Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Feeling good about control: Design considerations for accountability systems in schools
(USC Thesis Other)
Feeling good about control: Design considerations for accountability systems in schools
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, som e thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these w ill be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. ProQuest Information and Learning 300 North Zeeb Road, Ann Arbor, M l 48106-1346 USA 800-521-0600 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. FEELING GOOD ABOUT CONTROL: DESIGN CONSIDERATIONS FOR ACCOUNTABILITY SYSTEMS IN SCHOOLS Copyright 2000 by Steven M itchel Cantrell A Dissertation Presented to the FACULTY OF T H E GRADUATE SCHOOL U N IV ER SITY OF SO U T H E R N CALIFORNIA in Partial Fulfillm ent of the Requirements for the Degree D O C T O R OF PHILOSOPHY (PUBLIC ADM INISTRATIO N) August 2000 Steven M. Cantrell R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3018059 ___ ® UMI UMI Microform 3018059 Copyright 2001 by Bell & Howell Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. Bell & Howell Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. UNIVERSITY O F SOUTHERN CALIFORNIA TH E GRADUATE SCHOOL UNIVERSITY PARK LOS ANGELES, CALIFORNIA 90007 This dissertation, w ritten by ‘ O revenu Q sw rrts'L L . under the direction of h.J.& .......... Dissertation Committee, and approved by all its members► has been presented to and accepted by The Graduate School, in partial fulfillment of re quirements for die degree of DOCTOR OF PHILOSOPHY Dam of Graduate Studies D a te A u g u s t . . Z .’...2 000 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Steven M. Cantrell Peter J. Robertson, Chair FEELIN G G O O D ABO UT CO NTRO L: DESIG N CO N SID ER A TIO N S FOR ACCOUNTABILITY SYSTEMS IN SCHOOLS T ight organization controls allegedly do more harm chan good, but th e impact of control may depend upon the design characteristics o f the control system. Over 600 teachers in 64 schools were asked to describe the characteristics of the accountability syscem in place at their school. Teachers who describe the control system as tight, but also technically and socially sound, exhibited the highest level of morale (interaction, p<.05). Conversely, teachers working under tight control systems that were not technically and socially sound experienced the lowest level o f morale. These results call into question the predominately negative assumptions regarding the impact o f organizational control on em ployee morale. This theory of control system design also offers an explanation for the equivocal empirical results previously found by researchers examining control in schools. R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dedication To Rowin and the boys, Jackson and Clay. R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Acknowledgements What is a dissertation but the fru it o f patient guidance? I indeed have had patient guides. My chairs, Peter Robertson and Paul Adler, have read ten words fo r every one included herein. Their constant encouragement and exacting standards w ill continue to shape my work fo r years to come. I also want to thank committee members Yan Tang and Ken Merchant who sharpened my writing and kept me focused on finishing this task, so that my real work might begin. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. Table of Contents D e d i c a t i o n ....................................................................................................................................................................................ii A c k n o w l e d g e m e n t s ............................................................................................................................................................. iii L i s t o f T a b l e s ...........................................................................................................................................................................vii L i s t o f F i g u r e s .......................................................................................................................................................................viii CHAPTER 1 : INTRODUCTION....................................................................................................................................1 “ D e -d e m o n iz in g ” c o n tr o l...................................................................................................................................................... 1 A c c o u n ta b ility a n d c o n tro l.....................................................................................................................................................3 E m p iric a l e v id e n c e : w h a t w e k n o w a n d n e e d to k n o w ..........................................................................................4 P ra c tic a l im p lic a tio n s .................................................................................................................................-............................6 W h a t is to c o m e .........................................................................................................................................- .............................. 7 CHAPTER 2:.............................................................................................................................................. TECHNICAL AND SOCIAL ASPECTS OF MANAGEMENT CONTROL SYSTEMS............... 8 T o w a r d a T h e o r y o f M a n a g e m e n t C o n t r o l .....................................................................................................8 T h e G o als o f C o n tro l...............................................................................................................................................................9 S tre n g th s a n d W e a k n e sse s o f th e T h r e e F o rm s o f C o n tro l................................................................................ 10 P e rs o n n e l/C u ltu ra l C o n tro ls.......................................................................................................................................... 10 A ctio n C o n tr o ls .......................................................................................................................................... - ...................... 12 R e su lts C o n tr o l.......................................................................................................................................... - ...................... 12 D e s i g n i n g C o n t r o l Sy s t e m s : ........................................................................................................................................ 13 A T e c h n i c a l C o n t i n g e n c y T h e o r y o f M a n a g e m e n t C o n t r o l .........................................................13 In c re a s in g T e c h n ic a l K n o w le d g e .................................................................................................................................... 16 Id e n tify in g K ey C u ltu ra l D riv ers, K e y A c tio n s, a n d K ey R e s u lts ............................................................. 16 A c c u ra te T ra c k in g a n d M e a s u re m e n t...................................................................................................................... 18 G e rm a n e R e w a rd s an d P u n is h m e n ts ....................................................................................................................... 19 A T e c h n ic a l M o d e l o f M a n a g e m e n t C o n t r o l........................................................................................................... 20 A sse ssin g th e C o n tin g e n c y T h e o r y o f M a n a g e m e n t C o n tro l: S tr e n g th s .................................................... 21 B road c o n c e p tio n o f co n tro l...........................................................................................................................................21 R e c o g n iz e s th a t tig h t control m ay y ie ld u n in te n d e d n e g a tiv e c o n s e q u e n c e s ....................................22 C o n s id e rs th e in fo rm atio n al r e q u ir e m e n ts n e c e ss a ry for tig h t c o n tr o l...................................................2 4 W e a k n e s s e s o f th e C o n tin g e n c y M o d e l o f C o n tr o l............................................................................................... 25 U n e x p la in e d E m p iric a l F in d in g s ........................................................................................................................ — .25 U n e x a m in e d V ariables: C o n tro l S y s te m D e s ig n a n d Im p le m e n ta tio n ...................................................28 T h e Im p o rta n c e o f A ttitu d e s .......................................................................................................................................29 A n E x p a n d e d C o n t i n g e n c y M o d e l o f M a n a g e m e n t C o n t r o l ........................................................ 3 0 S o cial C h a ra c te ris tic s : T h e E n a b lin g R a tio n a le o f th e C o n tro l S y s te m .......................................................32 D e sig n F e a tu r e s ......................................................................................................................................... 32 D e sig n I n t e n t............................................................................................................................................ 34 Im p le m e n ta tio n P ro cesses.............................................................................................................................................35 I m p le m e n ta tio n as A d m in istra tio n ......................................................................................................................— .3 7 D e s ig n in g C o n tro l S y ste m s-P a rt T w o : T o w a r d a n E x te n d e d C o n tin g e n c y T h e o r y ...........................3 7 E m p iric a l S u p p o rt............................................................................................................................................................. 38 T w o T y p o lo g ie s o f M a n a g e m e n t C o n tr o l:................................................................................................................ 41 C o n tra s tin g th e C o n tin g e n c y a n d E x p a n d e d C o n tin g e n c y M o d e ls..............................................................41 H y p o t h e s e s E m e r g in g f r o m a S o c io - T e c h n i c a l M o d e l o f M a n a g e m e n t C o n t r o l 43 iv R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3: METHODS.............................................................................................................................................44 S a m p l e a n d D a t a C o l l e c t i o n .................................................................................................................................... 4 4 S u r v e y M e a s u r e s a n d D e s c r i p t i v e St a t i s t i c s .............................................................................................. 4 6 C o n tro l S y s te m T ig h t n e s s ................................................................................................................................................4 7 T e c h n ic a l K n o w le d g e ........................................................................................................................................................ 4 8 E n a b lin g ra tio n a le ................................................................................................................................................................. 4 9 M o r a le .........................................................................................................................................................................................5 0 P e rfo rm a n c e ................................................................................................................................................. - ........................... 51 O p e n in g Q u e s tio n s ...............................................................................................................................................................5 2 L a t e n t V a r ia b l e C o n s t r u c t i o n : C o n f i r m a t o r y F a c t o r A n a l y s i s ............................................ 5 4 C o n tro l S y s te m T ig h tn e s s ( C S T ) ................................................................................................................................. 5 6 T e c h n ic a l k n o w le d g e ( T K ) ..............................................................................................................................................5 6 E n a b lin g ra tio n a le ( E R ) ....................................................................................................................................................5 7 M o r a le .........................................................................................................................................................................................5 7 P e rfo rm a n c e ................................................................................................................................................. - ........................... 5 8 E x o g e n o u s V a ria b le s ........................................................................................................................................................... 5 9 R e l a t i o n s h i p s a m o n g L a t e n t F a c t o r s : T h e S t r u c t u r a l E q u a t i o n M o d e l .......................6 0 M o d e l S p e c ific a tio n a n d E v a lu a tio n ...................................................................................................................- ....... 63 M o d e r a t e d R e l a t i o n s h i p s a m o n g L a t e n t F a c t o r s :...............................................................................6 4 R e g r e s s io n A p p r o a c h e s t o In t e r a c t i o n ........................................................................................................... 6 4 M o d e l S p e c ific a tio n s U sin g 2 S L S ................................................................................................................................ 6 5 S u m m a r y ........................................................................................................................................................................................6 9 CHAPTER 4: RESULTS............................................................................................................................................... 70 S t r u c t u r a l M o d e l : T e s t i n g t h e D i r e c t E f f e c t s ...................................................................................70 O v erall M o d e l F i t ..................................................................................................................................................................70 A lte rn a te S p e c if ic a tio n s ......................................................................................................................................................71 A lte rn a te M o d e ls .................................................................................................................................................................... 73 T w o -S t a g e L e a s t S q u a r e s S t r u c t u r a l M o d e l : T e s t in g t h e I n t e r a c t i o n E f f e c t s ....7 3 S t r o n g S u p p o r t f o r a P a r t ia l T e s t o f t h e E x t e n d e d C o n t i n g e n c y T h e o r y ................... 78 T h e Im p o rta n c e o f T e c h n ic a l a n d S o cial C o n s id e r a tio n s ................................................................................79 T w o M o d e ra to rs a re B e tte r th a n O n e ..........................................................................................................................79 T h e M e a n i n g o f a P a r t ia l T e s t ................................................................................................................................80 CHAPTER 5:................................................................................................................................................ DISCUSSION......................................................................................................................................... 82 R e a l S c h o o l s , R e a l Is s u e s ............................................................................................................................................ 82 L e C o n te S e c o n d a ry S c h o o l........................................................................................................................................ 82 E x p o sitio n H ig h S c h o o l.................................................................................................................................................83 H ig h la n d S e c o n d a ry ........................................................................................................................................................85 S o m e O b s e rv a tio n s .......................................................................................................................................................... 86 C o n t r i b u t i o n s o f t h e s t u d y ........................................................................................................................................8 7 T h e o re tic a l C o n trib u tio n s ............................................................................................................................................8 7 P ractical C o n s e q u e n c e s .................................................................................................................................................8 7 L i m i t a t i o n s o f t h e s t u d y ...............................................................................................................................................88 I m p l i c a t io n s a n d F in a l T h o u g h t s ..........................................................................................................................9 0 C o n tro l a n d C o n s id e ra tio n : N o t s u c h s tra n g e b e d fe llo w s a fte r a l l ..........................................................91 BIBLIOGRAPHY:................................................................................................................................... 93 v R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ap p e n d k A ................................................................................................................................................................................. 105 LISREL 8.30 Input: Covariance M atrix................................................................................................................-1 0 5 A p p e n d ix B: LISREL 8 .3 0 S y n t a x f o r St r u c t u r a l Eq u a t io n M o d e l ......................................................106 A p p e n d ix C: SPS S S y n t a x f o r T w o -S t a g e L e a s t S q u a r e A n a l y s is ..........................................................107 A p p e n d i x D: A d d i t i o n a l V a r ia b l e s M e a s u r e d , b u t E x t r a n e o u s t o t h e T e s t e d M o d e l 108 Role Stress.........................................................................................................................-......................... 112 vi R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. L ist o f Tables T a b l e 3-1: D e s c r ip t iv e s f o r It e m s R e l a t e d t o C o n t r o l S y st e m T ig h t n e s s...............................................48 T a b l e 3-2: D e s c r ip t iv e s f o r It e m s R e l a t e d t o T e c h n ic a l k n o w l e d g e ......................................................... 49 T a b l e 3-3: D e s c r ip t iv e s f o r It e m s R e l a t e d t o E n a b l in g r a t io n a l e .............................................................. 50 T a b l e 3-4: D e s c r ip t iv e s f o r It e m s R e l a t e d t o M o r a l e ..........................................................................................51 T a b l e 3-5: P e r f o r m a n c e : St a n f o r d A c h ie v e m e n t T e s t , n in t h e d it io n ......................................................... 52 T a b l e 3-7: D e s c r ip t iv e s f o r O p e n in g Q u e s t io n s .........................................................................................................53 T a b l e 3-8: D e s c r ip t iv e s f o r D e m o g r a p h ic s a n d C o n t r o l s (C o v a r ia t e s).................................................... 53 T a b l e 3-9: F a c t o r l o a d in g s-M o r a l e ................................................................................................................................ 58 T a b l e 3-10: F a c t o r l o a d in g s-E x o g e n o u s V a r ia b l e s ..............................................................................................60 T a b l e 4-1: Su m m a r y o f T w o -St a g e L e a s t Sq u a r e s R e g r e s s io n A n a l y s is f o r V a r ia b l e s P r e d ic t in g t h e Im p a c t o f C o n t r o l Sy s t e m T ig h t n e s s o n St u d e n t P e r f o r m a n c e ....................... 74 T a b l e 4-2: S u m m a r y o f T w o-S t a g e L e a s t S q u a r e s R e g r e s s io n A n a l y s is f o r V a r ia b l e s P r e d ic t in g t h e Im p a c t o f C o n t r o l S y s t e m T ig h t n e s s o n M o r a l e ........................................................75 T a b l e A D -1: D e s c r ip t iv e s f o r It e m s E x t e r n a l t o t h e M o d e l ......................................................................... 1 10 T a b l e A D -I (c o n t .): D e s c r ip t iv e s f o r It e m s E x t e r n a l t o t h e M o d e l ......................................................... 111 T a b l e A D -2: D e s c r ip t iv e s f o r R o l e St r e s s It e m s .................................................................................................... 1 12 vii R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. L ist o f Figures F ig u r e 2 -1 : A C o n t in g e n c y M o d e l o f C o n t r o l S y s t e m D e s ig n ......................................................................... 20 F ig u r e 2 -2 : A n E x p a n d e d C o n t in g e n c y M o d e l o f C o n t r o l S y s t e m D e s ig n ...............................................39 F ig u r e 2 -3 : T h e p e r f o r m a n c e c o n s e q u e n c e s o f m is-s p e c if ie d a n d p r o p e r l y s p e c if ie d c o n t r o l SYSTEMS BASED ON THE LEVELS OF CONTROL SYSTEM TIGHTNESS AND TASK KNOWLEDGE......................41 F ig u r e 2 -4 a E n a b l in g Ra t io n a l e : T h e a t t it u d in a l a n d p e r f o r m a n c e c o n s e q u e n c e s o f m is- s p e c if ie d AND PROPERLY SPECIFIED CONTROL SYSTEMS BASED ON THE TK/CST MATCH AND THE PRESENCE o f e n a b l in g s o c ia l CHARACTERISTICS....................................................................................................4 2 F ig u r e 2 -4 b C o e r c iv e R a t io n a l e : T h e a t t it u d in a l a n d p e r f o r m a n c e c o n s e q u e n c e s o f m is- s p e c if ie d AND PROPERLY SPECIFIED CONTROL SYSTEMS BASED ON THE TK/CST MATCH AND THE PRESENCE OF COERCIVE SOCIAL CHARACTERISTICS.................................................................................................... 42 F ig u r e 2 -5 : T h e Im p a c t o f T ig h t C o n t r o l o n M o r a l e a n d P e r f o r m a n c e : T h e M o d e r a t in g E f f e c t s o f T e c h n ic a l K n o w l e d g e a n d E n a b l in g R a t io n a l e .................................................................... 43 F ig u r e 3 -1 : T h e Im p a c t o f T ig h t C o n t r o l- A St r u c t u r a l M o d e l ................................................................... 63 F ig u r e 3 -2 : T h e Im p a c t o f T ig h t C o n t r o l- A S t r u c t u r a l M o d e l w it h T h r e e- w a y In t e r a c t io n T e r m .............................................................................................................................................................................................. 68 F ig u r e 4 -1 : P a r a m e t e r Es t im a t e s : T h e Im p a c t o f T ig h t C o n t r o l ................................................................... 71 F ig u r e 4 -2 : C o n t r o l S y s t e m Im p a c t o n P e r f o r m a n c e a n d M o r a l e : .............................................................. 73 F ig u r e 4 -3 : T h e Im p a c t o f T ig h t Co n t r o l- A St r u c t u r a l M o d e l w it h T h r e e - w a y In t e r a c t io n T e r m .............................................................................................................................................................................................. 76 F ig u r e 4 -4 : E f f e c t o f C S T o n M o r a l e a t V a r io u s L e v e l s o f E n a b l in g Ra t io n a l e (E C ) a n d T e c h n ic a l K n o w l e d g e (T K )...........................................................................................................................................78 viii R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 1 : Introduction “The predictability o f one’ s behavior is the sure test of one’ s own inferiority ” — M ichael Crozier “On the contrary, it is by an understanding o f the laws which govern the process by which authority is generated that we gain our freedom” — Mary Parker Follecc Given a choice, people will opt to choose. Free will is one of the unquestioned assum ptions that inform our liberal worldview. Indeed, self-determination underlies both the civic and religious traditions dominant in the W estern world. Ideas that challenge or threaten to unmask th e belief in self-determination have been and continue to be m et with great resistance. People are reluctant to abandon the notion that they are the ultim ate arbiters of their own action. Yet, alongside this reluctance to cede autonomy, is th e recognition that cooperative action requires som e m inim um level of subordination to th e collective will. It is accepted that norms, conventions, laws and regulations, while they represent the subjugation of the individual to the collective will, also help society function more sm oothly. W illing submission for the good o f th e collective, however, is tenuous, difficult to maintain, and often stricdy circumscribed. “De-demonizing” control Given this bent toward autonom y, it is not surprising that many students of control (and related concepts, such as bureaucracy and accountability) are preconditioned tow ard negative conceptualizations of control. Bureaucratic control in US corporate settings has been called “all encompassing”, in that it “im pinges on the behavior of individuals” to the point where it “demands the worker’s soul” (Edw ards 1979). Such criticism is not merely reserved for the overbearing, hierarchical, and rule-bound bureaucracies, but also extends to those organizations whose management attem pt to hum anize bureaucracy through programs such as Quality o f W ork Life (QWL) or School-Based M anagem ent (SBM). Critics claim that programs like QWL represent an attem pt by m anagem ent “to control the hum an side of work for their own political advantage” (Fischer 1984). 