oxford: Pergamon Press.

Sergiovanni, T.J. (1990). Value-added leadership: How to get extraordinaryperformance in schools. New York: Harcourt Brace Jovanovich.

Sergiovanni, T. J. (1991). The principalship: A reflective practice perspective (2nd ed.).Boston, MA: Allyn and Bacon.

Silins, H.C. (1992). Effective leadership for school reform. The Alberta Journal ofEducational Research, 38, 317-334.

Silins, H.C. (in press). Leadership characteristics and school improvement. AustralianJournal of Education.

Stedman, L. (1987). Its time we changed the effective schools formula. Phi DeltaKappan, 69, 215-224.

Sweeney, J. (1982). Research synthesis on effective school leadership. EducationalLeadership, 39, 346-352.

Tichy, N.M., & Devanna, M.A. (1986). The transformational leader. New York: JohnWiley.

Van de Geer, J.P. (1971). Introduction to multivariate analysis for the social sciences.San Francisco, CA: Freeman.

Waldman, D.A., Bass, B.M., & Einstein, W.O. (1987). Leadership and outcomes ofperformance appraisal processes. Journal of Occupational Psychology, 60,177-186.

Wolcott, H. (1973). The man in the principals office: An ethnography. New York:Holt, Rinehart and Winston.

Yammarino, F.J., & Bass, B. M. (1990). Long-term forecasting of transformationalleadership and its effects among naval officers. In K.E. Clark & M.B. Clark(Eds.), Measures of leadership. West Orange, NJ: Leadership Library ofAmerica.

Yukl, G. A. (1989). Leadership in organizations (2nd ed.). Englewood Cliffs, NJ:Prentice- Hall.

Zaleznik, A. (1977). Managers and leaders: Are they different? Harvard BusinessReview, 55, 67-68.

Downloaded by [UNIVERSITYOFADELAIDE LIBRARY] at 19:19 26 March 2012

Transf SchoolLeader

Documents