1 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. T h e findings co be presented in chis current study suggest that our understanding of control will benefit from an expanded conceptualization o f control. Specifically, this stu d y presents findings that suggest an individual’s experience of organizational control will be determ ined in part by the characteristics of the control system in place. T h ese findings suggest chat organizational control is neutral and only in its design and im plementation does control positively or negatively affect those subject to it. It may be tem pting for critical scholars and perhaps others, to interpret any attem pt to redeem control as a capitulation to the management perspective of the organization. I have chosen a study population with these concerns in mind. T he workplace setting and work activity chosen for this study, schools and teaching, are noc cast easily as struggles betw een m anagem ent and labor or betw een ownership and the working class. T his is not to say that teachers have noc struggled against management. Certainly, the history of the rise of the teachers’ unions is th e history of maltreated workers fighting for better working conditions, workplace security, and wages (Cuban 1988). T here are indications that the tension betw een teachers and m anagem ent has lessened considerably. Current job conditions, security and wages are much better today than in th e past and are now competitive with other public seccor occupations. Although the labor history o f the educational system refleccs at lease some o f the labor- management animosity found in the private sector, schools differ from capitalist organizations. One primary difference is thac schools, like other public sector organizations, operate on behalf of society at large. Education may be considered a public good. Good schooling benefits both its recipients and society at large. If education is indeed a public good, it is difficult to portray m anagement control over teachers in terms of a profit motive. M anagem ent may have some self or career interest in controlling the actions of teachers, but the motivation is not profit. Some have claimed that schools follow factory like processes in order to reduce the costs of schooling large numbers of individuals (E aton 1990). Yet, these cost savings, except in rare cases of malfeasance, seldom end up in the pockets of those who run the system. Furtherm ore, unlike capitalist organizations, the public interest is at stake 2 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. w hen schools and/or teachers fail to perform their duties. T eachers are accountable to their supervisors, but accountability extends far beyond this m anagem ent-labor relationship to the individual child, her family and the com m unities within which they reside. Accountability and control Given the importance of the teaching task, and the wide distribution o f the benefits of good schooling, it is reasonable to hold schools and teachers accountable for their work. T h e question is noc w hether controls over ceaching are appropriate, buc which forms o f control are reasonable. Accountability, by definition, requires an account, a reporting o f an agent’s action to a principal (W agner 1989). In the case of schools, the question of who is accountable to whom and for what is often complicated. Multiple principals, such as management, parents, students, various levels of community, and businesses, certainly complicate the work o f schooling (C hubb and M oe 1988). Complexity, however, does not excuse schools or teachers from giving an account o f their work to each that holds a legitimate stake. T h e public rightly demands teachers’ best efforts and deserves the assurance that such efforts have been undertaken. Some control over the educational process is desirable and potentially serves even the least powerful in society. How much control should be exercised, and in what form, is open to debate. Teachers in the United States are subject to numerous external and internal controls. Some of these controls operate before a teacher is hired for the job. In many states, public school teachers are selected to work at a particular school within a given district only after satisfying the curricular and testing requirem ents necessary to obtain a credential. O ther controls place limits on teacher autonomy in areas such as textbook selection and content determ ination. Still other controls assess teacher performance, either through an appraisal process or through stu d en t perform ance on standardized tests. Less obvious, but serving control functions nonetheless, are professional developm ent programs, strong beliefs and cultures operating within the school or departm ent, and m em bership in one of the many professional associations that propound a particular pedagogy or 3 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. content. T his list is not exhaustive. T h e m echanism s of control are myriad: formal and informal, varying in design characteristics and im plem entation intent. T his study will focus on forms o f control th at operate at the school-level or below (i.e. departm ent or classroom). T his is noc to dim inish the im portance of state-level controls. Indeed, student performance gains have recently been shown for students in states with strong controls, specifically increased teacher-credentialing requirem ents (Darling-Hammond 2000). Increasing our understanding of how control operates at the school-level is im portant because th e school represents the locus for both policy im plem entation and teachers’ evaluation of their experience. T h e school site is where most policy is im plem ented. Most controls aim ed to influence che behavior of ceachers, regardless of their origin, are worked out ac the school or even classroom level (chose pertaining to entry into che profession would be an exception). Furtherm ore, it is the local level— school, department, and classroom— chat ceachers find most salient when evaluating their experience as ceachers (T albert and M cLaughlin 1994). T hus, che school prom ises to be an excellenc setting for uncovering che characteristics o f control that influence teachers’ perceptions of their work. Empirical evidence: what we know and need co know T o date, che empirical evidence is mixed as regards che impact of organizational controls on em ployee attitudes and behaviors. Scholars have, for quite some time, recognized che adverse effects of misguided organizational control (Argyris 1951; Hofstede 1978). Several theorists have responded to this challenge by outlining the conditions necessary to avoid unintended negative consequences of control (Kerr 1975; Law ler and Rhode 1976; M erchant 1985; M erchant 1998). T h ese works have provided practitioners with theory-based guidance for avoiding and correcting technical flaws in control system design, flaws th at frequently have human consequences. More recently, theorists have identified social features o f bureaucracy relevant to th e effective functioning of che control system (Adler and Borys 1996). T h e new challenge is co exam ine both 4 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. technical and social design considerations and test w hether these design and im plem entation features help predict the attitudinal and behavioral responses to control systems. T h e control system becomes increasingly important as pressure mounts for schools to be held accountable for student performance. M y research seeks to better understand the characteristics of control systems and their impact on school personnel. I w ant to know how school administrators and teachers have responded to an educational environm ent marked by ever-increasing dem ands for accountability. In this study I seek to exam ine two issues fundamental to our understanding of how administrators influence teacher behavior. First, do school administrators have system s in place to increase the likelihood that teacher effort is directed toward school goals? I will answ er this question by describing the nature o f control systems at the school level. T his entails answering several questions: Are control systems tight or loose? Can they be described as enabling teacher work? Are they coercive? Do they direct, improve and motivate teacher work? T h e second aim of this study is to determ ine how teachers respond to adm inistrative control. T h is second aim addresses the potency of managerial controls. I will exam ine boch the attitudinal and behavioral effects of managerial control systems on teachers. T o answ er these questions, I draw upon two theoretical bases. First, m anagem ent control system theory as presented and refined by Ken M erchant in two works (M erchant 1985; Merchant 1998) provides the technical basis for control system design. This framework proposes that control system effectiveness is a function of knowledge. It is impossible to design a well-functioning control system without detailed and explicit knowledge o f the inputs, processes and/or outputs critical to the organization's success. A technically sound control system does not attem pt to control more than is possible given the current state of knowledge. T h e appropriateness of the match betw een control and knowledge, called “fit”, determ ines the technical soundness o f the control system. W hile technical soundness is contingent upon knowledge, social soundness in design is contingent only upon a manifest recognition of the human desire to contribute. Social soundness in control system design is based upon an alternate conception of bureaucracy which suggests that an 5 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. em ployee’s experience of bureaucratic control depends on key features o f the control system, including both the design and im plem entation processes (Adler and Borys 1996). A control system is socially sound when it incorporates th e features that communicate to employees that the system exists to help em ployees do their work more effectively. T hese features can be defined as those that make the system intelligible, flexible, correctable, and valid. T h is perspective anticipates that teachers will accept organizational controls as useful guidance w hen such controls are designed and im plem ented in ways that enable teachers to improve their practice. T ak en together, these theoretical perspectives adopt a neutral stance toward control. T h e y do provide, however, a compelling rationale to reject simplistic attem pts to label all forms of control as negative and instead focus on the control system characteristics that are predicted co im pact the teachers' experience of control. T h e study draws upon survey data I have collected from a sam ple population of over 1200 teachers and 64 principals across four southern California school districts. T h e analysis utilizes structural equation modeling to test hypotheses generated from each of the theoretical perspectives outlined above. Five variables, four latent and one observed, are included in the model: 1. Control system tightness — four indicators o f tightness drawn from scales that assess the presence, form, and extent of site-level controls; 2. Enabling control — a new variable that addresses control system design intent and im plementation; 3. T ask knowledge — four indicators o f perceptions and knowledge o f the task; 4. Morale — th ree indicators, including scales of job involvement, caring, and alienation; and 5. Performance — three models were tested using different observed measures o f performance, including in-role perform ance, organizational citizenship behavior, and student performance on standardized tests. Practical implications In addition to its theoretical contribution, my research prom ises to have several practical implications. I am testing hypotheses that have consequences for control system design. My theory anticipates that control systems are less stressful and more likely to increase morale when teachers 6 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. judge the control systems as both fair and effective. If variations in control systems are experienced in ways anticipated by the theory, then districts and principals will have clearer guidelines for control system design. T his research will also explore the conditions necessary for effective control. Prior research suggests that the optimal control system design is contingent on contextual factors. Knowing when to im plem ent what type of control system will go a long way toward developing more effective control. Finally, both clearer guidelines for design criteria and knowledge regarding contextual constraints will help alleviate the dysfunctional side-effects of misguided administrative control. What is co come C hapter 2 outlines a theory of technical design considerations for m anagem ent control systems. T h is provides th e necessary backdrop to the models that I will test. T h e technical model, a latent variable representation o f the impact of control system tightness and task knowledge on teacher attitudes and behavior, is also presented in C hapter 2. I introduce theory that outlines the social considerations for control system design in C hapter 3. T h is ch ap ter concludes with several hypotheses that em anate from the social model of control system s. In C hapter 4, I provide details about the survey sample, the instrument, and the scales used to test the hypotheses. T h e latent variables are defined at that time. In Chapter 5, the models are formally tested using L ISR E L 8. Finally, in C hapter 6 I discuss the implications for educational adm inistration and for public management more generally. I conclude with suggestions for future research. Appendices follow. R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. Chapter 2 : T echnical and Social A spects o f M anagem ent Control System s “We have always to study in a p la n t how fa r the authority o f the management is real, how fa r it comes from fulfillingfunction, from knowledge and ability, and how fa r it is a nominal or an arbitrary authority. ” — M ary P arker Follccc The attitude most desirable fo r receiving orders is intelligent scrutiny, willingness to suggest changes, courtesy in the manner of suggesting, an d a t the same time no prejudice in regard to what is prescribed, but the assumption that the way prescribed is probably the best unless one can show some convincing reason to the contrary. — M ary Parker Follccc Toward a Theory o f M anagem ent Control Managers, including school principals and district administrators, use control system s to encourage some behaviors and discourage others. Control system s are p u t into place to align individual behavior with the goals o f the organization. In a world of unlim ited resources, control system s would track the entire set of behaviors th at have a significant im pact on organizational perform ance. For each em ployee, th e perfect control system would identify, confront and ultim ately, prevent any deviation from those behaviors known to contribute to organizational perform ance. Such a system would yield accurate and inform ative perform ance appraisals. C om pensation, along with other rewards and sanctions, could then be allocated in perfect conformity to each individual's behavior. Resources are indeed limited. T his means that, in most cases, control systems will fall far short of perfection. For most em ployees, this comes as good news, given the undesirablility o f a workplace beholden to Orwellian surveillance. For managers accountable to various stakeholders for the organization’s performance, however, the recognition that control systems are im perfect raises concerns regarding the efficacy o f their current and future control systems. In education, as with other areas of the public sector, high stakes accountability and limited resources increase the dem ands on the control system to be both effective and efficient. In such an environm ent it 8 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. becomes crucial to be clear about the goals of the control system, the strengths and weaknesses of various control forms, and the conditions necessary for effective control. The Goals o f Control Control system s are put in place to encourage appropriate behavior (M erchant 1998). Organizational perform ance falters when em ployee behavior is inappropriate. Inappropriate behavior frequently stem s from three sources: T h e em ployee does not know what work to do, the em ployee does not want to do the work, and/or the em ployee is not able to do the work. T h e control system is established to address these problems of direction, motivation and capacity. T he control system is designed to limit or eradicate inappropriate behavior. To the extent the control system succeeds, it may be said to control the work performed by the organization. T h e organization’s work, referred to as its technology, may be characterized as a combination o f the inputs used by the organization, the transformation processes used by the organization, and the outputs of the organization (Scott 1992). Control over work can address one or more of these stages. Personnel/Cultural controls are utilized to shape organizational inputs. Action controls are utilized to directly influence em ployee behavior. Results controls are utilized to shape organizational outputs, the products of organizational transformation. N ote the goals of control always concern behavior, even when the form of control targets som ething other than behavior. For example, personnel controls are often concerned with the selection process. T hey are aimed at hiring the right person for the job. T h e right person may be defined in several ways: as having the right prior experience, the right credentials, or the right disposition— or some combination of these and other characteristics. For teachers, input controls include prior training and the possession of a teaching credential that signifies the successful' completion of a state-approved teacher-training program. T h e object of this type of personnel control is the selection process, however, the aim o f personnel controls is to assure that persons entering che organization are likely to behave in ways that further the organization’s goals. 9 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Strengths and Weaknesses of the Three Forms o f Control Personnel/Cultural Controls Some concrol over organizational inputs can be achieved through personnel and cultural controls. Personnel controls involve selection and training of em ployees. Because replacem ent of employees is often expensive the organization has incentives to increase the probabilities that the individual remains the right person for the job. It is in the organization’s best interest to select the right person for the job and once hired, train this person in ways th at upgrade skills and address deficiencies. Cultural controls involve socialization. Socialization processes are designed to establish and reinforce the widely held norms of behavior that informally, yet powerfully, direct individual action toward organizational goals. Both personnel and cultural controls have the benefit of being relatively unobtrusive. Selection processes are most obtrusive to those not selected for the job. T raining processes can be framed in such a way as to be seen as a benefit of em ploym ent. Similarly, organizational socialization often is accomplished through mechanisms that are not often perceived by em ployees as mechanisms of control. E xam ples of this include organizational lore, u nstated assum ptions, sym bols o f inclusion/exclusion, and espirit de corp. Personnel/cultural controls can be designed to address all three control problems. Direction may be provided by selection, training and cultural influences that inform potential employees of the duties and requirem ents of the job as well as the goals of the organization. Motivation may be increased w hen the right person is selected for the job and w hen the organizational culture encourages successful completion of duties. Capacity is increased through training and the need for capacity is reduced through effective selection processes. Personnel/cultural controls are also fallible. Selection processes are subject co inform ation asymmetries (M erchant 1985a). The job seeker often has more knowledge about his or her personal characteristics— including characteristics the em ployer views as m ost relevant to success on the job— than even the most careful interview process can discern. T h e job seeker may selectively 10 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. reveal inform ation, carefully chosen to positively bias an interview er's assessm ent o f one's qualification for the job. Additionally, m any public sector jobs are subject to civil service em ploym ent processes that may remove a considerable am ount o f discretion from the hiring manager (Wilson 1989). Training m ay falter as a control mechanism for several reasons. First, it is often unclear what training needs exist. In many cases, it is in che individual’s best interest to mask deficiencies. Even when individual deficiencies are known, it may be unrealistic to address these at che individual level. Thus, even good training often can address only the deficiencies of (or the skills required by) groups rather than individuals. Second, it is necessary to establish strong causal relationships betw een the skills needed, che training adm inistered, and the outcomes desired. Finally, other forms o f problem s associated with any educational endeavor obtain— motivation, quality of training/trainers, expense, etc. Cultural controls are also problematic, but for different reasons. Foremost is the difficulty of selecting individuals whose values support or strengthen th e healthy organizational culture. In che case w here th e present organizational culture underm ines the goals of the organization, new personnel may be needed to help establish strong and healthy cultures. This is a difficult task. Cultures are m ost readily formed during th e founding of an organization and during transitions of leadership. Even during these times, establishing the type o f culture that reinforces organizational goals is noc guaranteed. Where patterns of relationships are deeply em bedded, as is the case within che public school system, significant cultural change is quice difficult, if not impossible (Sarason 1996). T o further complicate matters, effective cultural control may require that several cultures be altered sim ultaneously. Research into organizational cultures suggests that organizations may not have only one culture, but several, and chac some of these will likely run counter to w hat is desirable by th e standards of the organization's leadership (M artin and Siehl 1983; Morgan 1986; Smircich 1983). For these reasons, cultural controls, despite their many attractive features, often play a supporting role in the overall control system. 11 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Action Controls Control over organizational throughputs or processes can be accomplished through action controls. Action controls are designed to lim it undesirable and encourage desirable em ployee behaviors. Action controls can include behavioral constraints (i.e. com puter passwords), preaction review s (i.e. spending approvals), and action accountability (i.e. perform ance appraisals, standardized operating procedures) (M erchant 1998). Action controls are the most direct form of control because they specifically address the actions to be taken (or not taken) by the em ployee. T h e strength o f this class of control m echanisms lies in its direct nature. Action accountability controls, in particular, provide prescriptions for action that direct and motivate em ployees toward desirable behavior. Action accountability can also redress personal limitations by prescribing the series of actions required for successful task completion. Action controls also have a downside. T h ey are often costly to administer, difficult to specify, and are sometimes associated with negative em ployee attitudes. Results Control Perform ance results may also be used to inform control systems. Results control, or output control, focuses on an organization’s end product. Results control is especially desirable when m eans are indeterm inate or equifinality obtains among these means. In many cases, it is easier or m ore desirable to assess the end product and im pute proper actions than to assess the actions them selves. In such cases, employees are rewarded for the consequences of their actions rather than for the actions themselves. R esults controls direct em ployee effort by focusing their attention on organizational goals. M otivation is increased when employee rewards are tied to the accomplishment of organizational goals. T h e focus on results also creates incentives for employees to increase their skills and abilities or move to jobs where their current skills and abilities are valuable. Results controls are typically inexpensive to operate since results are usually m easured by the organization already (M erchant 1985b). 12 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. T h ere are som e downsides to results control. First and foremost, for som e jobs th e link between employee work quality and performance is not direct. This is often said to be the case with teachers. Any reward system that operates on th e basis o f student performance m ust account for the quality of the students (Willms 1992). It requires some degree of sophistication to compare students with different background characteristics, different motivation levels, different abilities, and different educational experiences. Furthermore, many educational outcomes are the product of collective, rather than individual efforts. Designing Control Systems: A Technical Contingency Theory o f Management Control Effective control system design requires more than simply having a com plete understanding of the various available forms of control. This is not to diminish the importance of understanding the uses and limits of each of the three forms of control. Such understanding is crucial, however, it is equally im portant to have detailed and com plete inform ation about the personnel/cultural characteristics, behaviors and results that the control system is designed to encourage or eliminate. This collection of information will be referred to as technical knowledge. Technical knowledge, as will be shown shortly, is a critical condition for an effective control system. Control system effectiveness is a function of the gap betw een desired effects of the control system and the actual behaviors undertaken by employees. It is important to understand that the control system is always focused on altering behavior. Regardless of the form the controls take (personnel/cultural, actions or results), controls are designed to motivate em ployees to alter their behavior and thereby receive rewards or avoid punishm ent. Thus, the control system cannot be effective unless it influences em ployee behavior. W hen the actual behaviors closely approximate those effects desired by the control system designers, the gap is small and control is said to be “tight.” At the other end of the spectrum, “loose” controls may exert uncertain effects on individual behavior. In this case, th e gap between actual and desired effects may be large. 13 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. A control system exhibits tightness when those subject to the control system alter their behavior to avoid sanctions or to gain rewards. Tightness is a function of the em ployee response to the control system. Control system tightness is necessarily a key com ponent of control system effectiveness, however, it is not sufficient to predict the control system’s im pact on performance. T his is because tight controls have unpredictable effects on employee behavior. T ig h t controls may direct employee behavior as intended or they may direct behavior in unanticipated and negative ways. W hen this happens, controls are said to have negative unintended consequences (Merchant 1998). Control system tightness is im portant (loose controls, after all, only indirectly influence behavior), but tightness cannot be considered in isolation from other factors know n to impact control system efficacy. T h e maxim um level o f tightness is intimately connected to technical knowledge, knowledge of the task to be controlled. T his limitation means that while tight control is desirable, it is not always possible. T ight control is not advisable without explicit knowledge of how the intended object of control — personnel/culture, action or results— relates to the organization’s goals (M erchant 1985 b; M erchant 1998). Som etim es the organization lacks technical knowledge because such knowledge would be too expensive and available resources would be better deployed elsew here. At other times, technical knowledge is simply unavailable, as is th e case when the know ledge required to undertake a given task is tacit (Polanyi 1962). W ithout the requisite technical knowledge, any attem p t to im plem ent tight controls may yield poor control and unintended negative consequences for the organization. In practice, nearly every control system suffers from insufficient technical knowledge (Daft and M acintosh 1981; Ouchi 1979). T h is inhibits an organization's ability to understand the extent to which an employee's actions contribute to organizational performance. T h e control system functions best when an organization can identify its performance goals and/or th e correct actions to accom plish these goals (O tley and Berry 1980), m easure these perform ance goals and track em ployee action, and reward the employees who have taken the specified actions or achieved the 14 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. desired results (M erchant 1985a). T hus, th e control system ’s ability to contribute to organizational performance is bounded by the availability o f pertinent technical knowledge. Technical knowledge is highly desirable, but usually costly and difficult to acquire. Indeed, an im portant function of the control system is to overcome deficits in this area and increase employee understanding of the processes and outcom es vital to the organization's success. Unless technical know ledge is sufficient, the negative behavioral consequences o f over-control threaten to undermine any benefits that might be achieved through tight controls. W hen the organization lacks technical know ledge or cannot com m unicate this knowledge to the individual, this provides an opportunity for individuals and groups to pursue their own interests at a potentially great cost to the organization. Teachers, for example, frequently believe that their actions are insufficient to increase student learning and/or that the system by which they are held accountable fails to take into account th e background characteristics o f the students they teach (Willms 1992). T ig h t controls, which in this case represent over-control, may lead teachers to respond actively by teaching to the test (even worse, manipulating stu d en t responses), or passively by limiting the am ount of energy they spend on planning and teaching (T albert and McLaughlin 1994). U nintended negative consequences stem from negative behaviors such as behavioral displacement, gamesmanship, operating delays and negative attitudes. T h ese behaviors pose a serious threat to organizational perform ance (M erchant 1998). T hese behaviors increase when the control system is unable to ascertain w hether the behavioral responses it evokes are in the best interest of the organization. T h e colloquial term “working the system” is apropos. Employees, who may even share many of the organization’s goals, recognize, and often resent, that what it takes to succeed within the organization is not necessarily the same as what it takes to contribute to the organization’s success. Unknowingly, the control system rewards negative behavior. Each negative behavior exploits information deficiencies w ithin th e control system . Behavioral displacement, gamesmanship, operating delays and negative attitudes are exacerbated by incom plete information regarding critical processes, outcomes and motivators. 15 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Negative employee responses to organizational controls are commonly theorized and reported (Agarwal and Ramaswami 1993b; Javvorski and M aclnnis 1989; M erchant 1985b; Ouchi 1979). All three forms, action control, results control, and to a much lesser extent, cultural/personnel controls, have the potential to evoke negative responses. Under conditions of increased scrutiny, as found with both behavioral and results controls, the employee becomes more aware o f the need to project an image of competence. In doing so, the em ployee may distort information, focus only on those “monitored” aspects of performance, and/or ignore the true mission of the organization. A tight control system will influence behavior, for even the negative actions outlined above are behavioral responses to the control system. T h e intended purpose of the control system, however, is to direct, motivate, and enable em ployees to execute actions that help the organization achieve its objectives. A control system simply cannot serve these functions unless it is able to identify, measure, and reward the actions or results that lead to the achievement of organizational goals. Increasing Technical Knowledge Identifying Key Cultural Drivers, Key Actions, and Key Results Personnel and cultural controls are less likely than other forms o f control to result in unintended negative consequences. T his said, it is of vital importance to identify the connection betw een personnel/cultural controls and the organizational goals that they intend to promote. Personnel/cultural controls are useless and may even work against the goals of the organization unless key cultural drivers, such as the criteria for selection, the assessm ent of training needs, and/or cultural symbols and stories, direcdy relate to the organizational goals. Controls cannot be effective unless they identify the key actions and/or results that are known to impact organizational goals, specify actions and/or results that the organization can track, and specify actions and/or results that employees are capable of undertaking. Frequently, as is often the case with professional work, the actions that constitute the work are not explicitly identified. Although a given role may encompass several identifiable tasks, the m ethods and procedures to be 16 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. used are not specified in detail or standardized. Instead, ic is pare o f one's professional responsibilities to adapt one's m ethods to the exigencies of the situation. An experienced teacher, for example, may know what motivational tools to em ploy for different students. Any attem pts to prescribe such actions to novice teachers m ust be attuned to the contextual nuances that bear significantly on the ultim ate results of such action. T his highlights the difficulties organizations face when attempting to link actions to results. T h e main point is this: If one set o f actions may lead to different outcomes under different circumstances and the correct set o f actions to take in any given circum stance is not codified, then it is logical to conclude that the organization has limited knowledge regarding what actions lead to desired results. W ith limited know ledge o f the key actions, action controls will likely be m isspecified. M isspecification can lead to behavioral displacem ent, such as m eans-ends inversion where em ployees' focus on sanctioned procedure causes them to lose sight o f organizational goals. Gamesmanship, operating delays, and negative attitudes further hinder th e effectiveness of behavioral controls (M erchant 1998). Gamesmanship refers to actions taken (i.e. the creation of budgetary slack, data manipulation) to manipulate performance indicators w ithout increasing the actual performance of the organization. Operating delay refers to the time lost due to the control in place. If each employee action requires formal approval, then it follows that the approval process will consume time formerly dedicated to the work itself. Decreased em ployee creativity may be an additional cost of rigid standard operating procedures. T h e organization then loses the benefits of incremental improvements that may occur when employees are encouraged to improve the system rather than merely subscribe to it (Blau 1955). Results controls may also bring about unintended negative consequences, primarily due to misplaced focus and effort. Organizational performance will suffer if the results control mechanism in place does not measure the range of outputs necessary for the organization's success. T h e danger is that what is measured becomes the foci of effort (Kerr 1975). If what is m easured is not directly related to the organization’s key results, then the control system is unlikely to direct individual effort toward its goals. 1 7 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. In schools, for exam ple, many recognize che incom plete character of standardized tests of student performance. Standardized tests m easure only a limited range of those qualities deem ed most important for a student’s success in life. Overemphasis on such an incom plete measure leads to problems that range from “teaching to the test” (a form of goal displacement and by implication, a misallocation of time and a potential invalidator o f the instrument) to outright cheating (Kolker 1999; Koretz 1992; RAND 1994). Accurate Tracking and M easurem ent T h e organization m ust also be able to accurately track the action and measure the outcome. M ost organizations lack th e resources to track em ployee action perfectly. E ven with sufficient resources, perfect monitoring o f employee actions is too costly. Instead, most organizations choose to rely on periodic observations or reviews, external reports of employee behavior (via clients, peers, subordinates, or supervisors), and em ployee self-reports. Recognition of potential sources of bias and efforts to overcome this bias are important to accurate attribution of employee actions. For professional work, th e m easurem ent o f action is complicated by the subtle shifts of emphasis that are nevertheless critical to the eventual outcome. We return to our classroom teacher. Suppose an evaluator sat in the back of this classroom with the intent of tracking this teacher's actions. An action oft cited as vital to student learning and motivation to learn is the teacher’ s transmission of a positive attitude toward— passion for—the subject matter. If one's passion-meter is not finely tuned, it may be easy to m istake for passion such expressions as verbosity or busyness. It is tem pting to dismiss th e exhibition of passion as unmeasurable and move on to other more easily measured actions such as time sp en t on task or attendance rates and so on. T o do this, however, is to risk a dim inished focus on chose actions that, while difficult to measure, have a significant impact on desired results. T h e potential for undesirable consequences exists for results controls w hen the required outcomes cannot be m easured completely and with precision (M erchant 1985a). Com pleteness is required for reasons similar to the argumenc made above for identification. Even if the result to be 18 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. m easured is identified correctly, incom plete measures tend to direct effort toward those measurable aspects of performance. Goal attainm ent is likely to be partial if em ployees focus only on a subset of critical performance goals. Imprecise measures create a different set o f problems. First, the measure is not worth much if it is im precise. Decisions m ade based on poor information tend to be poor decisions. If the measures are used to evaluate em ployee performance, poor measures may result in the improper application of sanctions and, consequendy, negative em ployee attitudes. Germ ane Rewards and Punishm ents T h e motivational power of rewards and sanctions depends heavily on the organization's ability to measure the accomplishments o f individuals and groups. Three necessary conditions for rewards and sanctions are congruence, controllability, and relevance. C ongruence means that the rewards (and sanctions) are aligned with results that benefit (and impair) organization performance. Controllability is necessary because m otivation is impacted severely if em ployees are held accountable for results that they have no control over (Bandura 1986). Relevance refers to the rewards and sanctions themselves, as opposed to the action and performance goals that serve as the criteria. Relevance means sim ply that the rewards are desirable and the sanctions are undesirable. Otherwise, they have lost their motivational impact. Technical knowledge helps alleviate some of this dysfunction. As Mary Parker Follett (1949) puts it, [ideally] “orders come from the work, not work from the orders.” Knowledge of the desired results and rhe required m eans to achieve these results reduces th e likelihood of behavioral displacement by making explicit the goals, the means to the goals and the necessary skills required for organizational performance. Knowledge of best practice helps em ployees direct their actions toward key organizational goals. T echnical know ledge is also essential for the creation and distribution of organizational rewards. Both ill-conceived and misallocated rewards will fall short of their motivational goals and may increase negative attitudes toward the organization. 19 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. A Technical M odel o f Management Control Contingency theorises contend th at neither tight control nor technical knowledge alone is sufficient to increase the likelihood that employee behavior will correspond to organizational goals. Yet contingency theory proposes that the combination of tight control and technical knowledge is powerful. A key premise of this research is that tight control cannot occur w ithout negative behavioral consequences unless organizations can both identify and measure key actions and key results and reward employees accordingly. W hen these conditions are impossible to satisfy, controls m ust be loosened so that the level o f control system tightness is properly m atched to the level of technical knowledge. We m ust also keep in mind that technical knowledge, however essential, is not a substitute for tight controls. It is the combination of control system tightness and technical knowledge that avoids problems of over-control and under-control. T h e model below (figure 2-1) makes explicit the key insight from the previous discussion regarding forms of control: Control system effectiveness is contingent on technical considerations. It is anticipated that knowledge of control system tightness (CST) alone is insufficient to predict performance. Instead, it is the interaction between C ST and TK that determ ines the effect of the control system on performance. In the section below, I define each of the variables in this model. Control System Tightness Technical Knowledge Organizational Performance Figure 2-1: A Contingency Model o f Control System Design Control system dghtness: T h e extent that employees alter their behavior due to the presence of organizational controls. T echnical knowledge: Knowledge of th e work performed by the organization. T his includes identification of the key cultural drivers, the key actions and/or the key results that lead to or 20 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. signify successful completion of the task. T his also includes the ability to m easure or track, as well as reward, these key actions and results. Technical knowledge may be com m unicated via formal or informal mechanisms, such as suggested by the following pairs (formal/informal): job descriptions / the work at hand, procedural m anuals / tacit understandings, pre- or in-service training / observations, performance expectations and performance reviews / reputation. Performance: T h e indicators o f goal attainment. Assessing the Contingency Theory o f Management Control: Strengths T he contingency theory o f m anagem ent control incorporates a num ber of features that m ake it useful as a guide to designing control systems. First, it conceptualizes control in broad terms. Second, it recognizes the potential for control systems to evoke unintended negative consequences. Third, it explicates the conditions necessary for control to operate as intended. Broad conception of control An effective control system m ust do more than recommend effective ways of working, it must also ascertain capacity for im plem enting recommendations and attend to training needs as required. It is recognized that the existence of standardized m ethods does not guarantee increased performance. At the school level, a set of prescriptions for practice may be in operation that has no dem onstrable effect on stu d en t perform ance. At the individual level, ev en so-called “best practices” may be difficult to im plem ent— much more so than the advocate for the change in practice realizes. David C ohen's “A Revolution in O ne Classroom: T h e C ase of Mrs. Oublier” (Cohen 1990) wonderfully illustrates such difficulties. Mrs. Oublier, equipped for “revolution” with California's new math framework, completely transformed her traditional, teacher-centered teaching. N ot really, but that is w hat she thought, and this is Cohen's point. W hat Mrs. O ublier unequivocally claims to be new practice is, when critically examined, “a remarkable melange of novel and traditional material.” Cohen's lesson is that serious changes in practice require not only new learning, but also unlearning prior practices. T h e State, in 21 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. mandating a new curriculum, assum ed teachers, when told, would adopt these “best practices” and student learning would improve. Unfortunately, the curriculum did not follow its own pedagogical advice. T o learn new ways o f teaching requires concentrated effort. T elling, it appears, is not teaching. Technical knowledge, especially o f a complex and nuanced task, m ust be developed and tested in an arena larger than one's own mind, lest the teacher, like Mrs. O., overrate her own understandings. Recognizes that tight control may yield unintended negative consequences Recent efforts at increasing teacher accountability have encouraged states and districts to adopt and increase the stakes for standardized testing, standardized content and standardized practices that prescribe some aspects of teachers' work. Standardized testing is familiar to most, however, the new tw ist is a trend toward tying stu d en t performance on standardized tests to rewards and sanctions at both the school and individual teacher level. C ontent standardization includes both what experts within a given discipline believe is essential for students to know and be able to do and what state or district curriculum frameworks tell teachers they need to cover in order to prepare students for the next level. Som etim es these are related. Prescriptions for practice include recommendations contained within the above mentioned frameworks, im plied or explicit methods required to prepare students to m eet national standards, and various evaluation practices that reward specific teacher behaviors. Contingency theory provides a predictive rationale for the otherwise unanticipated negative consequences connected to the newly imposed control mechanisms. A few teachers in Texas, one of the first states to tie school and teacher rewards to standardized testing of students, were recently convicted o f correcting stu d en t answers on the TASS exam ination prior to delivering these standardized tests to the state for scoring and reporting (Kolker 1999). Such cheating is clearly a negative consequence of “high-stakes” testing. T h e legitimacy of the accountability system is in jeopardy unless the state can provide assurances that student scores reflect their own efforts, rather that some form of manipulation by the teaching or administrative staff. Similar efforts to tie Los 22 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Angeles Unified School District teachers’ “merit pay” to student performance on standardized tests has provoked strongly negative reactions from union representatives who have threatened to reject any contract with such a provision. In this case, at least some o f the strong negative response can be linked to the widespread belief among teachers that the standardized tests are not valid measures of student performance. In contingency theory language, teachers believe that the available technical knowledge is insufficient to allow for tighter controls. Teachers' responses to content standardization have been notably m ixed. Interestingly, the differences have been along predictable disciplinary lines. M ath teachers, whose disciplinary training likely consisted of a series of similar courses in a similar order, have largely embraced the set of national standards developed by the National Council for Teachers of M athematics (NCTM). One explanation for their acceptance is that teachers of m athem atics believe that the requisite technical knowledge to develop such standards exists, especially among those best trained within their discipline. In stark contrast, attem pts to develop disciplinary standards similar to N C T M 's for social science teachers have been caught up in the “culture wars” (Diggins, Fonte, Learner, London, and Ravitz 1997; Nash, Crabtree, and D unn 1998). Lynne Cheney, former head of the National Endowment for the Humanities— the agency that commissioned standard setting task forces across several disciplines— derided the eventual history standards as “politically correct” and manifesting a “great hatred of traditional history” (Cheney 1994). T h e discourse over the National History Standards shed light on one im portant fact — there is little agreem ent as to the ultim ate purpose of teaching history within the pre-collegiate curriculum. A ttem pts to control what social science teachers teach has m et resistance from those who claim that the proposed standards are technically unsound. Standardized m ethods and practices have been even more controversial than standardized content. Some claim that teaching is an art, that its methods are difficult for teachers to describe, let alone for administrators to prescribe (Huberman 1993). Others, such as those involved in creating a national board for teaching, have rejected the mystification of teaching and asserted that there are 23 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. m arked and observable differences between expert and novice perform ance, and th at these differences can be taught (Shulman 1987). When states and districts began providing pay incentives to those successfully completing the National Board Certification process, m any cried “foul”, some citing studies that minority candidates failed to attain National Board Certification at a much higher rate than their white counterparts (Bond 1998). Again resistance increased, this tim e in the form of claims o f cultural bias, as control increased, this time in the form of incentive pay. T h e underlying d ebate— w hether we know enough to differentiate am ong teachers in term s of effective practice—has fueled the fire. Considers the informational requirements necessary for tight control T h e level of knowledge regarding organizational goals and processes has long been theorized as a key determ inant of the form o f controls available to the organization and th eir ultimate effectiveness (Follett 1973 (1925); Ouchi 1979). Several empirical studies have concluded that m any organizations seem to follow the theoretically anticipated im plem entation pattern. Organizations utilize tight controls over job functions that are well understood (or easiest to identify, measure and reward) and loose or no controls over areas least well understood (Agarwal and Ramaswami 1993b; Jaworski and Maclnnis 1989; Peterson 1984). Although there is scant empirical data regarding the impact of control systems on organizational performance, several studies have examined the effect of controls on job tension, dysfunctional behavior and information asymmetry, all of which are theorized to negatively impact performance. A study of financial controls found managers in uncertain environments more willing to shift profits from one year to the next than were managers operating in more certain environm ents (Merchant 1990). This suggests that controls are less effective where knowledge is lacking. Managers are likely to be more opportunistic when the threat of exposure is limited. Additional, b u t limited, support is found for the hypothesis that increased technical knowledge reduces the dysfunctional behavior associated with tight control. A replication and extension of Jaworski & M aclnnis’s (1989) study found that technical knowledge decreased dysfunctional behavior in two cases: 1. W hen higher 24 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. levels o f procedural know ledge accompanied process (action) controls'. 2. W hen higher levels of both procedural knowledge and performance knowledge accompanied output controls (Agarwal and Ramaswami 1993b). Although th e empirical support is not unequivocal as regards the effects of controls on undesirable attitudes and behaviors, it does suggest that when controls are in place, technical knowledge moderates this relationship and serves to reduce dysfunction. Weaknesses o f the Contingency M odel o f Control T h e contingency m odel of control system effectiveness, though useful, is incom plete. Results from empirical studies suggest that the structure-task relationship does n o t fully explain the variance in employee responses to control systems. T his may be due to incom plete specification. T h e model overlooks a critical driver o f control system effectiveness — the control system design and im plem entation process. Finally, although som e statem ents of the contingency theory recognize that em ployee attitudes impact performance, an expanded contingency theory should include employee attitudes as an outcome variable (or, at th e least, as a mediating variable). Unexplained Empirical Findings T h e available empirical evidence does not satisfactorily resolve the dilem m a w hether the net effect of bureaucratic structures, such as control systems, on attitudinal and behavioral outcomes is positive or negative. T h e research on formalization suggests that the structure-task fit alone is insufficient to predict attitudinai and behavioral outcomes. Formalization is th e aspect o f bureaucracy, along w ith centralization, m ost studied by organizational scholars. Formalization refers to formally prescribed work practices. T h ese practices are standardized through role and rules specifications. Formalization facilitates rationalization, as formalized structures have been designed to increase visibility and accountability for the critical work flows (Scott 1992, p32). Formalization also serves to separate the individual actor from the 1 It is important to note that process controls increased stress at all levels of technical knowledge, and for this reason, the authors concluded that managers should not rely on process controls. 25 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. office and the office from dependence on human foibles like emotion and opinion. By objectifying the office, the organizational hierarchy can more effectively control its m em bers’ behaviors (Zucker 1977). In order to understand better the mixed empirical findings regarding formalization's impact on em ployee attitudes and behaviors, it is useful to locate the debate more generally as concerning bureaucratic structures (formalized control systems being a type of bureaucratic structure) and their power to direct employee behavior. Organizational scholars have frequently view ed bureaucracy in conflicting term s. One branch has viewed bureaucracy in terms of its power to coerce effort from refractory employees and the other branch has focused on bureaucracy’s technical efficiency (Adler & Borys 1996). A ccording to the first perspective, bureaucratization is an expression o f th e desire for predictability and organizational control and necessarily constrains individual autonom y. Since people in general value autonomy, predictability for th e organization via tight controls comes at a cost, for such control undermines desirable attitudinal and behavioral outcom es o f its employees (Argyris 1964; McGregor 1960; M cN eil 1986). T h e se costs will be particularly high for “non routine” tasks like teaching, where initiative and creativity are required for effective functioning. T hus, bureaucratic control is seen as a threat, not only to teacher autonomy, b u t also to effective execution of the role itself (Johnson 1990). T h ere are several studies that support a negative assessm ent of organizational control. Hall’s (1968) research on the effects of bureaucracy on professionals infers that professionals hold mildly negative attitudes toward most aspects o f bureaucracy2. H e found that organizationally developed procedures conflict with professional orientation (attitudinal attributes indicative o f low morale). O ther studies have found higher levels o f formalization and centralization to decrease satisfaction among both private sector employees (Agarwal 1993; Agarwal and Ramaswami 1993a; Argyris 1964; 2 Em ployee acceptance of technical com petence as the criteria for employment was the exception to the anti-bureaucratic sentiment. This is a form of input control and as such would not be predicted to negatively influence em ployee attitudes and/or behavior. 26 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. McGregor 1960; Ramaswami, Agarwal, and Bhargava 1993) and public sector em ployees (Conley, Barcharach, and Bauer 1989; Firestone and Bader 1992). Several studies have also shown positive correlations betw een the control-related features of bureaucracy and negative experiences such as role stress. Organ and G reene (1981) found formalization to increase role conflict. Studies of sales em ployees have found formalization positively correlated with role conflict and role am biguity (Agarwal 1993; Dubinsky, Michaels, Kotabe, Lim, and M oon 1992; Johnston, Parasuram an, Futrell, and Black 1990; Ramaswami, Agarwal and Bhargava 1993). Studies in education show teachers in more highly bureaucratic schools to experience greater role conflict and role ambiguity (Pierce and Molloy 1990). Alternatively, scholars who have focused on the technical features of bureaucracy have emphasized bureaucracy’s unparalleled capacity for efficient coordination of work. T h e emphasis here is on organization as a form of cooperation. Given some degree of goal congruence, structures designed to facilitate and regulate work will be accepted if they are viewed as contributing to goal attainment. T hese scholars support their claims with research showing bureaucracy’s positive effects on target attitudinal and behavioral outcomes. Freeston (1987) found formalization to be associated with increased teacher commitment. Additionally, several outcomes of bureaucracy, namely clarity of goals, responsibilities, and procedures, have been shown to enhance teachers’ view of their work and their self-efficacy as teachers (Smylie 1990). Bullough and associates found member-teachers who adopted the National Diffusion N etw ork’s highly rationalized methods felt greater efficacy and security (Bullough, Gitlin, & Goldstein, 1984). Indeed, teacher efficacy has been associated with the availability of information regarding teaching and student performance (Bandura 1986) and with a measure of certainty regarding what constitutes effective performance (Smylie 1988). Since teachers overwhelmingly desire to feel efficacious, they often welcome efforts to enhance teacher effectiveness, even when those efforts increase centralization and formalization (Smylie 1990). In short, teachers want to do a good job and welcome structures and processes that are helpful. 27 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. T hese conclusions have been echoed in the role stress literature. Several studies suggest that formalization is non-signiflcantly or negatively correlated with role conflict and/or role ambiguity (Michaels, Cron, Dubinsky, and Joachim sthaler 1988; Organ and G reene 1981; Podsakoff, Williams, and T odor 1986). A meta-analysis of role stress studies found formalization to be weakly negatively correlated with role conflict and with role ambiguity (Jackson & Schuler 1985). Still other research suggests that while the direct effect of formalization may be negative or neutral, formalization’s indirect effect through mediators such as role stress, can negate or even reverse any demoralizing effects of formalization (Michaels, Cron, D ubinsky and Joachimsthaler 1988; Organ and Greene 1981; Podsakoff, Williams and Todor 1986). A nother possibility is the existence of decreasing marginal benefits for formalization. Engel (1969) suggested that the relationship between formalization and satisfaction may be curvilinear. In her study of physicians, those working within a context marked by m oderate levels of formalization perceived greater autonomy than did physicians working in either more or less bureaucratic settings. C ertainly confusion abounds regarding the im pact of control structures on attitudinal and behavioral outcomes. It might be said that since few of these studies directly test the contingency hypotheses set forth earlier, the claim that the contingency theory is incom plete is premature. It may be that the contingency theory is largely untested. It remains true, however, that the equivocal findings from the above studies cannot be explained by job type (i.e. level o f professionalism), public or private sector, or any other variable related to technical knowledge. Similar occupations and job sectors experience control differently. D egree of control paired with these task and job related variables is seemingly unable to capture all of what matters to individuals as they report their responses to bureaucratic structures such as control systems. Unexamined Variables: Control System Design and Implementation A series of direct tests of contingency theory hypotheses suggest that incorporating design and im plem entation within the contingency m odel could increase its explanatory power. T h e contingency perspective predicts that the most effective form of m anagem ent control will be 28 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. contingent on the nature o f the task to be controlled. Managerial forms that fit th e task would elicit favorable attitudinal and behavioral outcomes. M isfits would not. T h is is the basic prem ise o f contingency theory (Scott 1992). A few studies in educational settings have shown support for th e link betw een task and structure, but not for th e link between the task-structure m atch and positive attitudinal and behavioral outcomes. An early study by a group of Stanford researchers concluded that non-routine forms of teaching are associated with organic (less bureaucratic) forms o f m anagem ent (Cohen, Deal, Meyer, & Scott 1979). A recent extension o f this work also supported this half of the basic prem ise of contingency theory, but failed to find support for the hypothesis that organic forms of m anagem ent elicited favored attitudinal and behavioral outcomes from teachers engaged in non routine work (Rowan, Raudenbush, & Cheong 1993). It is not clear that those in non-routine jobs prefer less bureaucratic organizational forms. Sw idler’s (1979) research on alternative schools shows that these non-bureaucratic structures can lead to teacher burnout, in part, because o f the considerable tim e and energy spent on decisions th at would otherw ise be prescribed. M ore recently, studies have revealed the mixed effects of school-based m anagem ent on teacher morale (Malen, Ogawa, and Kranz 1990). Teachers, it seems, are not more favorably disposed toward organic/non-bureaucratic adm inistrative structures. T h ese researchers suggest that failure of organic m anagem ent to fulfill its expected role of facilitating q uality teaching and stu d en t achievem ent m ay be explained by inconsistent and partial implementation of these management systems (Rowan 1990; Rowan e t al. 1993). T h e Importance o f Attitudes Since the control system is designed to alter behavior in ways that help organizations attain th eir goals, it would be useful to track the attitudinal effects o f th e control system. Em ployee attitudes are the subject of numerous studies. Several early studies failed to find any strong relationship betw een satisfaction and perform ance (Petty, M cGee, and C avender 1984; Vroom 29 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1964). One influential meca-analysis reported an average true score correlation of 0.17 betw een job satisfaction and performance (Iaffaldano and M uchinsky 1985). T here is evidence to indicate that these conclusions are valid only at the individual level. At the organizational level, the link betw een em ployee attitudes and behavior is m uch stronger (O stroff 1992). O stroffs study o f 298 schools found that teacher satisfaction explained a significant portion of the variance across several school performance variables, including student math and reading achievement. Recent studies have given new life to the once dead issue of “happy workers are productive workers” (Wright and Cropanzano 2000), however, the direction of causality remains disputed (Riggs and Knight 1994). W hether attitudes cause behaviors or vice versa, attitudes are likely useful indicators of w h eth er th e control system will achieve its intended effects. Furthermore, if behaviors and attitudes have a reciprocal relationship a change in attitudes may serve as an early indicator of a change in behaviors. An Expanded Contingency M odel o f Management Control T hus far, emphasis has been placed on the fit between th e nature of the work task and the structure of the control system. Indeed, the interrelationship betw een these should inform control system design. As implied by the critique of the contingency m odel of control system design, there is an additional consideration, the social or human element, too often relegated to secondary status in much of the control literature. T his is not to malign the several fine analyses of behavioral responses to managerial control. Even in the best of these, however, individual behavior is analyzed as a response to the task-structure relationship. Few of these models, if any, include attitudinal variables. T his extension to control system theory gains its insight from recent theoretical work that attem pts to disentangle the mixed results reported by researchers examining employee responses to formalization (Adler and Borys 1996). Previous research, Adler and Borys maintain, is based on an underdeveloped conceptualization of formalization. Earlier conceptualizations have focused on the degree of formalization, but have neglected the social relations governing the design, content, and 30 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. implementation of these formalized structures. This expanded conception explicitly recognizes that em ployee responses to formalized structures are based in part on these social features of formalization. A theory that attends to these social aspects of formalization is potentially useful in explaining previously mixed empirical results. It is my intention to test this theory as it relates to control systems. T h e basic propositions of the contingency argument are not at issue. T h e extension proposed leaves the basic theory intact. W hat is added to the theory, however, is non-trivial. T h e addition of social characteristics to the model increases its predictive ability by providing an explanation for the em pirical observation that behavioral responses to similarly (or even ideally) m atched control systems frequently differ. T his extension also models the effect of control systems on attitudes. T h e addition of this outcome variable within the model makes explicit what was merely implied in previous contingency theoretic treatm ents of control systems: Mis-specifled control systems have negative attitudinal consequences in addition to the specified negative behavioral consequences. T h e basic contingency model outlined above will now be expanded to include the social characteristics of control systems. T hese social characteristics have been informed by and culled from a diverse stream of research and theory, from software design, to procedural justice and trust. This expanded contingency model anticipates that individuals' responses to the control system will be based not only on the task-structure match, but also on perceptions of enabling rationale informed by control system design characteristics, implementation processes and experience. T h e consistent message from the m ixed research findings may be stated as follows: Formal controls have the capacity to improve work, but this capacity is limited by the potentially demoralizing effect of formalized controls. T h e critical task is to identify those characteristics of the control system (including its im plem entation and design processes) that facilitate work and those characteristics that demoralize. 31 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. Social Characteristics: T he Enabling Rationale o f the Control System T h e work of explicating these social characteristics has ju st begun, for though previously theorized (Adler & Borys 1996), prior conceptualization has b een incom plete and not empirically tested. T h ese social characteristics are theorized to be im portant contributors to th e employee experience of control systems. T h ese social characteristics determ ine w hether th e rationale of the control system is perceived by em ployees as enabling or coercive. Enabling rationale is defined as the qualities of control system design and implementation features theorized to both facilitate work and extend employees' capability. I assess enabling rationale by examining two groupings o f social characteristics— those relevant to control system design characteristics and processes and those connected with control system im plementation. First, there is design. T his se t of characteristics incorporates the features of th e control system theorized to affect enabling rationale and design intent. N ext, there is im plem entation. T his second set o f social characteristics focuses on the implementation processes and chose who administer the control system. Design Features A control system with an enabling rationale will incorporate features that those subject to the system consider effective in guiding their task performance. Effectiveness is th e key criterion for enabling design features. Control system effectiveness is directly related to em ployee adoption of the goals and methods prescribed by the control system. T h e question becomes how to craft the control system so em ployees willingly comply. Willing com pliance preserves the em ployees’ autonomy and increases the chance that employees’ will use cheir creativity to im prove the system (Blau 1955). T h e design features that enhance a control system’s enabling rationale have been adapted from the software design and organizational learning literatures (see Adler & Borys 1996). This appropriation makes sense w hen the control system is conceptualized, first, as a technology for managing work that relies significantly upon the interface betw een the control system and the user, 32 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. and second, as an opportunity for organizational learning w herein the control system em beds “best practices” within organizational procedures. Four features characterize a control system with enabling rationale— perceived validity, correctability, flexibility, and intelligibility (Brown and D uguid 1992; Leonard-Barton and Sinha 1993). T hese features, while they often overlap in practice, are conceptually distinct. Validity refers to em ployee understanding of the system rationale. V alidity im plies good judgm ent. It is im portant that the control system advocates sound practices and outcom es lest individuals be forced to choose betw een being a good organizational citizen and being a good practidoner (G unz and Gunz 1994). Correctability is the process of allowing for mutual adjustm ent. Correctability is a function of the relationship betw een the em ployee and the control system . T h e control system and the practitioner w ork sim ultaneously to im prove the other. T h e control system directs employees toward best practices, while the practitioners alter the control system as necessary to maintain its fidelity to the best features of practice. While adjustm ent is necessary, so is flexibility. Sometimes, especially in com plex work, aspects of “best practice” are idiosyncratic. F or these aspects, no am ount of fine-tuning, however well intentioned, will help the control system direct em ployees toward best practice. Flexibility indicates that th e system is willing and able to m eet individual needs. Flexibility recognizes the idiosyncratic, and allows room for individualization when individualization is most prudent. Finally, control systems that exhibit the quality of intelligibility prom ote understanding of one's role and one's contribution to th e organization as a whole. U nderstanding also extends to the control system itself. Employees are b etter able to judge a control system's effectiveness if they understand th e role the control system plays in coordinating all individual roles to m eet organizational objectives. Thus, a key feature of the usable control system is that the system itself helps individuals understand its enabling rationale. T hese four features combine to produce a control system that not only m akes sense to those subject to it, b u t one that also recognizes where it cannot function and where it m ight err. T hese 33 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. characteristics seem especially relevant to control system s such as teacher evaluation where, depending upon the adm inistrative values underlying th e system , the rationale behind the evaluation and its various com ponents can be transparent or opaque to the teacher. W hen well understood, teacher evaluation can become a tool for im provem ent. Otherwise, the im provem ent claims tor teacher evaluation can become a cause for cynicism among educators (M cLaughlin & Pfeiffer 1988). Design Intent T h e design process is also important in establishing enabling control systems. T h e process of designing a control system for enablem ent is one that makes the system ’s enabling rationale salient from initial steps o f the design work right through trial runs. Although this now seems obvious, the long history of failed implementation is rife with examples o f designers’ neglect of both user needs and the cultural realities of the workplace (Sarason 1982; Wildavsky 1979). This is enhanced when che users— teachers in this case— are involved in the design process. User participation in design is not enough, however. T h e quality of the participation is what m atters. Research on participative decision making in schools has generated equivocal results (see M alen, Ogawa, & Kranz 1990). R ecent studies have found th e positive effects of participative decision making on innovation to be contingent upon conditions that support the quality of participation (Robertson, W ohlstetter, & Mohrman 1995). Research in software design, where user input is critical, provides a useful way to think about th ese issues. Control systems, like software, are organizational technologies, m echanism s for transforming inputs into outputs (Scott 1992). Successful design, in both cases, requires chat users find the technology helpful in producing their intended outputs, be they informed scudencs or eye catching illustrations. This research identifies four aspects o f the design process that enhance che quality of user participation: 1) a focus on users at the outsec; 2) an emphasis on enablem ent throughout the process; 3) user involvement in testing; and 4) che willingness and ability o f the designer to make changes based upon user recommendations (Gould 1988). 34 ^ — — i --------------------- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Such a design process will enhance teachers’ trust in the control system and school leadership. W hen users trust the control system, they are more likely to rely on it to improve their work. Users will rate a control system characterized by a design process with enabling rationale as more effective than an identical system designed w ithout regard for user input, feedback, and testing. Thus, a design intended for enablem ent yields trust in the control system. Individual perceptions o f the system, especially those concerning trust, inform action. Giddens (1990) states that trust in abstract systems is dependent on both the individual's experience and one's perceptions of the correctness of the principles underlying the structure. It seems odd to speak o f control systems building trust. Indeed, m ost theorists understand trust as either a substitute for control (Luhm an 1979) or a com plem ent to control (Sydow 1997/8). In order to explain the reciprocal relationship between trust and the control system, it is necessary to form a conception of trust wherein trust facilitates control. Following G iddens (1984), the control system, like all structures, has the capacity to both constrain and enable individual action. While the structure indeed influences action, the individual is not merely subject to the structure, but has the capacity to reflect on the structure and act accordingly. If the employees trust that the system's intent is to facilitate work and extend their capabilities, they will work with the control system rather than against it. Individual trust in the control system is essential to the system 's ability to productively adapt and, thus, continue to be seen as trustworthy. A control system designed for enabling rationale encourages individuals to suggest modifications that increase the control system ’s ability to facilitate work. T his ability, combined with a positive experience with the control system, increases trust in the system . T he cycle is reinforcing. Implementation Processes Im plementation processes matter, even for control systems designed with enabling rationale. Control system implementation includes those processes related to the administration of the control system. It is, in effect, the main interface betw een the individual and the system itself. As such, 35 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. characteristics of im plem entation procedures are often difficult to separate from characteristics of those individuals responsible for administering th e system . T h e distinction m ade here is between the actions taken by adm inistrators and the qualities o f th e administrators themselves. Neutrality, voice, and respect characterize th e actions that support fairness judgm ents and enhance the enabling effect of a control system . T h e characteristics o f administrators that enhance enabling rationale will be discussed in the following section. T h e main thrust of chis argum ent pertains to how the control system elicits compliance. T h e nature of the system im plem entation determ ines w hether employee com pliance is voluntary or coerced. Voluntary com pliance is possible because, as noted above, system s designed for enabling rationale contain features that facilitate work. W hile the control system’s features are essential to its enabling rationale, the im plem entation and adm inistration of the control system is also of great im port. T he quality of th e im plem entation process affects how the em ployee responds to the control system. Research on procedural ju stice alerts us to how th e im plem entation process affects an individual's willingness to com ply. It is clear th at unfair procedures lead to a host of negative behaviors and attitudes, including dissatisfaction and non-compliance with organizational rules and procedures (Lind and T y ler 1988). Fair im plementation procedures, by contrast, foster employee attributions of organizational legitimacy, which increase compliance (T h ib au t and Walker 1975; T yler and Lind 1992). In brief, individuals' perceptions o f justice are based on their assessment of the neutrality of the procedures, the consideration given the individuals during the implementation process, and the provision of opportunities for individuals to express th e ir opinions (T yler & Degoey 1995). Individuals who perceive higher levels o f involvement and greater outcome control are more likely to attribute fairness to the system . T h is combination o f involvem ent and voice affirm s their m em bership statu s and induces com pliance (T yler 1990). It follows that im plem entation processes that neglect these considerations threaten com pliance, underm ine enabling rationale and, by extension, impair the functioning of the control system . 36 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. Implementation as Administration A trustworthy adm inistrator can increase support for the authority o f the organization. I will follow Mayer, Davis, & Schoorm ann's (1995) conception of trust as based in com petence, benevolence, and consistency. T hese bases are applicable to both interpersonal and system trust. It is important to exam ine em ployee trust in both these areas because the system as im plemented may differ from the system as designed. T ru st in administrators or managers has been associated with outcomes such as organizational com m itm ent (Kramer and T yler 1996), cooperation (Axelrod 1984), and organizational citizenship behavior (Konovsky and Pugh 1994). R ecent investigations show that trust matters m ost in the face of negative outcomes (Brockner, Siegel, Daly, and T yler 1997). This is especially im portant for control system s due to the potential they hold for conveying information that the recipient may construe as negative. Performance evaluations may report poor performance or include suggestions for improvem ent, the organization may prescribe methods that are difficult to master or require much training, stakeholders may report dissatisfaction with the organization and/or its m em bers, or a change in incentives for performance may not value a given skill set. Em ployees who trust their adm inistrator maintain a positive connection to the organization, despite the sometimes discouraging effects of the control system. Designing Control Systems-Parc Two: Toward an Extended Contingency Theory Including enabling rationale within the m odel focuses attention on W eber's (1968:933) claim that bureaucratic power m ust “justify itself’ through either rational rules or personal authority. If W eber truly saw bureaucracy as “janus-faced” , as G ouldner (1954) attests, the addition of enabling rationale is a necessary corrective to prior uni-dim ensional conceptualizations. Enabling rationale helps reveal w hether em ployees subm it to organizational domination by virtue o f shared interest or by virtue of authority. W eber's (1968) discussion o f authority highlights two potentially competing sources: the office itself and knowledge. Pow er is exercised differently according to the source of authority. T h e office holder maintains power through discipline. T h e know ledge holder exerts power through technical expertise. A negative assessm ent of bureaucracy is more likely if one 37 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. emphasizes the authority o f the office, and assumes that the office holders draw no authority from their technical expertise. In such cases, bureaucracy is essentially coercive. A positive assessment o f bureaucracy, by contrast, em phasizes technical efficiency. Authority stem s from the task and employees comply because the rule or the method facilitates their work. T h e addition o f enabling rationale extends the model in ways that help reconcile the mixed empirical results reported above. T h e contingency model correctly identifies the technical aspects of control as necessary conditions for effective control. T he addition o f enabling rationale extends this insight to other aspects of the control system, namely design and im plem entation processes. It is not enough that the control system is technically sound, users must recognize the control system as technically sound. T h e design and implementation processes are means toward this end. If users judge a system to be effective, trustworthy, and just, it is rational to conclude that users will interpret their com pliance with such a system as being based on m ore than adm inistrative discipline. Furthermore, if they trust their administrator and agree to som e extent with the goals o f the organization, com pliance with the system will overlap to some ex ten t with their own self- interest. T he system becomes a way to meet mutually held goals. It comes as no surprise that previous studies have not satisfactorily explained the relationship betw een features of bureaucracy and the attitudes and behaviors associated w ith morale and performance. Even the “tru e” tests of the contingency framework explicated herein —those thac assess both control system tightness and som ething akin to technical knowledge— necessarily confound enabling bureaucratic structures with those structures intended to coerce effort. W ithout this additional specification by type of bureaucratic structure, it is reasonable to expect wide variance in attitudinal and behavioral outcomes. Empirical Support Each component of enabling rationale has previously been linked to favorable attitudinal and behavioral outcomes. T h e characteristics and consequences of enabling rationale, such as social support (Bacharach and Bamberger 1990), locus o f control (Fusilier, Ganster, and Mayes 1987), 38 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. organizational com m itm ent (King and Sethi 1997), and perceived influence (Jackson 1983), have had moderating or direct effects on morale-related attitudes and behaviors. A system that facilitates individual effectiveness, increases efficacy and decreases burnout, results in more positive attitudes toward the job, the organization and its clients (Kahili 1988). Effectiveness, at the individual level, is a com ponent of efficacy and an antidote to burnout (Maslach and Jackson 1981; Russell, Altmaier, and Velsen 1987). At the group level, groups that see themselves as more effective tend to set higher goals for themselves (Prussia and Kinicki 1996) and report greater satisfaction and organizational com m itm ent (Riggs and Knight 1994). T h e burgeoning literature on trust points toward similar, favorable attitudinal and behavioral outcomes. Interpersonal trust, an employee's trust of a supervisor, results in greater motivation and lower turnover intentions (Costigan, liter, and Berman 1998). T hese are both indicative of higher morale. At the organization level, the consequences o f employee attributions o f procedural justice include increased organizational com m itm ent (Kim and Mauborgne 1993), satisfaction (Folger and Konovsky 1989; Taylor, Tracy, Renard, and Harrison 1995), and job involvem ent (T ang and Sarsfield-Baldwin 1996). T h e enabling com ponents are an overwhelmingly positive influence on morale-related attitudes. Control System Tightness Technical Knowledge Enabling Rationale Organizational Performance Morale Figure 2-2: An Expanded Contingency Model of Control System Design Enabling rationale as a moderating variable potentially increases the explanatory power o f this m odel of control system effectiveness. Both technical knowledge and enabling rationale are 39 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. predicted to m oderate th e relationships betw een control system tightness and th e outcom e variables, morale and performance. An effective control system is only as tight as is technically feasible. An effective control system is also designed and im plem ented to enable em ployees to work more productively. Enabling rationale is anticipated to moderate th e relationship between tight control and morale and perform ance. E nablem ent minimizes th e potentially harmful effects of tight control (the potential for harm exists even under conditions of high technical knowledge). Enabling rationale helps clarify w hether the net effect of C ST on performance is positive or negative. T h is is to say that CST's impact on performance depends on the social characteristics o f the control system . If the control system is perceived as enabling by em ployees, then both morale and perform ance may increase. M orale increases because the system clearly values individuals. Perform ance increases because enabling control (control with enabling rationale) provides a com pelling rationale to em ployees otherw ise loathe to cede autonom y. In this way an enabling rationale m inim izes resistance to tight control. T his allows the control system to work as designed, not opposed or otherw ise subverted by disgruntled or m isguided employees. A control system designed and im plem ented with an enabling rationale counteracts the demoralizing attitudinal effects and the dysfunctional behavioral effects of tight controls m ost closely tied to the loss of autonom y. More importantly, enabling control serves as a su b stitu te for im perfect m onitoring. Em ployees who voluntarily com ply with the control system recognize its benefits and becom e increasingly self monitoring. T h e variables new to this model are defined as follows: Enabling rationale: the extent that employees perceive the control system to facilitate work. Morale: a basic psychological state indicative of one's attitudes coward work and the workplace (Organ and Ryan 1995). 40 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Two Typologies o f Management Control: Contrasting the Contingency and Expanded Contingency M odels T he first typology (see figure 2-3) flows from the initial contingency m odel of control. Given a level of control system tightness, the behavioral or performance outcomes are contingent on the level of technical knowledge. T ighter control requires more technical knowledge. If the technical knowledge available is insufficient to identify, measure or track, and reward key results or actions, then attempts at tight control are ill-advised. C ontrol hi S ystem T ightness io 2-3: The performance consequences of mis-specified and properly specified control systems based on the levels of control system tightness and task knowledge. T h e second set of typologies reflects the extended contingency model. Tw o figures are needed to reflect the three dimensions thought to im pact control system effectiveness. T h e upper typology (fig. 2-4a) addresses the consequences of varying levels of technical knowledge and control system tightness within the context of an enabling rationale. In this case, effective control combines high levels of control system tightness with high levels of technical knowledge. In the lower figure (fig. 2-4b), the com bination o f high control system tightness with high technical knowledge evokes negative responses from employees due to the coercive nature of the control system. In th e lower right hand com er o f both figures, high levels o f technical knowledge signify a missed opportunity for control. T each ers may resent this m issed opportunity, especially w hen com bined with administrative efforts to rationalize an essentially “toothless” control system. 41 T ask K n ow ledge lo hi Mis- Properly spedfied: spedfied: Negative Positive Properly Mis- spedfied: specified: Positive Negative R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Technical K now ledge lo hi C ontrol 1 System T ightness Mis- specified: Negative Enabling Control: Positive No Control: Mis- lo Neutral specified: Negative Enabling Rationale Figure 2-4a Enabling Rationale: The attitudinal and performance consequences of mis- specified and properly specified control systems based on the tk/cst match and the presence of enabling social characteristics. Technical K now ledge lo hi Mis- Coercive C ontrol hi specified: Control: System Negative Negative T ightness Mis- No Control: Specified: lo Negative Negative Coercive Rationale Figure 2-4b Coercive Rationale: The attitudinal and performance consequences of mis- specified and properly specified control systems based on the tk/cst match and the presence of coercive social characteristics. 42 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. H ypotheses Emerging from a Socio-Technical M odel o f M anagement Control T h e typologies above illustrate th e potential value of including a second dimension in this examination of control systems. Figure 2-5 shows how both technical knowledge and enabling rationale serve to moderate the relationship betw een control system tightness and the attitudinal and behavioral outcome variables. T h e hypotheses below state formally the expected impact of these dual moderators. Enabling R ationale M orale Control System Tightness Perform ance Technical know ledge Figure 2-5: The Impact of Tight Control on Morale and Performance: The Moderating Effects of Technical Knowledge and Enabling Rationale Each path that connects two variables or moderates the relationship between two variables is an implied hypothesis. N ot all hypotheses im plied by the model, however, are of equal im portance. This model is designed to test four hypotheses essential to th e extended contingency theory of control system design. H I: T echnical knowledge m oderates the relationship betw een control system tightness and performance. H2: T echnical knowledge m oderates the relationship betw een control system tightness and morale. H3: Enabling rationale moderates the relationship between control system tightness and performance. H4: Enabling rationale moderates th e relationship between control system tightness and morale. 43 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 3: M ethods Sam ple and Data Collection T h e nature o f this research required th e collection o f new data. Several of the hypothesized relationships have been previously studied in educational settings, however, prior em pirical studies have not exam ined the social and technical dim ensions o f control. Therefore, it was necessary to build a database th at contained m easures o f th e variables in both previously and newly hypothesized relationships. T h e bulk of the data was collected via a survey adm inistered to nearly 1200 teachers across four Southern California school districts and to principals at 64 schools. O ther data, such as school dem ographic and stu d en t perform ance data, w ere provided by th e California D ep artm en t of Education. Some socio-economic data came from CALWORKS, formerly AFDC (Aid to Families with D ependent Children). T h e schools in the survey sample reside in both urban (44) and suburban (20) attendance areas. T h ese schools were chosen, in part, because their student population reflects the diversity among those who live w ithin “the urban fringe” of Southern California. T h e sam ple consists of schools whose student populations, collectively, are 6% African American, 12% Asian, 45% Caucasian, 34% Hispanic and 4% other non-white. Among these schools, one has a student body whose majority is African American, 35 have a Caucasian majority, 16 have an Hispanic majority and in 12 schools no single racial group makes up a majority o f the stu d en t population. Economically diverse also describes the student population at schools within this sample. Free or reduced lunch is provided to 50% or more of the student body at 29 o f the schools, to 25% or more at an additional 11 schools. At 11 o f the schools, over one-third of the student population receives CALWORKS aid. A t over half of the schools sampled at least 15% of the student population receives CALWORKS aid. District superintendents or assistant superintendents gave m e permission to survey teachers at these schools. In each case, the site principal held the right of refusal. In one case this right was 44 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. exercised. For the elem entary schools in my sample, I surveyed all teachers currently teaching grade 3 or grade 5 students. At high schools, th e teachers who received surveys were selected randomly from staff lists identifying teachers of courses in the “core” disciplines o f math, English, science and social science. At the m iddle school level, only m ath and English teachers were surveyed. T hese departments were selected based on the need for a performance measure. Student test results are commonly reported for the subjects taught by teachers in these “core” disciplines. T h e random selection process was simple. I counted the num ber of letters in the last name of the first teacher on the departmental list, selected the teacher at that position on the list, then selected every third teacher thereafter until my quota was filled. Eight teachers from each discipline were surveyed, giving me a sample of 32 teachers at each high school and 16 teachers at each middle school. In some cases, eight or fewer teachers were currently teaching in a particular subject area, so the sample reflects the entire department. I also surveyed the principal at each school. T h e survey protocol closely followed Dillman's “total design m ethod” (D illm an 1978). I contacted each potential respondent four times. T h e initial survey packet consisted of the survey, a postage-paid business reply envelope (BRM), a cover letter explaining the project and the importance of teacher participation, a district cover letter encouraging teachers to participate (in all districts except #2), and a one dollar incentive. Teachers received the packet at their school address. I sent a reminder postcard one week after the initial survey packet. This postcard thanked those who had already returned the survey, reminded those who had not yet returned it to do so, and asked those who wished not to participate to send in the blank survey so their nam e could be removed from the follow-up list. T hree weeks following th e post card, I sent a first follow-up survey. A second copy of the survey was enclosed along with a BRM envelope and a second cover letter encouraging their participation. T hree weeks following the first follow-up, I sent a second follow-up survey. T h e follow-up packets were identical except for the text of the cover letter and the dollar. T his protocol took ten weeks to complete, seven weeks identified above, plus time at the front end for mailing and time at the back end waiting for the last survey responses. 45 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. I received responses from 54% (636/1182) o f the teachers and 81% (52/64) o f che principals. This response rate com pares favorably with those obtained by other researchers in published studies of teachers and principals that have used mail surveys. Although some survey researchers recommend response rates o f greater than 70% before m eaningful conclusions can be drawn (Fowler 1993), recent research on survey responses has concluded thac no new information about the sample is gained once the response rate reaches 50%. For this data set, no evidence o f non-response bias was found. T his was determ ined by comparing early and late respondents. T his comparison resulted in no significant differences for model parameters. Tw o structural equation models were run, one with equality constraints placed on the paths connecting latent variables, the other with no constraints. T h e chi-squares for the two models were not significantly different, meaning that the equality constraint holds. Thus, the t\vo samples are statistically equivalent; early and late respondents do not differ. Survey Measures and D escriptive Statistics T he majority o f the survey instrum ent measures were existing scales with well-established psychometric properties. A notable exception is my construction of a new m easure of enabling rationale. This was necessary due to the lack of previous empirical work on this concept. In some cases, such as with trust, it was necessary to combine elements from several scales to capture the theoretically distinct meaning that I im pute to the concept within my model. It is im portant to note that the num ber of variables measured exceeds the number of variables in my m odel. T his is for two reasons. First, I did not want to be dependent on measures with unknow n psychometric properties. Although I followed the advice given by experts on scale developm ent and question writing (Fowler 1993), I could not be confident that these new scales would exhibit acceptable psychometric properties. T hus, whenever my model required the construction of a new measure, I included one or more additional scales that represent some portion of the construct that the new variable was designed to tap. By including these additional variables, I not only provided an alternate measure, but also a means by which to establish external validity for m y new measures. 46 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. T hese additional scales, not explicitly identified in the model, include formalization, supervisor procedural know ledge, task routineness, procedural justice, trust, organizational citizenship behavior, in-role performance, and role stress (includes role ambiguity, role conflict, and role overload). T h e scale reliabilities and descriptive statistics for these additional variables are provided in Appendix D. Below, I describe the m easures and their properties as grouped by th eir associated latent constructs. Reliabilities are reported primarily as a mechanism for assessing th e applicability of the scales used for this survey population. It is fruitful for future users of the scales included within this instrum ent to have the ability to com pare the data derived from teachers during this survey administration with the published reliabilities that accompany data derived from other samples (Thompson 1998; Vacha-Haase 1998). N ote that although reliabilities are reported for each scale, not all items in each scale were included as part of the latent variables used to test the hypothesized model. T h e construction of the latent variables and their associated factorial validity is discussed in a subsequent section. Although the laten t variable construction details are left to a su b seq u en t section, some information about the latent constructs is provided below. T h e variables used to contruct the latent variables are denoted with a double asterisk (**). Additionally, the right hand side of each table displays item correlations for each o f the m odel’s latent variables. T h e abbreviations for the latent variables read as follows: Morale (M OR); Control system tightness (CST); Enabling rationale (ER); T echnical know ledge (TK); and th e observed perform ance variable (P E R F — Stanford Achievement Test, ninth edition). Control System Tightness T h e tightness o f organizational controls was assessed using three scales. All items used a five- point Likert-type scale anchored by strongly disagree (1) to strongly agree (5). Results control was measured b y five items, adapted from Jaworski and M aclnnis (1989), that assessed the existence and use o f results-oriented or performance goals. An exam ple is “If my 47 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. performance goals are not met, I would be required to explain why.” H igher scores indicate greater results control. I averaged the scores from all five items into a single index; o f results control (a=.88). Action control was measured by four items, also adapted from Jaworski and M aclnnis (1989), that assessed the existence and use o f process-oriented or procedural concrols. An exam ple is “My immediate supervisor monitors the extent to which I follow established procedures.” Higher scores indicate greater action control. I averaged the scores from all four items into a single index of action control (a=.91). Personnel/cultural control was measured using a five-item scale. T h ese questions, adapted from Jaworski and M aclnnis (1989) professional control scale, assessed the degree to which control is exercised via interaction among colleagues. An example is “T h e departm ent encourages job-related discussions between teachers.” H igher scores indicate greater professional control. I averaged the scores from all five items into a single index of professional control (a=.90). Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Perf CONTROL SYSTEM TIGHTNESS 0.20 1.00 0.46 0.13 0.27 5a** Results control 1 3.29 1.23 1 5 0.16 0.68 0.32 0.16 0.15 5b** Results control 2 2.92 1.28 1 5 0.19 0.91 0.38 0.09 0.18 5c** Results control 3 2.73 1.33 1 5 0.16 0.81 0.28 0.10 0.13 5d** Results control 4 2.82 1.33 1 5 0.21 0.87 0.38 0.11 0.16 5e Results control 5 2.58 1.36 1 5 0.19 0.67 0.47 0.09 0.14 4a** Action control 1 2.89 1.27 1 5 0.10 0.83 0.32 0.07 0.10 4b** Action control 2 3.09 1.30 1 5 0.13 0.86 0.37 0.05 0.13 4c** Action control 3 2.49 1.28 1 5 0.10 0.82 0.35 0.02 0.10 4d** Action control 4 3.04 1.41 1 5 0.19 0.85 0.40 0.15 0.15 3a Professional Control 1 4.08 1.10 1 5 0.05 0.25 0.23 0.02 0.06 3b Professional Control 2 3.55 1.21 1 5 0.03 0.28 0.20 0.08 0.17 3c Professional Control 3 3.91 1.08 1 5 0.14 0.31 0.38 0.10 0.14 3d Professional Control 4 3.96 1.17 1 5 0.14 0.31 0.32 0.06 0.09 3e Professional Control 5 3.16 1.21 1 5 0.05 0.34 0.25 0.03 0.10 Table 3-1: Descriptives for Items Related to Control System Tightness ** Included as part o f latent variable Technical Knowledge T%vo types of knowledge were measured, knowledge relevant to the task processes or methods and knowledge o f task results or outcom es. T h ese items were com bined in a single scale of technical knowledge. All items used a five-point Likert-type scale anchored by strongly disagree (1) to strongly agree (5). 48 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Technical knowledge was assessed with four item s developed for this study. T h ese item s operationalize M erchant's (1998) theory regarding the conditions that determ ine the effectiveness o f controls. T h e four item s are; “As a teacher, I know what stu d en t learning results are m ost important”, “As a teacher, I exert a significant influence on w hether students attain expected learning results”, “T h ere exists a clearly defined body of knowledge that guides my work” , and “I understand what I m ust do to help scudents attain expected outcom es.” Scores on this m easure were averaged so that higher scores indicate greater performance knowledge (a=.70). Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Perf TECHNICAL KNOWLEDGE 0.46 0.13 0.25 1.00 0.28 10b** Technical knowledge 1 4.40 0.71 1 5 0.41 0.11 0.13 0.76 0.06 10f** Technical knowledge 2 4.24 0.77 1 5 0.29 0.10 0.13 0.74 0.01 10x** Technical knowledge 3 4.17 0.8 1 5 0.36 0.05 0.25 0.68 0.14 10c** Technical knowledge 4 4.22 0.88 1 5 0.27 0.09 0.17 0.73 0.10 Table 3-2: Descriptives for Items Related to Technical knowledge ** Included as part of latent variable Enabling rationale A single five-facet scale was used to assess enabling rationale. All items used a five-point Likert-type scale anchored by strongly disagree (1) to strongly agree (5). Enabling rationale was assessed with twelve item s designed to capture a general measure o f system trust, along with the four facets of enabling rationale discussed previously: correctability, flexibility, understanding, and validity (a=.85). Item s included “T his school’s policies serve th e best interests of the school.” 49 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST E R TK Perf ENABLING RATIONALE 0.27; 0.46; 1.00! 0.25 0.28 7e Enabling rationale 1-Correctability1 2.92 1.20 1 5 0.04 0.16! 0.46 -0.04 0.03 7j** Enabling rationale 1-CorrectabiIity2 3.15 1.41 1 5 0.17 0.25 0.73; 0.22 0.16 7a** Enabling rationale 1-Flexibility 1 3.68 1.17 1 5 0.19 0.35 0.58 0.15 0.24 7b** Enabling rationale 1-Flexibility2 3.59 1.23 1 5 0.14 0.16! 0.48 0.10 0.26 7c Enabling rationale 1-Trustl 3.24 1.00 1 5 0.03 0.20 0.14 0.08 0.01 7d** Enabling rationale 1-Trust2 3.52 1.05 1 5 0.23 0.35 0.71 0.13 0.16 7g** Enabling rationale 1-Trust3 3.31 1.13 1 5 0.23 0.38 0.76 0.13 0.13 71** Enabling rationale 1-Trust4 3.09 1.36 1 5 0.19 0.32 0.66 0.14 0.03 7h** Enabling rationale 1-Understanding1 3.62 1.26 1 5 0.14 0.10 0.64 0.05 0.16 7k** Enabling rationale 1-Understanding2 3.49 1.24 1 5 0.19 0.31 0.73 0.12 0.24 7f Enabling rationale 1-Validity1 2.79 1.16 1 5 0.22 0.52 0.60 0.21 0.03 7i** Enabling rationale 1-Validity2 3.38 1.13 1 5 0.18 0.33: 0.65 0.22 0.12 Table 3-3: Descriptives for Items Related to Enabling rationale ** Included as p art of latent variable Morale Four acticudinal measures were used to assess morale: job involvem ent, alienation, caring for students and expectations for student success. All items used a five-point L ikert-type scale anchored by strongly disagree (1) to strongly agree (5). Job Involvem ent was assessed with six items taken from Lodahl and Kejner (1965). Alienation was measured with the five item scale developed by M iller (1967). Each o f these scales is widely used in organizational research. For both scales, items were averaged so th at higher scores indicate higher levels of job involvement (a=.76) and alienation (cc=.80). Teacher Caring: Teachers frequently view their jobs in terms of who they serve, the students. A loss of concern for one's students is indicative of low morale. Caring was measured using six items from Talbert and M cLaughlin's (1994) caring sub-scale of their twelve item service ethic scale. An example is “I am certain I am making a difference in the lives of my students.” High Expectations: A sense of futility over one's primary task is also indicative of low morale. M any teachers believe that their job would lack meaning if they did not believe that they make a difference in th e lives of their students. High expectations were measured using six item s from T albert and M cLaughlin's service ethic scale. Scores were averaged so th at higher scores reflect higher levels o f caring (a=.67) and expectations (a=.64). 50 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Perf MORALE 1.00 0.20 0.27 0.46 0.30 10d** Alienation 1 1.66: 0.93 1 5 -0.69 -0.16 -0.22 -0.54 0.02 10n** Alienation 2 1.82! 0.99 1 5 -0.72 -0.15 -0.19 -0.42 -0.01 10p** Alienation 3 1.55' 0.97 1 5 -0.60 -0.13 -0.26 -0.21 -0.06 10u** Alienation 4 2.68: 1.13 1 5 -0.65 -0.10 -0.13 -0.18 0.14 lOw* Alienation 5 1.881 0.88 1 5 -0.54 -0.12 -0.20 -0.40 -0.05 10a** Job Involvement 1 3.51 1.16 1 5 0.67 0.13 0.21 0.23 -0.09 10h** Job Involvement 2 2.72: 1.25 1 5 0.51 0.08 0.06 0.18 -0.14 10j* Job Involvement 3 3.63 1.06 1 5 0.20 0.20 0.03 0.14 -0.02 10k* Job Involvement 4 4.13 0.83 1 5 0.41 0.02 0.03 0.26 0.00 10m** Job Involvement 5 2.85 1.07 1 5 0.50 0.09 0.07 0.12 -0.17 10q* Job Involvement 6 3.82 1.08 1 5 0.34 0.02 0.09 0.11 0.00 12a Caring 1 4.65 0.70 1 5 0.30 0.05 0.07 0.20 -0.01 12d* Caring 2 4.64 0.63 2 5 0.50 0.16 0.18 0.31 -0.01 12e** Caring 3 3.39 1.20 1 5 0.48 0.05 0.15 0.21 0.01 12f** Caring 4 3.91 1.11 1 5 0.38 0.11 0.19 0.18 -0.06 12h** Caring 5 4.14 0.86 1 5 0.47 0.17 0.21 0.32 0.08 121 Caring 6 4.11 1.05 1 5 0.19 0.14 0.09 0.04 -0.01 12b Expectations 1 3.37 1.26 1 5 0.30 0.13 0.17 0.14 0.05 12c** Expectations 2 3.49 1.36 1 5 0.44 0.15 0.20 0.19 0.11 12g** Expectations 3 3.91 1.02 1 5 0.51 0.04 0.17 0.25 0.13 121 Expectations 4 2.93 1.33 1 5 0.37 0.15 0.02 0.05 -0.04 12j Expectations 5 2.45 1.19 1 5 0.44 0.07 0.14 0.09 0.14 12k Expectations 6 4.03 1.12 1 5 0.23 -0.01 0.09 0.20 0.29 Table 3-4: Descriptives for Items Related to Morale */** Pool of items for latent variables, ** Included as part of latent variable Performance Student Performance was measured by a com posite score of the standardized tests across the core subject areas. W hile stu d en t perform ance is only a proxy for teacher perform ance, it is reasonable to expect that standardized student test scores, adjusted for socio-economic factors, reflect the quality of their teachers. T he specific tests chosen for this com posite m easure overlap with the subjects taught by teachers within the sample. For high schools, these subjects are math, English, science, and social science. For m iddle schools, the subjects are m ath and English. Elem entary performance scores were derived separately for each grade level (3 and 5) from the average of the Stanford 9 achievem ent tests in the areas o f math, reading and language. Scores are reported annually by the California Departm ent o f Education. For ease of comparison, I used norm- referenced mean scores. 51 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Perf SAT/9E (Student Scores) 50.1 17.2 20 87 0.10 0.24 0.26 0.16 1.00 Table 3-5: Performance: Stanford Achievement Test, ninth edition Opening Questions T o iniciate in terest in the survey I asked several questions pertaining to the issue of accountability and teachers’ perceptions of the problem s at their school that justify increased accountability. Although grouped together, these questions have no scale properties. For the accountability questions, the responses were scaled strongly agree (1) to strongly disagree (5). In general, teachers were neutral as to whether standardized tests provide a useful measure of student achievem ent (la, x=3.04) and w hether the school's accountability practices serve students' long term interests (le , x=3.03). Teachers were again neutral, but in a direction leaning toward agreement, that district administrators (lb, x=3.26), school administrators (lc, x=3.38) and teachers (Id, x=3.35) should be held accountable for student achievem ent. Q uestions regarding problems used percentage, rather than agreement, as the anchor. Respondents identified the percent of teachers experiencing the problem stated in the question (l= none, 2=25%, 3=50%, 4=75%, 5=100%). T h e mean response for five of the six questions indicates that teachers perceive nearly tw enty five percent of their colleagues as having problem s in the following areas: Awareness of school goals (2a, x=1.97), accomplishing school goals (2b, x=2.20), motivation to m eet school goals (2c, x=2.21), believing chat administrators take goals seriously (2d, x=2.12), and lacking resources necessary to m eet school goals (2f, x=2.34). Teachers perceive that about half o f their colleagues believe that “there are insurmountable external factors that make it impossible to attain schoohvide goals” (2e, x=3.24). 52 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK PERF OPENING QUESTIONS-ACCOUNTABILITY & PROBLEMS 1a Accountability 1 3.04 1.07 1 5 -0.07 0.07 0.13 0.13 0.10 1b Accountability 2 3.26: 1.17 1 5 -0.07 -0.06 -0.02 0.09 -0.13 1c Accountability 3 3.38! 1.12 1 5 -0.01 -0.02 -0.03 0.12 -0.08 1d Accountability 4 3.35 1.12 1 5 0.11 0.07 0.10 0.15 0.01 1e Accountability 5 3.03! 1.05 1 5 0.13 0.23 0.39 0.17 0.15 2a Problems 1 1.97 0.97 1 5 -0.11 -0.13 -0.22 -0.11 -0.23 2b Problems 2 2.20 0.96 1 5 -0.07 -0.20 -0.25 -0.07 -0.21 2c Problems 3 2.21 1.01 1 5 -0.04 -0.20 -0.33 -0.11 -0.21 2d Problems 4 2.12: 1.15 1 5 -0.11 -0.36 -0.40 -0.12 -0.19 2e Problems 5 3.24 1.16 1 5 -0.15 -0.14 -0.20 -0.10 -0.28 2f Problems 6 2.34 1.11 1 5 -0.05 -0.21 -0.25 -0.16 -0.22 Table 3-7: Descriptives for Opening Questions Finally, I collected individual demographic data from teachers and school-level stu d en t demographic data from the California Department of Education and Calworks. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Perf DEMOGRAPHICS and CONTROLS Demo First Career* 1.20 0.62 1 2 -0.04 -0.01 0.09 -0.17 -0.10 Demo Years in District 12.25 10.43 1 38 -0.12 -0.16 0.03 0.16 -0.01 Demo Union Membership* 1.06 0.40 1 2 -0.08 0.07 0.04 -0.06 0.02 Demo Years Teaching 14.45 10.81 1 38 -0.11 -0.16 0.01 0.19 0.04 Demo Race** ** H r* ★ * ** ** ★ * ** *★ * ★ Demo Gender (1=F, 2=M) 1.35 0.61 1 2 -0.20 -0.13 -0.02 -0.18 0.08 Demo Schtype (1=HS, 2=MS, 1.65 0.76 1 3 0.14 0.29 0.07 0.09 0.02 3=Elementary) Demo Professional 1.30 0.56 1 2 -0.20 -0.04 -0.06 -0.16 0.00 Association* Demo Professional Journal* 1.17 0.48 1 2 -0.20 -0.02 -0.01 -0.21 0.01 School Enrollment 1843 958 400 3660 -0.48 -0.39 -0.26 -0.22 -0.36 White Students 0.44 0.21 0.04 0.74 0.19 0.36 0.18 0.15 0.88 Calworks (AFDC) 0.16 0.13 0.00 0.48 -0.25 -0.01 -0.43 -0.44 -0.91 Free/Reduced Meals 0.41 0.27 0.05 0.92 -0.19 -0.09 -0.24 -0.28 -0.90 Table 3-8: Descriptives for Demographics and Controls (Covariates) *1=yes, 2=no; ** Teacher's race was coded across five categories, descriptives meaningless R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. L atent Variable Construction: Confirmatory Factor Analysis Structural equation m odeling is a two-step process, requiring two m odels. T h e first step uses confirmatory factor analysis to establish the factorial validity of the m easurem ent model. T h e second step, the structural model, represents the relations am ong th e latent constructs. T h e m easurement model uses confirmatory factor analysis to link a set of observed measures to a smaller set of unobserved, conceptual (latent) variables. T h e latent variable is defined by what the observed measures have in common. T h e m easurem ent model confirm s (or disconfirms) the hypothesized relationships betw een the latent variables and th e observed m easures. Unlike exploratory factor analysis, where the observed measures are allowed to load on each factor, in confirmatory factor analysis, each observed measure is allowed to load on only one factor. T h e m easurem ent m odel confirms that th e latent variables are properly specified. In a model that exhibits good fit w ith th e data, each observed measure will have th e majority of its variance explained by the hypothesized common factor. A separate CFA is run for each latent variable to determ ine the factors that best represent the latent variable. T h ese factors then becom e indicators for the latent variable in the CFA for the com bined endogenous variables, the C FA for the combined exogenous variables, and in the structural (LISR EL) m odel to follow. T his transformation— from item s loading on factors and subsequently factors becom ing indicators— is similar to the widely used practice of combining items to produce scales and then using the scale mean as the indicator for th e variable of interest. Formally, the relationship betw een the observed measures and th e latent factors can be expressed as X =A ^ + 8 PC .U where X is a vector of the observed measures; ^ is a vector of common factors; A is a matrix of factor loadings (or weights) relating the observed measures to the common factors; and 5 is a vector of the residuals (or unique factors) (Long 1983). It is at the m easurem ent model stage that modifications to the m easurem ent of latent variables are made. At this stage, only the relationships between the observed m easures and the latent 54 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. variables are o f interest. Any modification to the m easurem ent m odel will only change how the latent variable is m easured. By design, it remains unknown how these measurem ent changes will alter the relationships among the latent variables. Since the relationships among the latent variables are unknown, modifications to the iatent variables at this stage do not jeopardize the integrity of the causal hypotheses. Causal relationships are only observed w ithin the structural model. O nce the relationships am ong th e latent variables are known, changes m ade to th e latent variables are suspect. T h ese changes will likely reflect sample-specific findings th at improve the structural model fit rather than general observations about the population from which the sample was drawn. For each latent variable I include below information as to th e initial item pool, the initial set of hypothesized factors, and the modifications made to construct the latent variable to be tested in the structural model. F it indexes are reported in this section for both th e individual latent variables and for the set o f exogenous latent variables. Good model fit describes the situation where each observed m easure loads on its predicted latent variable and shares common variance with the other measures loading on the same latent variable. T h e exogenous m easurem ent model contains those variables that are not explained by any other variables within th e model. T h e endogenous m odel contains those variables that are “caused” by another latent variable. A separate CFA is not run for the endogenous m odel because one of the two exogenous variables, performance, is an observed measure. T h e CFA for endogenous model and the CFA for the latent variable morale are identical. For this analysis, data were aggregated at the department level (n=144). Item scores for each teacher within a given departm ent were averaged to derive the departm ent score. This level makes sense for two reasons. First, student perform ance data is reported at the school level for each subject. Since departm ents are typically configured around specific subjects, it follows that student performance on a particular test is also an outcome of a given dep artm en t’s ability to educate its students in th e su b ject m atter. Second, departm ents serve adm inistrative functions as well. Departm ent heads, especially at the secondary level, are quasi-administrators who potentially play a significant role in observing and reporting teacher performance. 55 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Control System Tightness (CST) T h e initial pool of items for control system tightness included the individual measures for the results control, action control, and professional control scales. T h e professional control scale items were dropped from the latent variable control system tightness due to an insignificant contribution to the common factor. T he squared multiple correlations for each of the professional control items were below 0.10. T his means that the latent variable “predicts” less than ten percent o f the variance for each professional control item . T h e standardized relationship between professional control and control system tightness was only 0.02. T his indicates the presence of an extrem ely small am ount o f shared variance becween professional control and the latent variable. O ne results control variable was also dropped. T his item designated recognition for performance as part o f the control system. It is likely that teachers are infrequently recognized for performance. Several fit indexes suggest an adequately fitting model. Following Byrne's (1998) recommendation, I report scores from three different indexes. Goodness of fit index (GFI) = .96. Comparative fit index (CFI) = .98. Root m ean square error o f approximation (RMSEA) = .078. G FI and C FI are measures of the relative am ount of variance/covariance explained in the sample data (CFI accounts for sample size). For G FI and C FI a fit above .90 is considered acceptable. For RMSEA, which reflects the model fit for the approximated population covariance matrix, a score below .05 indicates good fit, between .05 and .08 an acceptable fit, and scores betw een .08 and .10 indicate a marginal fit (MacCallum, Browne, and Sugawara 1996). T h e items and their factor loadings are reported in table 3-10. Technical knowledge (TK) All four item s were included in the final model of technical knowledge. Each item loads on the common factor technical knowledge. F it indexes indicate an adequate fit (GFI=1.0; CFI=1.0; RMSEA=.00). T h e items and their factor loadings are reported in table 3-10. 56 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Enabling rationale (ER) N ine items of the original pool o f twelve were included in th e final model o f enabling rationale. T h e three items dropped all had low inter-item correlations (r<0.50) as per the reliability analysis. T h e initial CFA dropped only the item with a negative inter-item correlation (7c. “T his school’s policies have a predictable effect on teacher practice” ). Fit indexes for this initial five-factor model were acceptable (GFI=.95; CFI=.96; RMSEA=.070), however modification indexes revealed that the fit could be improved substantially by dropping two additional item s (7e. “T h ere are established procedures for changing th e school policies that no longer meet the needs of students or teachers”, 7f. “T h e performance goals help me determine w hether I have been successful with my students”). Dropping these items required some changes to th e initially hypothesized factors since the correctability and validity factors were left with a single measure. An exam ination o f the remaining items suggested a three-factor solution. The second C FA confirmed the improvement in the measure. T hese items grouped around three factors: support, lack of coercion, and validity. Fit indexes indicate an improved fit (GFI=.97; CFI=.99; RMSEA=.047). T h e item s and their factor loadings are reported in table 3-10. Morale T h e initial pool of items for morale (denoted by a single or double asterisk on table 3-4) included the individual measures for the alienation and job involvement scales. In addition, the item pool included six items from the caring and high expectation scales that reflect teacher seif-efftcacy. Morale for teachers, as defined herein, is related to an individual's connection both to the job and to the students. Based on an examination of the inter-item correlations within each scale, several items were dropped prior to the first confirmatory factor analysis (CFA) run. T h e initial model hypothesized that items from the same scale (job involvement, alienation, caring and high expectations) would load on their respective factors.3 This was largely the case. After dropping a few items that failed to load significantly on their predicted factor, only one question remained that loaded on a factor other than the predicted factor. This made substantive sense because the item, initially on the alienation scale, refers to rewarding work, a them e common to the items in the job involvement scale. Further analysis determ ined that the caring and expectation questions could be combined for the sake of parsimony, w ithout any reduction in fit for the model. T h e resulting latent variable (morale) included three factors: pride, satisfaction and efficacy. Fit indexes suggest an adequately fitting model (GFI=.95; CFI=.95; RMSEA=.065). T h e items and their factor loadings are reported in table 3-9. 3 In the structural model, these factors become indicators of morale. 57 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item Factor W eight 10d. My work gives me a feeling o f pride in having done the job well Morale-Pride 0.79 lOn. 1 very much like the tyoe o f work th a t 1 am doinq Morale-Pride 0.78 lOp. 1 really don't feel a sense o f pride or accomplish ment as a result of the tvoe o f work th a t 1 do Morale-Pride 0.58 10u. My work is my most rewardinq experience Morale-Satisfaction 0.82 10a. The major satisfaction in my life comes from teachinq Morale-Satisfaction 0.83 10h. I live, ea t and breath teachinq Morale-Satisfaction 0.59 10m. The most important things th a t happen to me involve my work. Morale-Satisfaction 0.60 12e. If 1 try really hard, 1 can g et through to even the most difficult or unmotivated students Morale-Efficacy 0.60 12f. I feel th a t it's part of my responsibility to keep students from droppinq out o f school Morale-Efficacy 0.55 12h. I am certain I am making a difference in the lives of my students Morale-Efficacy 0.58 12c. My expectations about how much students should learn are not as hiqh as they used to be Morale-Efficacy 0.51 12g. There is really very little 1 can do to ensure that most of my students achieve a t a hiqh level Morale-Efficacy 0.51 Table 3-9: Factor loadings-Morale Performance T h e initial conceptualization of performance tested a m odel combining che standardized test measures with principal’s assessm ents of teacher in-role performance, teacher’s organizational citizenship behavior and/or teacher’s assessments of colleagues' organizational citizenship behavior. Contrary to expectations, the data did not fit this m odel. In lieu of this laten t m easure of performance, standardized test scores were substituted as an observed measure o f performance. Standardized test scores were chosen, rather than one of th e other measures of performance, for two reasons. First, data were available for all departm ents at all schools. T h e o th er m easures of performance were dependent upon survey responses, eith er from the principals or th e teachers. Second, standardized test scores, for b etter or worse, are the m ost commonly used m easure of school performance. A latent m easure of performance would have ideally captured those elem ents of performance common across the three measures collected. Since this did not happen it was both expedient and practical to use standardized test scores as th e sole measure of performance. 58 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Exogenous Variables Before proceeding to th e structural model, a final C FA was run incorporating all observed m easures of the three exogenous latent variables (control system tightness, technical knowledge, enabling rationale)"5 . M easures w ere allowed to load on th e factors identified for each latent variable. Thus, control system tightness has two indicators, technical knowledge has four (each item serves as an indicator), and enabling rationale has three. T his procedure confirmed th e fit of the exogenous model (GFI=.92; CFI=.96; RMSEA=.057). T o reiterate, the factors derived from the C FA for the individual latent variables were used as indicators for the exogenous CFA (the combined C FA for ail exogenous variables). T h e exogenous CFA confirmed the fit of the m easurem ent m odel to be used within the structural model. T h e exogenous CFA also established the proper factor loadings. T hese are needed to weight the com posite indicators used in the structural model. Ef the structural model were sm aller (i.e. had few er parameters), then individual items could be used as indicators for all latent variables. C om posite indicators were necessary because, if all item s were included, the sam ple would not have been large enough to accom m odate th e num ber o f estim ated param eters. Com posite indicators significantly reduce the num ber of estim ated parameters. Again, this practice is similar to com bining items to form a com posite scale w ithin other analytic strategies. A regression model could theoretically accom m odate single item s instead o f scales, how ever th e sam ple size requirem ents would increase drastically. Furthermore, w hen items are combined, especially Likert- type survey items, this increases the continuous quality o f the indicator necessary to m eet the assumptions of maximum likelihood estimation. 59 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item Factor Loading W eiqht 4a. My immediate supervisor monitors th e extent th a t I follow established procedures CST-Action control 0.85 4b. My immediate supervisor evaluates th e m ethods I use to teach students CST-Action control 0.94 4c. My immediate supervisor modifies my teaching m ethods when desired results are not obtained CST-Action control 0.77 4d. I receive feedback on the m ethods I use to attain performance qoals CST-Action control 0.89 5a. Specific performance qoals are established for my iob CST-Results control 0.64 5b. My immediate supervisor monitors th e extent to which I attain my performance qoals CST-Results control 0.97 5c. If my performance goals are n o t met, I would be required to explain why CST-Results control 0.77 5d. I receive feedback from my immediate supervisor concerning the extent to which I achieve my qoals CST-Results control 0.68 7J. I may appeal an administrative decision th at affects me personally w ithout fear of retribution Enabling Rationale - Support 0.74 71. This school's policies are applied the same way for all teachers Enabling Rationale- Support 0.66 7a. This school's policies support and encourage innovative teaching Enabling Rationale- Support 0.74 7h. Teachers a t this school are told w hat to do, not why they need to do it ER- Non-coercive 0.75 7k. Administrators at this school do not make clear th e connection between their mandates and th e educational implications for students ER - Non-coercive 0.82 7b. Established procedures prevent me from doing w hat I think best serves th e needs of students ER - Non-coercive 0.60 7g. This school's policies serve the best interests of th e school Enabling rationale - Valid 0.85 7d. This school's policies are consistent with our school's basic values Enabling rationale - Valid 0.78 7i. Performance goals are based on high standards of teaching practice Enabling rationale - Valid 0.60 10b. I understand what 1 must do to help students attain expected outcomes Technical knowledqe 0.76 10f. As a teacher, 1 know w hat student learning results are most im portant Technical knowledqe 0.68 10x. As a teacher, I exert a significant influence on w hether students attain expected learninq results Technical knowledqe 0.58 10c. There exists a clearly defined body of knowledge th at guides my work Technical knowledqe 0.59 Table 3-10: Factor loadings-Exogenous Variables Relationships among L atent Factors: The Structural Equation M odel T h e main purpose o f this statistical analysis is to establish the plausibility of a theoretical model and to estimate the effects of several control system characteristics on individual attitudes and behaviors. The model is properly understood as representing hypothesized causal relationships. Given that the data are nonexperim ental, it will be impossible to establish causality. In true 60 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Popperian fashion, the best [ can do is disconfirm th e relationship betw een th e data and the hypothesized relationships among the variables contained in the model. W ith these goals in mind, I chose latent variable structural equations m odeling as the best available m ethod to exam ine the com plex relationships contained within the m odel. T h e term structural equation modeling conveys two equally im portant ideas, that “the causal processes under study are represented by a series of structural (i.e. regression) equations, and that these structural relations can be modeled piccorially to enable a clearer conceptualization o f the theory under study” (Byrne 1998). SEM approaches, like traditional regression, are based on che general linear model. Latent- variable SEM approaches, however, are better suited to th e goals of my research and the limitations o f my data than is traditional regression. First, the m odel focuses on explanation rather than prediction. I w ant to understand as m uch as possible about the interrelationships am ong my predictor variables. I am interested in knowing how well my predictor variables explain morale and performance, b u t it is equally vital that I understand th e relative im portance o f each predictor. W hen predictors are correlated, as they are within m y m odel, the regression coefficient for a predictor variable is influenced both by correlation w ith the criterion variable and by its intercorrelation with other predictor variables. In regression, isolating the effects o f one variable from all other variables on a common criterion variable is accomplished by partialing out the effects o f all other variables. In SEM, the path coefficients for this model, representing th e strength o f the relationships betw een the predictors and criterion variables, are equivalent to partial regression coefficients. W hen the assum ptions o f regression are m et by the sam ple data, then the two approaches produce identical results. For most social science data, however, th e assumptions underlying regression are unrealistic. Some of the most im portant assumptions concern th e measures used to define the construct of interest. Regression approaches assume perfect reliability, that is, that the observed variables perfectly capture the underlying concept (no specification error) with no error in m easurem ent. For this model, it would be unrealistic to assume perfect reliability, especially for those scales that 61 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. assess attitudes. SEM overcomes these lim iting assumptions im plicit in regression approaches about the reliability of measures. SEM uses the logic of factor analysis in its inclusion of m ultiple indicators to reduce specification error and its use of reliability estim ates to reduce m easurem ent error4. Although it is common within regression approaches to use m ultiple indicators of a single or overlapping construct, this often creates problems of m ulti-colinearity that destabilize regression coefficients (bouncing betas). L atent-variable SEM approaches, by com bining the logics of regression analysis and factor analysis, elim inate many of the problems o f multi-collinearity within constructs, since th e set of collinear predictors is replaced by a single com posite predictor (Maruyama, 1998: 75). Latent-variable SEM approaches also provide better information for variance decompostion. T h e ability to com bine measures designed to assess the same underlying theoretical construct helps researchers separate three types o f variance: T rue score com m on variance, tru e score unique variance, and error variance. T h e first two types of variance are not error. T ru e score common variance is attributable to the theoretical construct of interest. T ru e score unique variance is attributable to som ething beyond the theoretical construct o f interest, perhaps m ethod variance or even a second construct. T h e sum o f these two types o f variance equals the reliability of the measure (a). Error variance is what remains (1-a). Partitioning the two types o f true score variance requires m ultiple indicators. Latent-variable SEM approaches isolate the variance related to the construct of interest (true score common variance) by using principles consistent with the logic of factor analysis (Maruyama 1998). T h e factor is defined by the shared variance which eliminates a substantial portion o f the true score unique variance. By utilizing m ulti-trait, m ulti-m ethod approaches (M TM M ), the researcher can eliminate even the com m on m ethods portion of this shared variance. Although this represents the ideal, M TM M is not possible with my data, and therefore the variance attributable to a common method will remain. 4 M easurem ent error is calculated w ithin the SEM program unless the latent variable is indicated by a single, observed variable. In this case, the researcher provides the estim ated m easurem ent error. 62 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Enabling Rationale Morale Control System Tightness Performance Technical knowledge Figure 3-1: The Impact of Tight Control- A Structural Model M odel Specification and Evaluation I estim ated the hypothesized structural model (see Figure 3-1) using Joreskog and Sorbom's (1994) L ISR E L structural equations program. T h e program analyzed an 18x18 variance/covariance matrix representing the latent, observed and covariate (control) variables. For ease of depiction, the model does not include the paths between the latent variables and the five covariates (years of teaching experience, percentage of white students at the school, percentage of students receiving free or reduced price meals, percentage of students enrolled in CALWORKS, and school enrollm ent). T h e model as hypothesized would necessarily include the moderating influence of enabling rationale and technical knowledge (represented by gray arrows). Due to programming difficulties, the moderator hypotheses were not tested using L ISR E L . A second analysis using two- stage least squares regression will test the moderator hypotheses. S tudent perform ance and the five covariates were m easured using objective demographic inform ation. For this reason I assum ed that th ese m easures contained very little random m easurem ent error. M easurement error for these observed variables (including the covariates) was estimated at 0.01. T h e factor loading for each of these single-item constructs was set to 1.0. 63 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. I tested the eight substantive relationships (see Figure 3-1) w hile controlling for the potential confounding influence of the five dem ographic covariates. I controlled for the effects of the covariates by including them as exogenous variables in the m odel and allowing these covariates to correlate among themselves and to predict the five substantive latent constructs (Markel and Frone 1998). I report th e overall model fit with th e same fit indexes used to evaluate the fit of the confirmatory factor analyses. T h e goodness of fit index (G FI) and comparative fit index (CFI; Bentler 1990) are measures of the relative am ount of variance/covariance explained in th e sample data compared to the baseline null model. A score between .90 and 1.0 represents an acceptable fit. T h e root mean square error of approximation (RMSEA; Steiger 1990) reflects the model fit for the approximated population covariance matrix. A RMSEA score below .05 indicates good fit, between .05 and .08 an acceptable fit, and scores between .08 and .10 indicate a marginal fit (MacCallum, Browne and Sugawara 1996). Moderated Relationships among Latent Factors: Regression Approaches to Interaction Several of my hypotheses make predictions about the effects o f control system tightness that are contingent on the levels of technical knowledge and enabling rationale. Stated differently, both technical knowledge and enabling rationale are theorized as moderators of the relationship between control system tightness and morale and performance. T hese m oderated relationships are examined using two-stage least squares regression (2SLS). 2SLS is a m ethod for estimating product interactions in structural equation models (Bollen 1995). 2SLS accounts for measurement error and is robust to violations of multivariate normality (Bollen and Paxton 1998; Li and Harmer 1998). 2SLS does not account for full information. T h is means that the m ethod does not make use of all the data, does not estim ate all the param eters of the model, and cannot give a reliable estimate of the overall fit of th e model (Joreskog 1998). E ven with these limitations, 2SLS is considered the most reasonable approach to use when the variables involved in the interaction are latent (Joreskog 1998). 64 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. M odel Specifications Using 2SLS T h e 2SLS m ethod first substitutes one observed variable (called a scaling indicator) for each latent variable in the regression equation and then uses instrum ental variables (IVs) to estim ate the parameters (Bollen and Paxton 1998). T h e goal is to elim inate any correlation between the scaling indicator and th e error term (the disturbance, in structural equation m odel term inology). Instrum ental variables are useful, when well-chosen, because they correlate with the scaling indicator, but not with the error term. In the first stage the instrum ental variables are regressed on the scaling indicator. T h e resulting coefficients are then used to create a predicted value of the scaling indicator. Because the predicted value is based on observed characteristics, it is uncorrelated with the disturbance (Foster and McLanahan 1996). T h e predicted values of the scaling indicators are then used to estim ate the outcome variable of interest. Reference to the standard regression model will help clarify this process. In this equation, latent variables (L) fill the space usually reserved for observed variables within the standard regression m odel. Assume a model with two main effects and one interaction term. T h e resulting equation is Y = a + b iL [+ b 2 L ,2 + b 3 L iL 2 + e . [X .2 ] Assume each of the two latents has two indicators. T h e scaling indicators for the latent terms are represented by xi and xj. xt = Lt+Si. [X.3a] x2 = a 2+^.2|L|+52. [X.3b] X 3 = L - 2 + S 3 . [X.3c] X4 — GC4-hX4iL#2“ ^ 4 * [X.3d] Solving the equation for each scaling indicator, we get L (= Xi-5i. [X.4a] L2= X 3 - 5 3 . [X.4b] T hen substitute the scaling indicator for each term on the right-hand side of the standard regression equation. T h e resulting equation is similar to the standard regression equation, except for the inclusion o f the error terms: 65 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. yi=0Cyi+Pii(xI-St)+Pi2(X3-83)+Pi3(X|-5i)(X3-S3)+Cl- [X.5] Solve chis equation and we define che composite disturbance (ui) as u^-PuSi-PuSs-PnfxiSs+X jS^iSs)-^,. [X-6] T h e final equation Vl = O C y I+P1 l+p 1 2 ^ 3 +P 13^ L ^ 3+U 1 [X.7] is alm ost identical to the standard regression equation with one im portant caveat. T h e right-hand side variables are correlated with the disturbance term. T h is violates a critical assum ption of ordinary least squares (OLS) regression. 2SLS is designed to overcome this violation. In th e first step o f this two step process, the IVs are regressed on each of the right-hand side variables. T h e IVs are correlated with the exogenous variables but are uncorrelated with the disturbance term . In the second step, the predicted values are substituted for the original right- hand side variables. T h e most difficult aspect of the procedure is the selection o f IVs. T h e 2SLS estimator requires at least one IV for each latent variable and latent variable interaction included in the equation. (In fact, one IV is needed for each right-hand side variable, but since observed measures can be used as IVs, one need only worry about constructing IVs for the laten t term s on the right-hand side) Furthermore, these IVs m ust not be correlated with the disturbance (residual) term. Bollen (1998) offers several “rules” for selecting instrum ental variables for an equation w ith a product interaction of latent variables. T hese rules are broken down into rules for inclusion and rules for exclusion. IVs should include all exogenous observed variables and ail nonscaling3 indicators of exogenous latent variables. In order to m eet the condition that the IVs ac least equal the num ber o f latent variables (including interactions), IVs may need to be created by m ultiplying observed indicators. 3 A latent variable requires that an indicator be identified as “scaling” in order to establish the m easurem ent scale o f the variable. Typically, the indicator w ith the highest factor loading is selected as the scaling indicator. 66 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. IVs should noc include scaling indicators as these are correlated w ith th e general disturbance term . Neither should the IVs include any variables directly or indirectly affected by the dependent variable of the second stage equation (the one that includes the interaction term). 67 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Enabling Rationale CSTxERxTK Control System Tightness Morale Technical Knowledge Figure 3-2: The Impact of Tight Control- A Structural Model with Three-way Interaction Term Following Bollen (1995), the saw-tooth lines indicate interactive relationships. T h e interaction model (see figure 3-2) includes three latent main effect terms (control system tightness, enabling rationale, and technical knowledge) and one three-way interaction term (CSTxERxTK). Covariates, which are not represented in figure 3-2, are included within the model as observed exogenous variables. T h e covariates are also utilized as IVs. I chose as scaling indicators those observed variables with the highest factor loading on the latent construct. Non-scaling indicators were used as IVs. T hree additional IVs were constructed from product terms of the non scaling latent indicators. Each of the non-scaling indicators was used to construct the interaction term IVs. To evaluate the quality of the IVs, three tests were em ployed. First, a simple counting rule determined that the num ber of IVs exceeded the num ber of right-hand side variables. Second, each squared multiple correlation coefficient (R2) from stage one of the two-stage process was examined. An R2 value below .10 suggests that the IVs may reduce the quality of the estimates. Third, when 68 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. the equation includes more IVs than right-hand variables, as is the case w ith m y equation, then the m odel is over-identified and tests may be run to determ ine w hether the IVs are uncorrelated with the disturbances. Bollen suggests the Basmann (1960) test o f over-identification (Bollen and Paxton 1998). Summary Survey data were collected from 636 teachers (54% response rate) and 52 principles (81%). Confirmatory factor analysis was used to construct the latent variables. For this sample, all latent variables exhibited satisfactory psychometric properties. Structural equation m odeling was used to exam ine the eight main effects as presented in figure 3-1. Two-stage least squares regression was used to examine the hypothesized interaction effects (figure 3-2). Results are reported below. 69 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 4: Results T h e bias against tighc control systems appears unwarranted. Recall that despite equivocal empirical results (Rowan 1990), th e preponderance o f the literature anticipates negative attitudinal and behavioral outcomes for tight control systems (M cNeil 1986). Contingency theorists have dismissed unequivocally harsh judgm ents of control systems by way of the grand “it depends.” For contingency theorists, attitudinal and behavioral responses to the control system depend on the system ’s ability to overcome inform ation deficiencies. According to these theorists, technical knowledge is the critical contingency variable. My research has examined an additional contingency variable, enabling rationale. T h e hypotheses set forth in this study are based on the proposition that control systems differ in terms o f their technical and social characteristics (i.e. enabling rarionale) and that teachers’ reactions to such systems differ accordingly. T here is strong, but qualified, support for the extended contingency theory. Structural Model: Testing the Direct Effects T h e LISR EL models (fig. 4-1, fig. 4-2) tested the direct effects of technical knowledge and enabling rationale. T h e results provide mixed support for the hypothesized model. N one of the hypothesized paths to performance were significant (standardized path coefficients, (3 or y, are shown in Figure 4-1). Both the paths from technical knowledge and enabling rationale to morale were significant (p < .05) and in a positive direction. Morale was positively related to both the technical characteristics (technical knowledge; P=.83; z=4.15) and social characteristics (enabling rationale; P=.45; z=3.13) of control systems. Control system tightness exerts no direct effect on morale. Overall M odel Fit M odel fit is adequate to test the stated hypotheses. T h e overall chi-square for the structural model (n=144) was 130.53 with 95 degrees of freedom (p<0.009). T h e fit indexes for the structural 70 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. m odel indicate an adequate fit o f the model to the data (GFI=0.91; CFI=0.97; RMSEA=0.046; 90 percent confidence interval for RM SEA = (0.018 ; 0.068)). T h e confidence interval for RMSEA includes the .05 standard for good fit and the upper boundary exceeds this standard. Given a confidence interval that includes the standard for a good fitting model, it is not possible to reject a null hypothesis of not-good fit (MacCallum, Browne, and Sugawara 1996). T h e probability that RM SEA represents a good fit is 0.67. Therefore, the model represents an adequate, though not optimal, fit to the data. Enabling Rationale Morale Control System Tightness 1.04 Performance Technical knowledge Figure 4-1: Parameter Estimates: The Impact of Tight Control (Covariates modeled, but not pictured) p < .0 5 * p < .0 1 * ’ p<_001’ ** note: Model 2 <P J resuits shown in gray, all other paths are non-significant Alternate Specifications Although not central to the theory being tested, the attitude-behavior debate does play a part in the depiction of this model. Research has not satisfactorily resolved w hether morale is more likely to influence performance or visa-versa (Scott 1992). Thus, the direction of the arrow could run from morale to performance (pI2 ), from performance to morale (P2 1 ), or both directions simultaneously. W ith this in mind, three models were run, one testing morale as a predictor of performance (P,2), 71 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. one testing performance as a predictor of morale (P2I), and one testing the reciprocal relationship (PI2 and p2t). D ue to problems of identification this non-recursive model was specified differently and will be discussed separately. Across models, th e causal direction of the performance-morale relationship did not influence th e strength or direction of the other path estim ates. T he two recursive models (see figure 4-1, above) provided nearly identical estimates. T h e param eter estimates from m odel one (performance influences morale) are shown in black and the significant param eter estim ates for m odel two (morale influences perform ance) are shown in gray. In both models, th e perform ance-m orale relationship was non-significant. T he third model allowed a reciprocal relation between the two variables. Model three is shown belovv (fig. 4-2). T his m odel tested only the significant paths from the earlier models. Eliminating the non-significant paths reduced the num ber of parameters estim ated by the model. T h is was necessary for che model to be identified, m eaning that a unique solution for the values o f the parameters could be found. Since the non-recursive model increases the num ber of parameters to be estimated, dropping the non-significant param eters was necessary to reduce the num ber of parameters in order to test the reciprocal relationship between morale and performance. T h e path coefficients and overall model fit (X2=132.64, d f 99; GFI=0.91; CFI=0.97; RMSEA=0.043, (c.i. 0.012; 0.065) p<.05=0.67) were not significantly different from the other two models. T h e strength o f the relationships for the covariates, pictured below, did not deviate across models. 72 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Enabling Rationale P 3 5 = 0 .6 4 " " Control System Tightness Morale Technical Knowledge Performance E nrollm ent ^ 2% W hite ^3 Teacher Experience C alw orks % Free M eals Figure 4-2: Control System Impact on Performance and Morale: Significant param eter estimates. Including Covariates p<.05* pc.01** p<.001*** Alternate M odels As m entioned in che m ethods section, the initial conceptualization o f the model included a latenc performance variable instead of the single standardized measure o f student performance used in the models above. T h re e alternate m odels w ere run that substituted a m easure of extra-role performance, organizational citizenship behavior, for the standardized stu d en t perform ance measure. T h e first alternate model used che principal’s assessment o f che departm ent m em bers’ collective organizational citizenship behavior. T h e second used an aggregate of che individual self- reported organizational citizenship behavior. T h e third model used teacher perceptions of their departmental colleagues’ organizational citizenship behavior as che outcom e variable. None of chese models show ed a significant relationship betw een control system tightness and extra-role performance. Two-Stage Least Squares Structural Model: Testing the Interaction Effects T he two-stage least squares regression (2SLS) model tested th e m oderating influence of technical knowledge and enabling rationale on the control system cightness-m orale and control system tightness-perform ance relationship. T h e 2SLS analysis follow ed a sim ple four-step hierarchical process. First, the outcome variable is explained as a function of the covariates. Second, 73 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. variables hypothesized to directly influence the outcome variable are entered into the equation. Third, two product terms, representing the moderated independent variables, are regressed on the dependent variable. Fourth, the model is expanded to include a three-way interaction term. T he interaction effect is significant if both the change in R2 betw een the direct and interaction models and the T value of the interaction term(s) are significant. T h e F test for significant AR2 is yielded by the following equation: F = (R2, - R2,)/(K2 - K,) - (1 - R22)/(N-K2 -1) [Y.l] where K represents the num ber of predictors, N is the sample size, and the subscripts 1 and 2 refer to the original and expanded equations (Jaccard, Turrisi, and Wan 1990). Table 4-1 Summary of Two-Stage Least Squares Regression Analysis for Variables Predicting the Impact of Control System Tightness on Student Performance Predictor and Outcome B SE P P AR2 R 2 Student Performance (STAR) Step 1: Correlates White (%) Teacher Experience Free/Reduced Lunch Enrollment CALWORKS (AFDC) Step 2: Main Effects Control System Tightness Enabling Rationale Technical knowledge Step 3: Interaction Effects CSTXER CSTXTK Constant 74 41.580 4.916 0.169 0.083 -20.856 5.264 -2.068 0.787 -29.912 9.041 0.787 1.270 0.763 1.699 2.152 2.501 0.233 0.355 0.061 0.279 32.958 13.295 0.520 0 .0 0 0 0 .072 0 .0 4 4 -0.334 0.001 -0.116 0 .010 -0 .222 0.001 0.036 0 .537 0 .027 0 .6 5 4 0 .0 6 4 0 .3 9 2 0.026 0.513 0.009 0.827 0 .0 1 4 .869 -.005 .864 .001 .864 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4-2 Summary of Two-Stage Least Squares Regression Analysis for Variables Predicting the Impact of Control System Tightness on Morale Predictor and Outcome B SE P P AR2 R 2 eacher Morale (MOR) Step 1: Correlates .045 White (%) -1.314 1.740 -0.133 0.452 Teacher Experience -0.032 0.028 -0.111 0.250 Free/Reduced Lunch 0.565 1.793 0.073 0.753 Enrollment 0.237 0.271 0.107 0.384 CALWORKS (AFDC) 0.403 3.075 0.024 0.896 Step 2: Main Effects .136 .181 Control System Tightness 0.505 0.446 0.190 0.259 Enabling Rationale 0.981 0.576 0.276 0.091 Technical knowledge 2.679 0.851 0.642 0.002 Step 3: Interaction Effects -.007 .174 CSTXER* -0.927 0.639 -0.228 0.150 CSTXTK* 0.190 1.998 0.027 0.925 Step 4: Three-way Effects .019** .200 CSTxERxTK 1.983 0.902 0.388 0.030 Constant 4.627 4.507 0.301 'n o te : The two-way interactions w ere dropped in step four, two-way interaction results taken from step three ** The F test is statistically significant (3.16, 8,9) In the 2SLS model, as with the L ISR EL model, the covariates explain nearly all of the variance in student perform ance (R'=0.869). For the student perform ance model, the addition of main and interaction effects does not increase the variance explained. N one of the hypothesized main or interaction effects are significant at the .05 level. T h e 2SLS analysis provides strong support for this study’s main hypothesis: T he impact of control systems on teacher attitudes is contingent on the control system ’s characteristics. T h e covariates explain only a small portion of the variance in morale. W hen the main and interaction effects are entered into the model, the signs of the hypothesized relationships are in the predicted direction. Interestingly, and consistent with contingency theory, control system tightness is not directly associated w ith morale. Control system tightness, however, does have a moderated effect on morale. T h e 2SLS m odel supports the hypothesis that the social and technical characteristics of the control system influence a teacher’s attitudinal response to a control system. T he three-way interaction term (C STxER xTK ) was positive and significant (P=.39; p=0.030). T h e 2SLS analysis also corroborates the findings of the earlier structural model. Morale is influenced by both technical knowledge (P=.64; p=0.020) and enabling control (P=.28; p=0.091). T h ese results, while consistent 75 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. w ith che structural equation model results, differ slightly due to the addition o f the interaction terms. Namely, the betas drop for boch technical knowledge and enabling control, the latter falling below the level of statistical significance. An F cest of the r-square difference between steps two and four in the 2SLS model (table 4-2) confirm ed the presence o f the interaction effect. It has been argued th a t product terms, by them selves, do not represent statistical interactions. T his is because the product term is often correlated with its com ponents. For interaction analysis it is necessary to partition the variance to separate the main and interaction effects. This is accom plished through a hierarchical two-stage least squares regression. T h e hierarchical procedure entails entering the covariates on step one, main effects on step two, two-way interaction effects on step three, and three-w ay interactions on step four. T h e resulting changes in R2 are then exam ined for significance using the F statistic (Jaccard, Turrisi and Wan 1990). T h e present study yielded a statistically significant F statistic (F=3.16, 8,9). A significant F indicates the presence o f an interaction effect (see formula Y.l). T h e main and the significant 3-way interaction effect from the 2SLS analysis are depicted in Fig. 4-3. Enabling R ationale CSTxERxTK ,,=0.39" C ontrol S ystem T ig h tn ess M orale .=0.64*** Technical know ledge Figure 4-3: The Impact of Tight Control- A Structural Model with Three-way Interaction Term *p<0.10; **p<0.05; ***p<0.01 76 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. T h e strength of the three-w ay interaction effect is derived from the AR2 statistic (0.019), m eaning that the three-way interaction term increases th e am ount of explained variance by nearly 2%. Alternatively, the three-way interaction term increases the explanatory power of the equation by over 10%. T h e nature of the interaction effect can be estim ated by using the P values from the 2SLS analysis (table 4-2) to create a regression equacion, then substituting high (+1 standard deviation), m edium (m ean) and low values (-1 s.d.) for enabling control and technical knowledge, then examining the effect on morale across the range of control system tightness. As the chart below illustrates (fig. 4-4), the impact o f control system tightness on morale is not constant, b ut varies as a function of enabling control and technical knowledge. At low levels of CST, the level of morale for any given combination of technical knowledge and enabling control differs only slightly. For ease o f exposition, figure 4-4 reflects differences am ong departm ents adjusted so that the highest possible morale score would b e a 5.0 and the lowest possible score a 1.0, just like a Likert scale. While the differences reflect small numbers, the meaning of a 1.0 difference reflects the difference between any two consecutive anchor points on the scale. T h e entire range of morale at low control system tightness is equivalent to about one quarter of a point on a five-point Likert-type scale. At high levels of CST, differences in control system characteristics have dramatic effects on morale. Here, the range for morale spans an entire point on the five-point Likert-type scale. As expected, the most dramatic differences exist betw een control systems characterized by high technical knowledge/high enabling control and low technical knowledge/low enabling control. W hen T K /E R is high, control system tightness has a positive im pact on morale. A one-unit increase in C S T results in a three-quarter point increase in morale (P=.742). When T K /E R is low, control system tightness has a negative im pact on control. A one-unit increase in CST results in a third of a point decrease in morale (p=\36). 77 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0:875 HiEC HiTK HiEC MidTK HiEC LoTK MidECHiTK MidEC MidTK MidECLoTK LoECHiTK LoEC MidTK LoECLoTK 0.625 0.375 0.125 Figure 4-4: Effect of CST on Morale at Various Levels of Enabling Rationale (EC) and Technical Knowledge (TK ) T h e interpretation is straightforward: When the control system is loose, its other characteristics are inconsequential. A loose control system does nothing to alter one’s current perceptions about the job. Conversely, a tight control system produces an attitudinal response. W hen the control system is technically sound and those administering it are deem ed helpful and reliable, the attitudinal response is positive. Alternatively, a control system seen by teachers as inadequate provokes negative attitudinal responses. Strong Support for a Partial T est o f the Extended Contingency Theory Teachers’ attitudes toward work are influenced strongly by the control system. Tight control does not always evoke negative attitudinal and behavioral responses. T hese results are consistent with expectations from a contingency theory perspective. Moreover, these results suggest that the addition of enabling rationale increases the explanatory power of the contingency theory of control system design. 78 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. The Importance o f Technical and Social Considerations T h e technical and social conditions hypothesized as necessary for effective control had direcc and moderating influences on teacher attitudes. T h ese results were consistent across m ethods. W ithin the LISREL model (fig. 4-1) the direct relationship between morale and each conringency variable was statistically significant—enabling rationale (|3=.45; z=3.13) and technical knowledge (P=.83; z=4.15). Furtherm ore, no main effects of control system tightness on morale were found. T his is consistent with expectations that one cannot know the effect o f control system tightness on morale without specifying th e type o f control and determining w hether th e requisite technical knowledge is present. Within the two-stage least squares “moderated-effeccs” model, (fig. 4-3) the direct relationship betw een technical knowledge and morale was statistically significant (P=.64; p=0.020). T h e strength o f the relationship betw een enabling rationale and morale was statistically significant only if the standards of significance are relaxed to p < .10 (P=.28; p=0.091). The strength o f these direct effects are consistent with che L IS R E L model. In che 2SLS model, the relationship betw een control system tightness and morale was non-significant (P=. 19; p=0.259). T his is an im portant aspect o f the contingency argument. If che effect of control system tightness on morale is conditional upon one or m ore other variables, control system tightness should explain little or none o f che variance in morale. Of course, this is only meaningful in the presence of a statistically significant m oderated effecc. Two Moderators are B etter than One T his study’s most im portant finding was the presence of statistically significant moderators of che control system cightness-morale relationship. T h e type of control system m atters. W hen high technical knowledge and high enabling rationale characterize che control system , control system cightness positively impaccs teachers’ morale. Conversely, when the control system lacks these characteristics, control system tightness negatively impacts teacher morale. It is interesting to note chat the presence (or absence) of both characteristics was necessary for control system tightness to 79 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. positively (or negatively) im pact morale. N eith er characteristic alone was sufficient to account for the variation in the control system tightness to morale relationship. T h is finding highlights the value of extending the contingency theory o f control system design to include variables that address the quality of social relations throughout the control system design and im plem entation processes. The Meaning o f a Partial Test Absent from above was any mention o f the effects of control system tightness on performance. T his is due to the fact that the lion’s share o f the performance variance betw een departments was explained by the covariates (control variables) (R2=0.87). The individual covariates included in the m odel were strongly related to the standardized achievem ent test used as th e performance measure. Five covariates were measured, three school-level stu d en t background characteristics, school size and teacher experience. E xcept for teacher experience (P=0.07; p=0.044), all other covariates exhibited strong relationships w ith the performance variable— percenc o f white students at the school ((3 =0.52; p=0.000); percent o f students at the school receiving free or reduced price meals (P = -0.33; p=0.001); percent of stu d en t families at the school receiving financial assistance from the state (P = -0.22; p=0.001); and school size (P = -0.12; p=0.010).6 While the implications o f this are staggering, similar covariates have explained similarly large portions of the variance in previous studies. Several studies have found socioeconomic status to account for greater than half the variation in educational attainm ent betw een schools (Bryk and Raudenbush 1992; Rowan, Raudenbush, and Kang 1991; Willms 1992). Som e comfort (for those of a less deterministic bent) may be found when comparing student background characteristics within the school. T hese studies report that the sam e background characteristics explain only about 7% of the performance variation w ithin schools (Bryk and Raudenbush 1992). U nfortunately, the data requirements for this type o f within school analyses could not be m et w ith the present data set. 6 These coefficients are taken from model one (see figure 4-1). Note th a t these relationship differ only slightly from those reported in model th ree (see figure 4-2). Consistent across models, all covariates except for teacher experience exhibit non-significant relationships with morale (fig. 4-2). 80 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thus, the cest of th e extended contingency theory model can be labeled a partial test. Better data are required to test th e performance implications of control system design w ithin schools. 81 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Chapter 5: Discussion “The arbitrary command ignores one of the mostfundamentalfacts o f human nature, namely, the wish to govern one's own life. ’’-M a ry Parker FoIIetc This study makes an em pirical contribution to a growing body o f theory that calls for a reexamination o f the predom inantly pejorative view o f control and related constructs. Strong empirical support was found for the central theoretical tenet underlying this study— that responses to control will vary according to the technical and social characteristics of the control system. Real Schools, Real Issues One goal of this study is to provide a framework for examining control systems in schools. This framework includes a language for categorizing forms of control (personnel/cultural, actions, and results), some criteria for when control should and should not be attem pted (technical knowledge), and some guidance for the design and implementation of control systems (enabling rationale). T h e language of theory and statistical tests, however, often is quite distant from the language of schools. At times in this exposition, schools and the educational context recede far into the background. Further, and closer, study is needed to examine these processes in action, where the real work of schooling occurs. The three composite vignettes that follow do not replace the need for in-depth fieldwork, but hopefully might illustrate the usefulness of the control system design theory articulated here and thereby motivate further research. Le Conte Secondary School Teachers at Le C onte Secondary School shrug indifferently w hen asked to com m ent on evaluation. “I’ve been here twenty-three years and have been evaluated only six or seven times. No matter, it’s more of a nuisance, really.” T h e overwhelming majority of Le Conte’s teachers place little or no value on formal classroom observations. When evaluation or observation is m entioned, 82 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. teachers are apt to assum e the reference is to something being done to them by administrators to m eet administrative needs, rather than those of teachers or students. T he tone becomes noticeably more rancorous w hen the subject changes to student evaluation. (Student evaluation is often used as a proxy for school performance and, implicitly or explicitly, teacher performance.) “T h o se tests tell us nothing... T h e newspaper acts like that’ s the only m easure of how good a school is doing... Kids don’t care about [standardized tests]. T h ey fill in the bubbles as quickly as they can, then spend the rest of the tim e period doodling on the test booklet.” D espite the universal disregard for standardized tests, th e school faculty has not im plem ented any alternative measure o f student performance. Staff com m ents regarding professional developm ent (routinely called “in-service” or “pupil-free” days) indicated considerable dissatisfaction along with the same powerlessness to alter the status quo. District-administered professional developm ent days occur four times p er year with teachers assigned to specific sections, sometimes by grade level and other times by subj’ect m atter taught. One additional day is allocated to schools, usually prior to the students’ first day of school in September. Le C onte uses this day for m eetings (whole faculty and departm ent) and classroom preparation. Exposition High School Exposition H igh School offers its teachers three m ethods of evaluation: 1) A traditional adm inistrative evaluation, with a pre-observation m eeting to decide th e focus and a post observation debrief; 2) P eer review utilizes a trained “peer-coach” instead of an adm inistrator and doubles the num ber o f observations each year from three to six; 3) Evaluation by portfolio, developed in conjunction with participation in a self-study group, which is th en publicly presented/defended as evidence of teacher growth. Teachers appear to enjoy the options, but a hint o f cynicism accompanied their descriptions of options other than the one they chose for themselves. Those choosing adm inistrative evaluations questioned the level of accountability in peer and self- study groups. T hose opting for peer-evaluation noted the benefits of having more observation done by someone with both subject-m atter and pedagogical expertise. “It goes much deeper than filling 83 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. out a standard form,” one teacher rem arked. Teachers com pleting portfolios frequently used superlatives to describe their experience. “[Putting together the portfolio] has been the m ost valuable learning I have done in m y entire career.” T h ese teachers suggested that the private and sem i-private nature o f the other evaluations robbed those teachers o f the honest self-reflection provided by a public demonstration o f growth. Student evaluation at Exposition High is a combination o f standardized tests and twelfth grade dem onstrations of m astery. E xposition’s faculty does not em brace standardized tests, but reluctantly accepts the need for standardized measures o f student performance. “If our kids are learning to think, then they will perform on these tests anyway.” A recent attem pt by the state to im plem ent a curriculum -based standardized test, aborted in the face o f political pressure, was nonetheless applauded by Exposition faculty as indicative of a next wave of testing, much more favorable to measuring what they value. Student demonstrations of mastery, many faculty members com m ented, provide what current versions o f standardized tests cannot. T h e dem onstration is loosely modeled after the doctoral thesis. Students, under th e guidance of a com m ittee consisting of at least one faculty m em ber and two community members, select a problem o f interest and work toward a public presentation o f w hat they have learned. Teachers report that the dem onstration rivals the fall football games in terms of student engagement and parental attendance. A teacher is hired full tim e to coordinate professional developm ent at Exposition. T h e staff is surveyed each year to determ ine priorities for professional developm ent. T h e program, according to the coordinator, focuses narrowly on one or two aspects of teaching. “Relevance is critical. If these teachers cannot see that what they are learning can be applied to the classroom, then we’re all wasting our time. T his doesn’t m ean we go for quick fixes. Sometimes w e’ll spend four or five days a year on a single topic like student assessment. Teachers practice what they are learning between sessions and we revisit areas of difficulty.” Exposition uses th e state maximum of eight professional developm ent days per year, a num ber that initially had to be negotiated with a now supportive community. 84 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Highland Secondary T h e principal of Highland Secondary blames “wimpy administrators” for protecting ineffective teachers. “I use evaluation as an im portant tool to send a message to those who should not be teaching that they need to improve or leave. I have restored the m eaning o f a “satisfactory” rating. It takes years to dismiss a ceacher for poor performance, but I’ll be dam ned if I give a satisfactory rating to a crummy teacher.” Teachers at Highland are split in their opinions regarding evaluation. Some agree with the principal that the tim e is now to support rigorous standards for teachers. O thers see teacher evaluation as a weapon, used imperiously by th e principal to tarnish the reputations o f those with whom she has a personal vendetta. Students at Highland, like those at L e Conte and Exposition, take standardized tests. T est results are used at the student level for placem ent decisions and at th e school level to identify weaknesses. T est scores are disaggregated by race, gender, and teacher. T h e principal, armed with each teacher’s student test results, requires teachers to subm it an action plan aimed at improving test scores. Teacher reaction to this policy is mixed. Some are pleased by th e emphasis on the basic skills and motivated by the feedback from the hard data. Others give no credence to the data, citing difficulties o f interpretation (“How different are students at the 45th and 55th percentiles?” “How are my efforts distinguishable from all the other influences on students these days?”), and thus see test scores as unwelcome intruders. Professional developm ent at Highland similarly evokes a m u ltitu d e o f responses. “T h e principal usually chooses someone entertaining.” “I’ve learned a few tricks from presenters, but not much more than that.” “Fads! In-service days are simply a tribute to what is in vogue. T hese people m ake all kinds of money telling those of us on the front-lines what to do.” Although responses are mixed, there are strikingly few com m ents at the extrem es. T eachers are neither enthusiastic nor despondent about professional developm ent. Most responses fill the middle range betw een slight am usem ent and mild dissatisfaction. T h e principal defends the professional developm ent program, saying that her teachers need to be exposed to and cajoled to adopt current 85 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. ideas about “best practice” in teaching. “T oo many o f these teachers see them selves as learned, rather than learning. Professional developm ent is about changing that perspective.” Some Observations At Le Conte Secondary School, the lack of a discernible control system does not change the fact that teachers at che school are relatively powerless. T hey may, and probably do, close their doors and have dominion over their classrooms (for better or for worse). Teachers at L e C onte have no clear way to evaluate the quality of their work. Additionally, they appear to have resigned themselves to the fact chat professional developm ent is so in name only. L e C onte may represent the Faustian bargain in reverse: Teachers have preserved their autonomy at th e cost of knowledge and power. Administrators at both Exposition and Highland appear to have students’ best interest in mind. T h e nature of the concrol system at each school is quite different. Exposition is a work-in-progress: T hey have not settled on “one best way” for either their evaluation practices or their assessm ent of student achievement. Teachers do not seem to mind the ambiguity of work at Exposition. Instead, they appear to embrace the challenge of doing things better. At Highland, the principal, not the teachers, comes to the foreground. She is the control system. T he principal is a doer, and uses her power as effectively as she can. T e st data is analyzed and teachers are expected to improve. Teachers are her “work-in-progress.” Although it is possible co fie chese schools inco che cheorecical framework and m ake predictions about the level of teacher morale based on the location within the framework, chis would be a disservice to che framework and, more importancly, to the schools. T h e framework alerts us to the salient aspects of control systems in general. It alerts us to potential threats imposed by the system and helps us identify levers co improve not only the system, but also che work at hand. 86 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Contributions o f the study Theoretical Contributions T his study, by expanding the conceptualization o f bureaucratic control to include both technical and social characteristics, takes a significant step forward toward explaining why earlier empirical studies have been equivocal regarding the effects of control. T h e addition o f social and technical characteristics allows for a m ore precise test o f key features of W eber’s theory of bureaucracy most relevant to control systems, nam ely hierarchy of authority and standardization of work processes. T h e forms o f control examined here are most often exercised through these two mechanisms. W ithout technical knowledge, standardization is unlikely to improve th e work and hierarchy of authority becomes an arbitrary exercise of power. W ithout an enabling rationale, standardization turns workers into drones and hierarchy of authority descends to coercion. Thus, the theory propounded explicitly recognizes that the design and im plem entation of control systems are social processes that also dem and technical com petence. As the data show, control system s that do not attend to both technical and social considerations are likely to adversely impact morale. Practical Consequences T h e practical consequences flow directly from the theoretical contributions. Teachers, it turns out, discriminate among control systems (replication will show w hether these findings will extend to other occupations as well). A control system is designed to direct, motivate and increase the capacity of workers. Thus, it is in the best interests o f the control system designer to attend to those factors that facilitate the goals of the control system. Control system design is a multi-faceted, multi-stage process. Attention to both technical and social aspects of the design and im plem entation tasks is essential for effective control system functioning. One cannot control what one cannot understand. Teachers recognize when the control system is failing to assist them become b etter at their chosen task. Likewise, teachers resent a design and implementation process that fails to recognize ceachers as willing participants in helping 87 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. students succeed. M ost teachers have come to th e profession with a strong sense of purpose (FuIIan 1993). A socially inept control system neglects this fact. T h e control system designer does well to heed the portion of the Hippocratic Oath that reads “above all do no harm.” W hen teachers do not perceive the control system as beneficial to their work, morale suffers. Although the current study does not make a strong case for the morale- perform ance link, common sense dictates th at students are not best served by demoralized teachers. Thus, establishing a tight control system without obtaining th e necessary conditions of technical and social com petence will likely work against the intentions of the control system designer. T he data show that the tighter the control system, the more serious are the attitudinal consequences of neglecting either technical or social considerations. Lim itations o f the stu dy T h e cross-sectional nature of the study design presents some limitations. First, because data were measured at a single point in time, it is not clear w hether a change in control system characteristics will bring about any real changes in morale. It is possible that the control system design implications of the study will have their intended effects. It is also possible that short-term efforts to increase technical knowledge may reveal deficiencies in practice that will adversely impact morale, at least in the short run or until such deficiencies are overcom e. This happens because the easiest way to increase technical knowledge is to pay close attention to current practice in order to identify best practices. Teachers’ work most frequently occurs behind closed doors, in the absence of other adults (Hargreaves 1993; Lortie 1975; Rosenholtz 1989). Replacing this norm of privacy with a norm of continual im provem ent will require sustained effort, over and above control system design efforts. Similarly, initial efforts at enablem ent may not have their in ten d ed effect. A sudden willingness to involve formerly disenfranchised teachers in decision m aking and design processes may increase skepticism and/or feelings of work overload. Either consequence could adversely affect morale. Indeed, similar occurrences followed the administration of school-based management 88 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. programs, whereby programs ostensibly designed to increase teacher em pow erm ent were m et with resistance and/or fatigue (Rowan 1990). Another shortcoming of the current study concerns common-mechod bias. T h e data for the key findings relating control system characteristics and morale were collected via a single survey instrum ent. Com m on-m ethods bias is noc a concern for the perform ance m easure. Both standardized test results and the principals’ assessm ent of teachers’ organizational citizenship behavior were collected independendy from teacher reports of control system characteristics. A final concern for most cross-sectional studies is causal priority. T h e hypothesized model presents control system tightness as causal for morale and performance. It is possible that the causal direction is reversed. For example, poorly performing schools could be subject to tighter control. Given this scenario, better performing schools could afford their teachers relatively more freedom. W hile plausible for the direct effect o f control system tightness on perform ance or morale, the reverse causal direction scenario breaks down for the moderated effect hypothesis supported by this study. It is difficult to even construct a sensible sentence using a reverse-causal construction. M orale would have to cause control system tightness only under the conditions of high task knowledge and high enablem ent. T his said, it remains true that a cross-sectional study cannot assert causality, but m ust rely on strong theory for its causal grounding. A follow-up to the current study could help establish three things: first, w hether a change in control system tightness anticipates a change in teacher morale; second, w hether alterations to control system design along the lines recom m ended by this study, keeping tightness constant, will have their intended effects; lastly, w hether it is possible to reduce such common-methods effects as consistency biases, halo effects and prim ing effects by separating the survey questions regarding the control system in place from those survey questions regarding teacher attitudes (or even using different informants). A reduction of com m on-m ethods bias would increase the confidence with which study-based recommendations could be made. T h e study also found that the covariates overpowered the m odel’s hypothesized variables as predictors of the performance criterion variables. A follow-up study would require longitudinal data 89 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. at th e student-level. A . com parison of individual stu d en ts’ perform ance across tim e would effectively control for those covariates related to stu d en t background characteristics. T his strategy relies on the fact that, for most students, background characteristics are relatively stable over short periods o f time (say one to four years). Unfortunately, th e availability o f stu d en t data, especially student-level data that allows comparisons among teachers and departments, is quite limited. Implications and Final Thoughts T h e results of this study suggest several implications for those designing and/or administering control systems. First, the control system is like a toolbox. W ithin the box, each tool has a purpose, a potential to facilitate work, and various limitations. Clarity about the requirements o f the work and the proper ways o f using each tool are essential elements to successful completion of the task. Failure to attend to these conditions introduces the possibility that wrong tools will be selected and/or that the tools selected will be used improperly. T h e tools at hand for the control system designer are three types o f controls: personnel/cultural, action, and results. T h e ir use is dictated by the information available regarding the identity of the behavior to be controlled and the ability to m easure and reward such behavior. T his requires the control system designer and administrator(s) to critically assess whether the existing conditions allow for the use of a given tool or set o f tools. U nm erited optim ism in this assessm ent will likely result in over-control and adverse consequences for both th e individuals subject to the mis-specified control system and the organization that depends on their cooperation. Second, control system design and im plem entation are social processes. For teaching, as with other com plex tasks, the technical knowledge necessary for effective control is sim ply not available to the organization without extensive cooperation. Control through cooperation may require a mind- shift away from the belief that all control system s will be opposed. As a strategy for eliciting compliance, persuasion will be far superior to coercion. T hinking in terms o f persuasion is made easier when one makes the assum ption that teachers want to do their jobs well. Such thinking is credited for the positive turn-around at autom obile factories adopting cooperative and enabling 90 R eproduced with perm ission o f the copyright owner. Further reproduction prohibited without permission. w ork processes (Adler 1993). I f autom obile factories can encourage w orkers toward greater productivity by operating as if they believed their employees w anted to do quality work, how much easier should it be for schools, w here the product is student success? T hird, trust is essential. T h o se subject to the system m ust trust the system itself as well as those adm inistering the system . T h e re are close parallels becw een th e th ree elem ents of adm inistrative trust—com petence, consistency of treatm ent, and benevolence— and the three com ponents of an enabling control system — validity, non-coercion, and support. As the trust literature makes clear, building crust is m uch more difficult than eroding trust. T h e control system designer and administrator have th e doubly difficult task of building trust in both the system and the person(s) administrating the system . H onest, detailed communication regarding the state of the control system, including the design intent and technical merits, will likely go a long way toward establishing the necessary trust. Control and Consideration: N ot such strange bedfellows after all T his study resurrects and provides support for insights articulated nearly three-quarters of a century ago by a prominent m em ber of Frederick Taylor’s Scientific M anagem ent Society, Mary Parker Follett. In her discourse on “the giving of orders”, F ollett offers sound advice, entirely consistent with this study’s em pirical findings. First, she warns managers against giving arbitrary directions. Arbitrary directions, she asserts, undermine pride of work. If there exists a better work process, it is best to use persuasion rather than force. We are wise to understand the inherent value o f good work to the worker and not to underm ine a sense o f responsibility through th e issuing of arbitrary commands (Follett 1949 in Graham 1995, pl25-127). Consonant w ith her admonitions against arbitrariness, Follett charges her readers to reconsider th e nature of power. She offers the distinction between “power over” and “power-with.” Power-with is “a j'ointly developed power, a co-active, not a coercive power (p. 102-103).” At the heart of Follett’s writings, and central to the theory elaborated in this study, is the pairing o f control and consideration. Consideration is used in a dual sense: referring to both 91 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. thoughtfulness and deliberation. At its best, a control system gains its authority from the situation and its power from willing subjugation. T h e situation defines what is needed for individuals to successfully com plete the task. W illing subjugation is the product of thoughtful and deliberate interaction among those who direct and undertake the task. Only w hen control is properly considerate will individuals proffer their autonomy for the good of the collective and, thus, feel good about control. 92 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Bibliography Adler, Paul S. 1993. “T h e Learning Bureaucracy.” Pp. 111-194 in Research in Organizational Behavior, vol. 15, edited by B. M. Staw and L. L. Cummings. Greenwich, CT: JAI Press. Adler, Paul and Brian Borys. 1996. “T w o T ypes o f Bureaucracy: Enabling and Coercive.” Administrative Science Quarterly 41:61-89. Agarwal, Sanjeev. 1993. “Influence of formalization on role stress, organizational commitment, and work alienation o f salespersons: A cross-national study.” Journal o f International Business Studies 24:715-739. Agarwal, Sanjeev and Sridhar N . Ramaswami. 1993a. “A ffective Organizational Commitment of Salespeople: An expanded M odel.” Journal o f Personal Selling and Sales Management 13:1-22. Agarwal, Sanjeev and Sridhar N . Ramaswami. 1993b. “M arketing Controls and Em ployee Responses: T h e Moderating Role of Task Characteristics.” Journal o f the Academy o f Marketing Science 21:293-306. Aiken, M and J Hage. 1966. “Organizational Alienation: A Comparative Analysis.” American Sociological R eview 31:497-507. Argyris, Chris. 1951. The Impact o f Budgets on People. N ew York: Controllership Foundation. Argyris, Chris. 1964. Integrating the individual and the organization. London: Tavistock. Axelrod, R. 1984. The Evolution o f Cooperation. N ew York: Basic Books. Bacharach, Samuel B. and Peter Bamberger. 1990. “Exit and Voice: Turnover and M ilitancy Intentions in Elem entary and Secondary Schools.” Educational Administration Quarterly 26:316-344. Bandura, A. 1986. Social foundations o f thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice Hall. Bausmann, R. L. 1960. “On finite sample distributions of generalized classical linear identifiability test statistics.” Econometrica 45:939-952. 93 R eproduced with perm ission of the copyright owner. Further reproduction prohibited without permission. Beehr, T . A., J. T . Walsh, and T. D . Taber. 1976. “Relationship o f stress to individually and organizationally valued states: Higher order needs as a moderators.” Journal o f Applied Psychology 61:41-47. Bentler, P eter M. 1990. “Comparative fit indeces in structural equations.” Psychological Bulletin 107:238-246. Blau, Peter M. 1955. T he Dynamics o f Bureaucracy. Chicago: University o f Chicago Press. Bollen, Kenneth A. 1995. “Structural equation models that are nonlinear in latent variables: A least squares estimator.” Pp. 223-251 in Sociological M ethodology, edited by P. M. Marsden. Cambridge, MA: Blackwell. Bollen, K enneth A. and Pamela Paxton. 1998. “T w o-Stage L east Squares Estimation o f Interaction Effects.” Pp. 125-151 in Interaction and Nonlinear Effects in Structural Equation M odeling, edited by R. E. Schumacker and G. A. Marcoulides. Mahwah, N ew Jersey: Lawrence Erlbaum Associates. Bond, Lloyd. 1998. “Culturally R esponsive Pedagogy and the A ssessm ent o f Accomplished Teaching.” Jounal o f Negro Education 67:242-254. Brockner, Joel, Phyllis Siegel, Joseph Daly, and Tom Tyler. 1997. “W hen trust matters: T h e m oderating effect o f outcom e favorability.” A d m in istra tive Science Quarterly 42:558-583. Brown, John Seely and Paul Duguid. 1992. “Enacting design for the workplace.” Pp. 164-198 in Usability: Turning Technologies into Tools, edited by P. S. Adler and T . W. Winograd. New York: Oxford Univeristy Press. Bryk, A. and Steven Raudenbush. 1992. Hierarchical linear models in social and behavioral research: Applications and data analysis methods. Newbury Park: Sage. Byrne, Barbara. 1998. Structural Equation M odeling with LISREL, PRELIS, and SIMPLIS: Basic Concepts, Applications, and Programming. Mahwah, N ew Jersey: Lawrence Erlbaum Associates. Cheney, Lynne. 1994. “T h e End of History.” in Wall Street Journal. N ew York, N ew York. Chubb, John E. and Terry M. Moe. 1988. “Politics, Markets, and the Organization of Schools.” American Political Science R eview 82:1065-87. 94 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Clark, Murray C. and Roy L. Payne. 1997. “The nature and structure o f workers' trust in management.” Journal o f Organizational Behavior 18:205-224. Cohen, David. 1990. “A Revolution in One Classroom: T h e Case o f Mrs. Oublier.” Educational Evaluation and Policy Analysis 12:327-45. Conley, S, S Barcharach, and S Bauer. 1989. “The school work environm ent and teacher career dissatisfaction.” Educational Administration Quarterly 25:58-81. Cook, John and T o b y Wall. 1980. “N ew work attitude m easures o f trust, organizational com m itm ent and personal need non-fulfilm ent.” Journal o f Occupational Psychology 53:39-52. Costigan, Robert D, Selim S liter, and Jason J Berman. 1998. “A multi-dimensional study of trust in organizations.” Journal o f Managerial Issues 10:303-317. Cuban, Larry. 1988. T h e managerial imperative and the practice o f leadership in schools. Albany: State University o f N ew York Press. Daft, Richard L. and Norman B. Macintosh. 1981. “A T entative Exploration into the Amount and Equivocality o f Information Processing in Organizational Work Units.” Administrative Science Quarterly 26:207-224. Darling-Hammond, Linda. 2000. “Teacher Quality and Student Achievem ent: A Review of State Policy Evidence.” Education Policy Analysis Archives v8:l. D iggins, John Patrick, John D . Fonte, Robert Learner, H erbert London, and Diane Ravitz. 1997. “Symposium: National Standards.” Society 34:9-13. Dillman, Donald A. 1978. M ail and Telephone Surveys: T h e T otal Design Method. N ew York: John W iley. Dubinsky, Alan, Ronald E. M ichaels, Masaaki Kotabe, Chae U n Lim, and H ee- Cheol Moon. 1992. “Inluence o f role stress on industrial salespeople's work outcomes in the United States, Japan, and Korea.” Journal o f International Business Studies 23:77-100. Eaton, William Edward. 1990. Shaping the superintendency: a reexamination o f Callahan and the cult o f efficiency. N ew York: Teachers C ollege Press. Edwards, Richard. 1979. Contested Terrain. New York: Basic Books. F irestone, W illiam A. and B. D . Bader. 1992. R edesign in g teaching: Professionalism or Bureaucracy I, vol. 63. Albany, N ew York: SU N Y Press. 95 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Fischer, Frank. 1984. “Organizational Expertise and Bureaucratic Control: Behavioral Science as Managerial Ideology.” Pp. 172-195 in Critical Scudies in Organization and Bureaucracy, edited by F. Fischer and C. Sirianni. Philadelphia: T em ple University Press. Folger, R and MA Konovsky. 1989. “Effects of procedural and distributive justice on reactions to pay raise decisions.” Academy o f M anagement Journal 32:115- 130. Follett, Mary Parker. 1949. “T h e Giving of Orders.” Pp. 16-33 in Freedom Sc Co ordination: Lectures in Business Organization by M ary Parker Follett, edited by L. Urwick. London: M anagement Publications Trust. Follett, Mary Parker. 1973 (1925). “Power.” Pp. 66/87 in Dynamic Administration: The C ollected Papers o f M ary Parker Follett, edited by E. M. Fox and L. Urwick. London: Pitman. Foster, E. M ichael and Sara McLanahan. 1996. “An Illustration o f the U se of Instrumental Variables: Do Neighborhood Conditions A ffect a Young Person's Chance of Finishing High School?” Psychological M ethods 1:249-260. Fowler, Floyd J. 1993. S u rvey Research M ethods. N ew bury Park: Sage Publications. Freeston, K. R. 1987. “Leader Substitutes in Educational O rganization.” Educational Administration Quarterly 23:45-59. Fullan, Michael. 1993. Change Forces: Probing the D epths o f Educational Reform. London: Falmer. Fusilier, Marcelline R., Daniel C. Ganster, and Bronston T . Mayes. 1987. “Effects of Social Support, Role Stress, and Locus o f Control on H ealth.” Journal o f Management 13f:517-528. Gouldner, Alvin W. 1954. Patterns o f Industrial Bureaucracy. N ew York: Free Press. Gunz, Hugh P and Sarah P Gunz. 1994. “Professional/organizational commitment and job satisfaction for em ployed lawyers.” Human Relations 47:801-828. Hall, Richard. 1968. “Professionalization and Bureaucratization.” Am erican Sociological R eview 33:92-104. 96 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Hargreaves, Andy. 1993. “Individualism and Individuality: Reinterpreting the Teacher Culture.” Pp. 51-76 in Teachers' Work, edited by J. W. Little and M. McLaughlin. N ew York: Teachers College Press. Hofstede, Geert. 1978. “T he Poverty of M anagement Control Philosophy.” Academy o f Management R eview 3:450-461. Huberman, Michael. 1993. “T h e Model o f the Independent Artisan in Teachers' Professional Relations.” Pp. 11-50 in Teachers' Work, edited by J. W. Little and M. M cLaughlin. N ew York: Teachers C ollege Press. Iaffaidano, M. T . and P. M. M uchinsky. 1985. “Job satisfaction and job performance: A meta-analysis.” Psychological Bulletin 97:251-273. Jaccard, James, Robert Turrisi, and Choi K. Wan. 1990. Interaction effects in multiple regression. Thousand Oaks, CA: Sage. Jackson, Susan E. 1983. “Participation in decision making as a strategy for reducing job-related strain.” Journal o f Applied Psychology 6:3-19. Jaworski, Bernard J. and Deborah J. M aclnnis. 1989. “M arketing Jobs and Management Controls: Toward a Framework.” Journal o f Marketing Research 26:406-419. Johnston, Mark W., A Parasuraman, Charles M. Futrell, and William C. Black. 1990. “A Longitudinal Assessm ent of the Impact o f Selected Organizational Influences on Salespeople's Organizational C om m itm ent D uring Early Employment.” Journal o f Marketing Research 27:333-344. Joreskog, Karl G. 1998. “Interaction and N on-linear M odeling: Issues and Approaches.” Pp. 239-250 in Interaction and Nonlinear Effects in Structural Equation M odeling, edited by R. E. Schum acker and G. A. M arcoulides. Mahwah, N ew Jersey: Lawrence Erlbaum Associates. Joreskog, Karl G. and Dag Sorbom. 1994. “L ISR E L ”. Chicago, Illinois: Scientific Software International, Inc. Kahili, S. 1988. “Symptoms o f professional burnout: A review o f the empirical evidence.” Canadian Psychology 29:284-297. Kerr, Steven. 1975. “On the Folly o f Rewarding A while Hoping for B.” A cadem y of Management Journal 18:769-83. 97 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kim, Chan W and R enee A Mauborgne. 1993. “Procedural justice, attitudes, and subsidiary top managem ent compliance with multinationals' corporate strategic decisions.” Academy o f Management Journal 36:502-526. King, R.C. and V. Seth i. 1997. “T h e m oderating effe ct o f organizational com m itm ent on burnout in information system s professionals.” European Journal o f Information System s 6:86-96. Kolker, Claudia. 1999. “T exas Offers Hard Lessons on School Accountability; Education: C heating alleged after job security is linked to test scores. California, other states have mimicked plan.” in Los Angeles Tim es. April 14, 1999. Los Angeles, CA. Koretz, David. 1992. “W hat Happened to T e s t Scores, and Why?” Educational Measurement: Issues and Practice :7-l 1. Konovsky, Mary and S. D ouglas Pugh. 1994. “Citizenship behavior and social exchange.” Academy o f Management Journal 37:656-669. Kramer, Rod M. and T om R. Tyler. 1996. T rust in Organizations. Thousand Oaks, CA: Sage. Lawler, Edward E. and John Grant Rhode. 1976. Information and Control Systems in Organizations. Pacific Palisades, CA: Goodyear. Leonard-Barton, Dorothy and D eepak Sinha. 1993. “D eveloper-user interaction and user satisfaction in internal technology transfer.” Academy o f Management Journal 36:1125-1139. Lewicki, Roy J. and B.B. Bunker. 1996. “D eveloping and maintaining trust in work relationships.” Pp. 114-139 in Trust in organizations: Frontiers o f theory and research, edited by R. M. Kramer and T . R. Tyler. Thousand Oaks: Sage. Li, Fuzhong and Peter Harmer. 1998. “M odeling Interaction Effects: A Two-Stage Least Squares Exam ple.” Pp. 153-166 in Interaction and Nonlinear Effects in Structural Equation M odeling, edited by R. E. Schum acker and G. A. Marcoulides. Mahwah, N ew Jersey: Lawrence Erlbaum Associates. Lind, E. Allan and T om Tyler. 1988. The social psychology o f procedural justice. N ew York: Plenum Press. Lodahl, T . and M. Kejner. 1965. “T h e definition and m easurem ent o f job involvement.” Journal o f Applied Psychology 49:24-33. 98 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Long, J. Scott. 1983. Confirmatory Factor Analysis: A Preface to LISREL, vol. 33. Beverly Hills, CA: Sage. Lortie, D. 1975. Schoolteacher: A Sociological Study. Chicago: University o f Chicago Press. Luhman, Nicolas. 1979. Trust and Power. Chichester: Wiley. MacCallum, R.C., M.W. Browne, and H.M. Sugawara. 1996. “Power analysis and determination o f sample size for covariance struccure modeling.” Psychological M ethods 1:130-149. Malen, B, Rodney T . Ogawa, and J Kranz. 1990. “What do we know about school- based management? A case study o f the literature- a call for research.” in Choice and Control in American Education, Volume 2: The Practice o f choice, decentralization, and school restructuring., edited by W. H. C lune and J. F. W hite. London: Falmer Press. Markel, Karen S. and Michael R. Frone. 1998. “Job Characteristics, Work-School Conflict, and School O utcom es Among A dolescents.” Journal o f Applied Psychology 83:277-287. Martin, J and C Siehl. 1983. “Organizational Culture and Counterculture: An Uneasy Symbiosis.” Organizational Dynamics 12:52-64. Maruyama, Geoffrey M. 1998. Basics o f Structural Equation M odeling. Thousand Oaks, CA: Sage. Maslach, Christina and Susan E Jackson. 1981. “T h e measurement o f experienced burnout.” Journal o f Occupational Behavior 2:99-113. Mayer, RC, JH Davis, and FD Schoormann. 1995. “An Integrative M odel of Organizational Trust.” Academy o f Management Review 20:709-734. McAllister, Daniel J. 1995. “Affect- and Cognition-Based Trust as Foundations for Interpersonal Cooperation in Organizations.” Academy o f M anagement Journal 38:24-59. McGregor, Douglas. 1960. T h e Human Side o f Enterprise. N ew York: McGraw- Hill. M cN eil, Linda M. 1986. Contradictions o f Control: School structure and school knowledge. N ew York: Routledge and Kegan Paul. Merchant, Kenneth A. 1985a. Control in Business Organizations. Boston: Pitman. 9 9 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Merchant, Kenneth A. 1985b. “Organizational Controls and Discretionary Program D ecision Making: A F ield Study.” Accounting, Organizations, and Society 10:67-85. M erchant, K enneth A. 1990. “T he effects o f financial controls on data m anipulation and m anagem ent myopia.” Accounting, Organizations, and Society 15:297-313. Merchant, Kenneth A. 1998. Modern Management Control Systems. Upper Saddle River, N ew Jersey: Prentice-Hall. M ichaels, Ronald E., W illiam L Cron, Alan J D ubinsky, and Erich A Joachimsthaler. 1988. “Influence o f Formalization on the organizational com m itm ent and work alienation of salespeople and industrial buyers.” Journal o f M arketing Research 25:376-383. Miller, A. 1967. “Professionals in Bureaucracy: Alienation Am ong Industrial Scientists and Engineers.” American Sociological R eview 32:755-67. Morgan, Gareth. 1986. Images o f Organization. Newbury Park: Sage. Nash, Gary B., Charlotte Crabtree, and Ross E. Dunn. 1998. H istory on Trial: Culture Wars and the Teaching o f the Past. N ew York: Alfred A. Knopf. Organ, Dennis W. and Charles N. Greene. 1981. “T h e effects o f formalization on professional involvement: a compensatory process approach.” Administrative Science Quarterly 26:237-252. Organ, D ennis W. and Katherine Ryan. 1995. “A m eta-analytic review of atdtudinal and dispositional predictors of organizational citizenship behaviors.” Personnel Psychology 48:775-802. Ostroff, Cheri. 1992. “T he Relationship Betw een Satisfaction, Attitudes, and Perform ance: An Organizational L evel A nalysis.” Journal o f A pplied Psychology 77:963-974. Otley, David and Anthony Berry. 1980. “Control, Organization, and Accounting.” Accounting, Organizations and Society 5:231-244. Ouchi, W illiam G. 1979. “A Conceptual Framework for the design of organizational control mechanisms.” Management Science 25:833-847. Peterson, Kent D. 1984. “Mechanisms of Administrative Control over Managers in Educational Organizations.” Administrative Science Quarterly 29:573-597. 100 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Petty, M. M., G. W. M cG ee, and J. W. Cavender. 1984. “A meta-analysis o f the relationship betw een individual job satisfaction and individual performance.” Academ y o f Management R eview 9:712-721. Pierce, C. Mark and G eoffrey N . Molloy. 1990. “Relations betw een school type, occupational stress, role preceptions and social support.” Australian Journal of Education 34:330-338. Podsakoff, Philip M and Scott B. Mackenzie. 1994. “Organizational Citizenship Behaviors and Sales U nit E ffectiveness.” Journal o f M arketing Research 31:351-363. Podsakoff, Philip M, Larry J Williams, and William D Todor. 1986. “Effects of organizational form alization on alienation am ong professionals and nonprofessionals.” Academ y o f Management Journal 29:820-831. Polanyi, Michael. 1962. Personal Knowledge: Toward a Post-Critical Philosophy. N ew York: Harper Torchbooks. Prussia, Gregory E. and Angelo J. Kinicki. 1996. “A Motivational Investigation of Group E ffectiveness U sing Social-Cognitive T h eory.” Journal o f A pplied Psychology 81:187-198. Ramaswami, Sridhar N ., Sanjeev Agarwal, and M ukesh Bhargava. 1993. “Work Alienation o f M arketing Employes: Influence o f Task, Supervisiory, and Organizational Structure Factors.” Journal o f th e Academ y o f M arketing Science 21:179-193. R A N D . 1994. “T h e U se and M isuse of T est Scores in Reform D ebate.” RAND Institute on Education and Training Policy Brief, Santa Monica, CA. Riggs, Matt L. and Partick A. Knight. 1994. “T h e Im pact of Perceived Group Success-Failure on M otivational Beliefs and A ttitudes.” Journal o f Applied Psychology 79:755-766. Rizzo, J. R., R. J. House, and S. J. Lirtzman. 1970. “Role conflict and ambiguity in com plex organizations.” Administrative Science Quarterly 15:150-63. Rosenholtz, Susan J. 1989. Teacher's Workplace: A social organizational analysis, vol. 95. W hite Plains, NY: Longman. Rowan, Brian. 1990. “Com m itm ent and Control: Alternative Strategies for the Organizational Design o f Schools.” Review o f Research in Education 16:353-389. 101 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Rowan, Brian, Stephen W. Raudenbush, and Sang Jin Kang. 1991. “School Climate in Secondary Schools.” Pp. 203-223 in Schools, Classrooms, and Pupils: International Studies o f Schooling from a M ultilevel Perspective, edited by S. W. Raudenbush and J. D . Willms. San Diego, CA: Academic Press. Russell, D, E Altmaier, and D Van Velsen. 1987. “Job-related stress, social support, and burnout among classroom teachers.” Journal o f A pplied Psychology 12:269- 274. Sarason, Seymour B. 1982. T h e Culture o f the School and the Problem o f Change. Boston: Allyn and Bacon. Sarason, Seymour B. 1996. Revisiting “The Culture o f the School and the Problem o f Change”. N ew York: Teachers College Press. Scott, W.R. 1992. Organizations: Rational, Natural, and Open Systems. Englewood Cliffs, NJ: Prentice Hall. Shulman, L ee S. 1987. “K now ledge and Teaching: Foundations o f the N ew Reform.” Harvard Educational Review 57:1-22. Smircich, Linda. 1983. “C oncepts o f culture and organizational analysis.” Administrative Science Quarterly 28:339-358. Smylie, Mark. 1990. “Teacher Efficacy at Work.” Pp. 48-66 in Teachers and Their Workplace, edited by P. Reyes. Thousand Oaks: Sage. Steiger, J. H. 1990. “Structural m odel evaluation and modification: An interval estimation approach.” Multivariate Behavioral Research 25:173-180. Sydow, Jorg. 1998. “Understanding the Constitution o f Interorganizational Trust.” Pp. 31-87 in Trust Within and Between Organizations: Conceptual Issues and Empirical Applications, edited by Christel Lane and Reinhard Bachmann. N ew York: Oxford. Talbert, Joan E. and Milbrey W. McLaughlin. 1994. “Teacher Professionalism in Local School Contexts.” American Journal o f Education 102:123-153. Tang, T hom as L and Linda J Sarsfield-Baldwin. 1996. “D istributive and procedural justice as related to satisfaction and com m itm ent.” SANI Advanced Management Journal 61:25-31. 102 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Taylor, Susan M, Kay B Tracy, Monika K Renard, and Kline J Harrison. 1995. “D u e process in performance appraisal: A quasi-experim ent in procedural justice.” Administrative Science Quarterly 40:495-523. Thibaut, John and L Walker. 1975. Procedural Justice: A psychological analysis. Hillsdale, NJ: Erlbaum. Thom pson, Bruce. 1998. “Five M ethodology Errors in Educational Research: T h e Pantheon o f Statistical Significance and Other Faux Pas.” in A m erican Educational Research Association. San Diego, CA. Tyler, Tom . 1990. Why people obey the law. N ew Haven: Yale University Press. Tyler, Tom . 1994. “Psychological M odels of the Justice Motive: Antecedents o f D istributive and Procedural Justice.” Journal o f Personality and Social Psychology 67:850-863. Tyler, T om and Peter Degoey. 1995. “Collective Restraint in Social Dilemmas: Procedural Justice and Social Identification Effects on Support for Authorities.” Journal o f Personality and Social Psychology 69:482-497. Tyler, T om and E. Allan Lind. 1992. “A relational model o f authority in groups.” Pp. 115-191 in Advances in Experimental Social Psychology, vol. 25, edited by M. Zanna. N ew York: Academic Press. Vacha-H aase, T . 1998. “Reliability generalization: Exploring variance in m easurement error affecting score reliability across studies.” Educational and Psychological Measurement 58:6-20. Vroom, Victor H. 1964. Work and Motivation. N ew York: Wiley. Wagner, Robert B. 1989. Accountability in Education. N ew York: Routedge. Weber, Max. 1968. Economy and Society. Berkeley: University of California Press. W ildavsky, Aaron. 1979. Speaking Truth to Power. Boston: Little, Brown and Company. Willms, J. Douglas. 1992. Monitoring School Performance: A Guide for Educators. London: T h e Falmer Press. Wilson, James Q. 1989. Bureaucracy. N ew York: Basic Books. 103 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wright, Thomas A. and Russell Cropanzano. 2000. “Psychological W ell-Being and Job Satisfaction as Predictors o f Job Performance.” Journal o f Occupational Health Psychology 5:84-94. Zucker, Lynne G. 1977. “T h e Role o f Institutionalization in Cultural Persistence.” American Sociological R eview :726-743. R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix A LISREL 8.30 Input: Covariance Matrix PCONTROL OCONTROL PERFK1 PERFK2 PERFK3 PROCICL PCONTROL OCONTROL PERFKX PERFK2 PERFK3 PROCK1 SUPPORT NCOERCE VALID PRIDE SATIS EFFICACY STAR WHITE MEALS CALWOR PUBENR YRS EXP 0 .54 0 .44 0.01 0.01 0 . 0 1 0.02 0 . 2 1 0 .15 0.24 0 . 2 0 -0.04 0 .22 2.74 0.05 - 0.01 0 . 0 0 -0.24 -1.04 0 .44 0 .00 0.01 0.01 0 .01 0.16 0 . 1 2 0 .2 0 0.16 0.00 0 .25 17 04 02 00 -0.24 -0.89 0 .25 0 .10 0.09 0 .11 0 .05 0.03 0.03 0 .18 0 .11 0.32 1.28 0.00 -0 .02 - 0.01 -0.13 0 .23 0 .34 0 .13 0 .09 0 .02 -0.03 0 .02 0.19 0 .03 0 .22 0 .53 0 .01 0.00 0 .00 -0.03 0 .76 0.32 0.09 0.02 0.04 0.05 0.23 0.06 0.25 2.45 0.02 -0.04 - 0.01 -0.07 0.42 0.49 0.08 0.04 0 .04 0.24 0 .23 .23 .84 .01 -0.03 - 0.02 0 . 0 0 1.05 SUPPORT NCOERCE VALID PRIDE SATIS EFFICACY STAR WHITE MEALS CALWOR PUBENR YRS EXP SUPPORT 0.36 0.24 0 .25 0.17 0.09 0.29 2.36 01 03 -0 .02 -0.08 0 .14 NCOERCE 0.35 0 .21 0.14 0 . 0 2 0 .26 3 .54 0 . 0 2 -0.04 -0.03 -0.16 0 . 1 2 VALID PRIDE SATIS EFFICACY 0.29 0 .19 0.06 0 .35 2.41 0.02 -0.03 -0 .02 -0 .13 -0 .23 1.45 1.27 0 .73 1.58 0.00 - 0.02 - 0.02 -0 .18 -0.43 92 33 11 08 07 01 16 -0.54 2.33 5.49 0.04 -0 . 04 - 0.02 -0 . 54 - 2.20 STAR WHITE STAR WHITE MEALS CALWOR PUBENR YRS EXP 295 .86 2.84 -3.90 - 1.20 -5.87 9.76 0 .04 -0.03 0.00 -0.07 -0 .10 MEALS 0.07 0 .03 0 .02 -0 .25 CALWOR PUBENR YRS EXP 0.02 0 .00 - 0.12 0.92 1.16 52.70 105 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix B LISREL 8.30 Syntax for Structural Equation M odel I Full SEM Model - MOR/'PERF Endogenous ! 1st Model - VARS Weighted using 2nd Order Factor LX's DA NI=47 NO=144 LA; SCHDEPT SCHTYP CALWOR DEPT MEALS PERFK1 PERFK2 PERFK3 PR0CK1 PUBENR RA1 RA3 RA4 RA5 RAS SCH# SPROCK_3 STAR STAR_5 WAVE WHITE YRS_EXP PRIDE SATIS EFFICACY SUPPORT NCOERCE VALID PCONTROL OCONTROL PRIDEM SATISM EFFICACM PCONTROM OCONTROM SUPPORTM NCOERCEM VALIDM PERFK1M PERFK2M PROCK1M RA1M RA3M RA4M RA5M RA6M PERFK3M CM=SEM1112.CM SE PCONTROL OCONTROL PERFK1 PERFK2 PERFK3 PP.OCK1 SUPPORT NCOERCE VALID PRIDE SATIS EFFICACY STAR WHITE MEALS CALWOR PUBENR YRS_EXP / MO NY=13 NX=5 NE=5 NK=5 LY=FU,FI LX=FU,FI SE=FU, FI GA=FU,FR BE=FU,FI PH=SY,FR PS=DI,FR TE=SY TD=SY LE CST TK ER MORALE PERF LK WHITE MEALS CALWORKS ENROLL TENURE FR LY(2,1) FR LY(4,2) LY(5,2) LY(6,2) FR LY(8,3) LY(9,3) FR LY{11,4) LY(12,4) FI TE(13,13) TD(1,1) TD(2,2) TD(3,3) TD(4,4) TD(5,5) FR BE(2,1) BE (3,1) BE(4,1) BE(5,1) BE(4,2) BE(5,2) BE(4,3) BE(5,3) BE(4,5) FR TE(11,10) TE(8,7) VA 0.01 TE(13,13) TD(1,1) TD(2,2) TD(3,3) TD(4,4) TD(5,5) VA 1.0 LY(1,1) LY(3,2) LY(7,3) LY(10,4) LY(13,5) VA 1.0 LX (1,1) LX(2,2) LX(3,3) LX(4,4) LX(5,5) OU RS SC MI AD=200 COMPUTE CSTxTK = pcontrom*(perfklm*.72) . COMPUTE CSTxEC = pcontrom*(supportm) . COMPUTE CSTECTK = pcontrom*(supportm)*(perfklm*.72) . COMPUTE CSTECIV1 = ocontrom*(ncoercem) . COMPUTE CSTECIV2 = ocontrom*(validm) . COMPUTE CSTTKIV1 = ocontrom*(perfk2m*.58) . COMPUTE CSTTKIV2 = ocontrom*(procklm*.59) . COMPUTE ThreelVl = ocontrom*(ncoercem)* (perfk2m*.58) . COMPUTE ThreeIV2 = ocontrom*(validm)* (procklm*.59) . EXECUTE . COMPUTE Pride = ((all*.79)+(al2*.78)+(al3*.58)) . COMPUTE Satis = ((al4*.82)+(jil*.83)+(ji2*.59)+(ji5*.60)) . 106 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. COMPOTE Efficacy = ((sec3*.SO)+(sec4*.55)+(sec5*. 58) +(see2*.51)+(see3*.61)). EXECOTE. COMPOTE MOR = ((pride *.86) + (satis*.54) + (efficacy*.39)) . EXECUTE. Note: Observed variables were mean-centered before constructing interaction variables. A ppendix C SPSS Syntax for Two-Scage Lease Square Analysis *2-Stage Least Squares. TSET NEWVAR=NONE . 2SLS MOR WITH white yrs_exp pubenr meals calworks [stage 1] pcontrol support perfkl [stage 2, include 1] cstxtk cstxec [stage 3, include 1&2 ] cstectkm [stage 4, include 1&2 ] /INSTROMENTS white yrs_exp pubenr meals calworks ocontrol ncoerce valid prockl perfk3 csteciv2 csttkiv2 threeivl /CONSTANT . R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 Appendix D: Additional Variables M easured, but Extraneous to th e T ested M odel Several ocher item s were included on this survey. T hese item s are not central to the hypotheses tested, but instead represent the variables that, prior to this study, offered the most plausible alternative explanations for the phenom ena of interest. Survey items used a five-point Likert-type scale anchored by strongly disagree (1) to strongly agree (5). Form alization was m easured using seven item s that assessed the use o f rules in the workplace. Aiken and Hage's (1966) item s were used in original form. An exam ple is “How things are done here is left up to che person doing the work (reverse coded).” Scores were averaged into a single index of formalization (a=.73). Supervisor procedural knowledge assessed the extent that supervisors m et M erchant's (1998) conditions for determ ining the effectiveness of action controls. This was a new scale. T h e th ree items are; “M y supervisor has accurate knowledge o f my classroom practice”, “My supervisor knows what actions result in higher student performance”, and “M y supervisor has been able to persuade m e to refine my teaching methods to improve studenc perform ance.” Scores on this m easure w ere averaged so chat higher scores indicate g reater supervisor procedural knowledge (a=.88). Task routineness was assessed with five items taken from Aiken and Hage's (1966) measure. An example is “Many o f my work tasks are che same from day to day.” Scores on this measure were coded and averaged so that higher scores indicate more routine work (a=.70). Procedural Justice was assessed with nine items adapted from T yler and D egoey (1995). Respondents were presented with a scenario concerning professional developm ent. T h e questions were designed to tap respondents' assessments of the three justice conditions — neutrality, voice, and respecc — theorized by T yler and others (T yler 1994; T y ler and Degoey 1995; Tyler and Lind 1992). For example, respondents were asked w hether “T h e decisions of the individual or com m ittee would treat you fairly.”; “ ... would be influenced by your views.”; 108 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. “...would respect your position as a teacher.” Item s were averaged, w ith higher scores indicating greater procedural justice (a=.95). T rust in O ne's Supervisor was assessed using items drawn from several studies (Brockner, Siegel, Daly, and T y ler 1997; Clark and Payne 1997; Cook and Wall 1980; Lewicki and Bunker 1996; McAllister 1995). T h ese items were selected co rep resen t tru st as defined by the underlying dim ensions of com petence, benevolence and consistency. T h is perspective is elaborated by Mayer, Davis and Schoorman (1995). Item s included “M y supervisor shows good judgm ent when m aking decisions” ; “I can usually trust my supervisor to do what is good for m e”; “My supervisor applies the same rules for all teachers.” Scores on these twelve items were averaged, with higher scores indicating higher levels of trust in one's supervisor (a=.97). Organizational citizenship behavior (OCB) was assessed w ith fourteen item s adapted from Podsakoff and M ackenzie's (1994) scale of the same name. T h e scale assesses the extent to which teachers believe that they and their colleagues act in a m anner consistent with the good of the school, as opposed to acting strictly out of self-interest. An exam ple is “Willingly gives of his or her time to help o th er teachers who have work related problem s.” T eachers were asked to answer fourteen questions on the OCB scale using three reference groups: self (a=.82), their departm ent (a=.91), and their entire school faculty (a=.91). For each reference group, scores were averaged so higher scores reflect higher levels of organizational citizenship behavior. In-role performance was assessed using the Los Angeles Unified School District evaluation form 98.32 (1995). LAUSD and other districts within Southern California use these questions as part of the principal's assessm ent of teacher performance. T h e three prompts are: “Consistently achieves objectives” , “Planning and preparation is ev id en t” , and “Classroom performance indicates professional com petence.” I asked principals to respond to these questions at the departm ent or grade level. High school principals responded for the four core departm ents, middle school principals for two, and elementary school principals for grades 3 and 5 (a=.87). 109 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Star CONTROL SYSTEM TIGHTNESS 0.20 1.00; 0.46 0.13 0.27 10e Formalization 1 2.26 1.20 1 5 -0.09 -0.03 -0.36: -0.03 -0.16 10i Formalization 2 1.91 1.07 1 5 -0.14 -0.07 -0.32: -0.06 -0.20 101 Formalization 3 2.42 0.95 1 5 -0.12 0.10 -0.03; -0.02 -0.02 lOo Formalization 4 2.27 1.00 1 5 -0.18 0.01 -0.21 -0.18 -0.01 10r Formalization 5 3.36 1.09 1 5 0.01 0.32 0.21 0.05 0.20 10t Formalization 6 2.32 0.99 1 5 -0.21 0.02 -0.24 -0.15 -0.01 10y Formalization 7 3.35 1.07 1 5 0.00 0.23 0.01 -0.05 0.07 TECHNICAL KNOWLEDGE 0.46 0.13; 0.25 1.00 0.28 Supervisor 8m Procedural Knowledge 1 3.49 1.33 1 5 0.23 0.49 0.55 0.22 0.16 8n Sup Proc Knowledge 2 3.55 1.22 1 5 0.26 0.45 0.63 0.16 0.18 8o Sup Proc Knowledge 3 2.99 1.32 1 5 0.21 0.48 0.55 0.08 0.04 11b Task Routineness 1 1.98 0.99 1 5 -0.25 -0.12: -0.11 -0.22 0.01 11f Task Routineness 2 2.33 1.06 1 5 -0.21 -0.11 -0.12 -0.18 -0.01 111 Task Routineness 3 3.71 1.1 1 5 0.07 0.11 0.17 0.15 0.14 11o Task Routineness 4 3.40 1 1 5 -0.07 0.07 0.02 0.04 0.04 11v Task Routineness 5 2.88 1.09 1 5 0.06 -0.03 0.01 -0.01 -0.09 ENABLING RATIONALE 0.27 0.46 1.00 0.25 0.28 9c Procedural Justice- 3.53 1.11 1 5 0.24 0.18 0.43; 0.24 0.11 Neutralityl 9d Procedural Justice- 3.47 1.07 1 5 0.22 0.23 0.50 0.19 0.08 Neutrality2 9i Procedural Justice- 3.58 1.11 1 5 0.20 0.21 0.51 0.12 0.13 Neutrality3 9f Procedural Justice- 3.65 1.06 1 5 0.22 0.21 0.46 0.22 0.14 Standinql 9g Procedural Justice- 3.82 1.00 1 5 0.24 0.24 0.52 0.23 0.12 Standing2 9h Procedural Justice- 4.14 0.89 1 5 0.17 0.14 0.37 0.18 0.19 Standing3 9a Procedural Justice-Trusti 3.17 1.11 1 5 0.24 0.13 0.38 0.21 0.15 9b Procedural Justice-Trust2 3.62 1.09 1: 5 0.15 0.19 0.41 0.20 0.16 9e Procedural Justice-Trust3 3.74 1.02 1 5 0.18 0.18 0.49 0.16 0.11 8d T rust-Benevo lence 1 3.64 1.38 1 5 0.19 0.34 0.55 0.10 0.15 8h Trust-Benevolence2 3.62 1.26 1 5 0.15 0.33 0.48, 0.07 0.14 8i Trust-Benevolence3 3.63 1.31 1 5 0.13 0.35 0.54 0.03 0.15 8j Trust-Benevolence4 3.59 1.32 1 5 0.18 0.36 0.57 0.09 0.14 8a Trust-Consistencyl 3.89 1.17 1 5 0.15 0.33 0.45 0.10 0.16 8e Trust-Consistency2 3.66 1.34 1 5 0.20 0.36 0.57 0.06 0.15 8f Trust-Consistency3 3.33 1.40 1 5 0.18 0.42 0.60 0.05 0.12 8k Trust-Consistency4 3.80 1.12 r 5 0.12 0.30 0.42; 0.11 0.16 8b Trust-Competency 1 3.81 1.17 1 5 0.19 0.39 0.59 0.08 0.17 8c Trust-Com petency2 3.69 1.19 1 5 0.16 0.40 0.58; 0.07 0.12 8g T rust-Competency3 3.71 1.33 1 : 5 0.16 0.34 0.55 0.05 0.16 8I Trust-Com petency4 4.04 1.13 1: 5 0.17 0.36 0.51 0.07 0.17 Table AD-1: Descriptives for Items External to the Model Heading indicates the corresponding model variables for these extra variables 110 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. Item # Item Mean S.D. Lo Hi Correlations w/Latents MOR CST ER TK Star ORGANIZATIONAL CITIZENSHIP BEHAVIOR 13a1 OCBA-DEPARTMENT 3.82 0.93 1 5 0.07 0.24 0.26 0.07 0.13 13a2 OCBA-SCHOOL 3.48 0.97 1 5 0.12 0.23 0.33 0.12 0.20 13a3 OCBA-SELF 4.33 0.74 1 5 0.23 0.14 0.16 0.27 0.01 13b1 OCB B-D E P ARTM ENT 3.56 1.03 1 5 0.16 0.19 0.22 0.09 0.03 13b2 OC8B-SCHOOL 3.31 1.05 1 5 0.14 0.25 0.31 0.08 0.12 13b3 OCBB-SELF 3.78 1.12 1 5 0.31 0.11 0.13 0.23 0.01 13c1 OCBC-DEPARTMENT 3.69 1.03 1 5 0.06 0.31 0.29 0.08 0.13 13c2 OCBC-SCHOOL 3.38 1.06 1 5 0.08 0.37 0.43 0.08 0.17 13c3 OCBC-SELF 4.25 0.85 1 5 0.18 0.18 0.15 0.20 0.06 13d1 OCBD-DEPARTMENT 3.66 0.98 1 5 0.11 0.28 0.25 0.15 0.10 13d2 OCBD-SCHOOL 3.43 1.02 1 5 0.17 0.32 0.38 0.13 0.20 13d3 OCBD-SELF 4.27 0.84 1 5 0.15 0.22 0.19 0.21 0.05 13e1 OCBE-DEPARTMENT 3.98 1.00 1 5 0.15 0.28 0.29 0.16 0.13 13e2 OCBE-SCHOOL 3.64 1.03 1 5 0.18 0.29 0.41 0.20 0.14 13e3 OCBE-SELF 4.48 0.7 1 5 0.28 0.23 0.13 0.25 0.02 13f1 OCBF-DEPARTMENT 3.18 1.02 1 5 0.11 0.28 0.28 0.05 0.08 13f2 OCBF-SCHOOL 3.05 1.00 1 5 0.07 0.26 0.34 0.01 0.09 13f3 "" OCBF-SELF 3.53 1.14 1 5 0.16 0.14 0.10 0.12 0.04 13g1 OCBG-DEPARTMENT 3.27 1.00 1 5 0.10 0.26 0.26 0.08 0.06 13g2 OCBG-SCHOOL 3.09 0.94 1 5 0.10 0.29 0.38 0.08 0.09 13g3 OCBG-SELF 3.60 1.11 1 5 0.23 0.21 0.15 0.30 0.04 13h1 OCBH-DEPARTMENT 3.27 1.04 1 5 0.14 0.29 0.28 0.06 0.11 13h2 OCBH-SCHOOL 3.20 1.01 1 5 0.14 0.28 0.29 0.03 0.18 13h3 OCBH-SELF 3.62 1.21 1 5 0.23 0.20 0.17 0.20 0.06 13i1 OCBI-DEPARTMENT 3.32 0.97 1 5 0.15 0.30 0.27 0.10 0.07 1312 OCBI-SCHOOL 3.16 0.95 1 5 0.12 0.32 0.33 0.11 0.16 1313 OCBI-SELF 3.81 1.11 1 5 0.24 0.17 0.16 0.20 0.06 13j 1 OCBJ-DEPARTMENT 3.79 0.95 1 5 0.16 0.29 0.21 0.18 0.20 13J2 OCBJ-SCHOOL 3.67 0.96 1 5 0.10 0.28 0.18 0.08 0.24 13j3 OCBJ-SELF 4.05 1.04 1 5 0.19 0.17 0.13 0.28 0.08 13k1 OCBK-DEPARTMENT 3.58 1.08 1 5 0.10 0.15 0.22 -0.05 0.11 13k2 OCBK-SCHOOL 3.39 1.09 1 5 0.15 0.19 0.24 0.00 0.09 13k3 OCBK-SELF 4.36 0.85 1 5 0.21 0.17 0.20 0.15 0.04 1311 OCBL-DEPARTMENT 3.65 1.04 1 5 0.14 0.16 0.26 0.04 0.09 1312 OCBL-SCHOOL 3.50 1.05 1 5 0.15 0.19 0.24 0.01 0.07 1313 OCBL-SELF 4.36 0.85 1 5 0.21 0.17 0.20 0.15 0.15 13m1 OCBM-DEPARTMENT 3.80 1.00 1 5 0.10 0.17 0.25 0.06 0.06 13m2 OCBM-SCHOOL 3.61 1.07 1 5 0.12 0.21 0.28 0.02 0.07 13m3 OCBM-SELF 4.40 0.88 1 5 0.21 0.23 0.20 0.20 0.09 13n1 OCBN-DEPARTMENT 3.77 0.97 1 5 0.15 0.17 0.25 0.08 0.13 13n2 OCBN SCHOOL 3.61 0.98 1 5 0.20 0.19 0.26 0.01 0.12 13n3 OCBN-SELF 4.27 0.89 1 5 0.24 0.19 0.21 0.16 0.05 13o1-4 In-Role Performancel 4.36 0.85 1 5 0.07 0.15 0.12 0.08 0.16 13p1-4 IRP2 3.65 1.04 1 5 0.01 0.03 0.13 0.00 0.13 13q1-4 IRP3 3.50 1.05 1 5 0.06 0.18 0.20 0.04 0.15 Model Heading indicates the corresponding model variables for these extra variables 111 R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. R ole Stress Role Stress was measured using Rizzo, H ouse, and Lirtzman's (1970) role conflict and role am biguity scales. Role conflict is an eight-item scale (a=.82). Role am biguity is a six-item scale (a=.80). Role overload (a=.81) was assessed by com bining items taken from two three-item scales, one from Beehr, Walsh and Taber (1976) and the other from the M ichigan Organization Assessment Questionnaire. Item # Item_______________ Mean_ _ S.D. Lo Hi Correlations w/Latents MOR CST ER TK Star ROLE STRESS 11a Role ambiguity 1 2.24 1.03 1 5 -0.26 -0.30 -0.41 -0.34 -0.11 11e Role ambiguity 2 2.55 1.08 1 5 -0.12 -0.07 -0.01 -0.29 0.10 11k Role ambiguity 3 2.21 1.02 1 5 -0.26 -0.37 -0.35 -0.41 -0.13 11n Role ambiguity 4 2.59 1.11 1 5 -0.30 -0.32 -0.40 -0.36 -0.11 11r Role ambiguity 5 1.75 0.82 1 5 -0.31 -0.21 -0.31 -0.49 -0.15 11x Role ambiguity 6 2.48 1.04 1 5 -0.27 -0.45 -0.40 -0.31 -0.15 11d Role Conflict 1 2.31 1.09 1 5 -0.07 -0.18 -0.32 -0.09 -0.23 11] Role Conflict 2 2.44 1.16 1 5 -0.21 -0.18 -0.32 -0.17 -0.12 111 Role Conflict 3 2.73 1.29 1 5 -0.14 -0.15 -0.28 -0.15 -0.19 11p Role Conflict 4 2.96 1.37 1 5 -0.03 -0.06 -0.14 -0.05 -0.03 11s Role Conflict 5 2.88 1.21 1 5 -0.26 -0.14 -0.38 -0.10 -0.15 11u Role Conflict 6 2.75 1.23 1 5 -0.11 -0.11 -0.22 -0.09 -0.15 11 w Role Conflict 7 2.09 1.08 1 5 -0.24 -0.13 -0.34 -0.16 -0.21 1 iy Role Conflict 8 2.44 1.28 1 5 -0.30 -0.08 -0.20 -0.15 -0.12 11c Role Overload 1 2.74 1.29 1 5 -0.20 -0.09 -0.20 -0.14 -0.09 11 g Role Overload 2 3.23 1.34 1 5 -0.14 -0.14 -0.14 -0.12 -0.05 11 h Role Overload 3 3.37 1.26 1 5 -0.15 -0.06 -0.11 -0.06 -0.04 11m Role Overload 4 2.15 1.01 1 5 -0.17 0.02 -0.13 -0.22 -0.14 1 iq Role Overload 5 3.66 1.23 1 5 -0.03 0.04 0.01 0.03 -0.02 n t Role Overload 6 3.71 1.20 1 5 0.03 0.04 0.03 0.01 0.04 Table AD-2: Descriptives for Role Stress Items R eproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Comparative study of organizational commitment in the public and private sectors: The case of transportation agencies in Thailand
PDF
Business political influence on agency regulatory decision making: The efficacy of corporate political activity in Taiwan's CATV license awarding
PDF
Does adding nonfinancial value drivers to a summary financial measure improve the learning and performance of managers?
PDF
Accountability in public -private partnerships
PDF
Biopharmaceutical strategic alliances: Interorganizational dynamics and factors influencing FDA regulatory outcomes
PDF
Demographic diversity, team process, and team performance: Assessing moderator effects of cognitive conflict management practices and task interdependence
PDF
Androgyny and transformational leadership: Effects of gender and sex -role identity in the collectivistic context of Taiwan, R.O.C.
PDF
California charter schools: Including students with disabilities
PDF
Essential features of federalism: A study of federal evolution in India's Capital Territory
PDF
Board involvement in monitoring and strategy making: Antecedents and consequences
PDF
Building inspectors: Bureaucrats, professionals or heroes?
PDF
Customer satisfaction, customer bargaining power, and financial performance
PDF
Information sharing in work groups: A transactive memory approach
PDF
Corporate ethical orientation and crisis management before and after September 11, 2001
PDF
Information systems capabilities and indirect relational value: Sustaining networked organizations
PDF
A socio-technical approach to support collaborative engineering design
PDF
Essays in supply chain management
PDF
Inventory and pricing models for perishable products
PDF
Defining the advice -seeking request in electronic solicitations in boundary spanning environments
PDF
Banking on the commons: An institutional analysis of groundwater banking programs in California's Central Valley
Asset Metadata
Creator
Cantrell, Steven Mitchel
(author)
Core Title
Feeling good about control: Design considerations for accountability systems in schools
School
Graduate School
Degree
Doctor of Philosophy
Degree Program
Public Administration
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
business administration, management,education, administration,OAI-PMH Harvest
Language
English
Contributor
Digitized by ProQuest
(provenance)
Advisor
Adler, Paul (
committee chair
), Robertson, Peter J. (
committee chair
), Merchant, Kenneth A. (
committee member
), Tang, Shui-Yan (
committee member
)
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c16-73016
Unique identifier
UC11328301
Identifier
3018059.pdf (filename),usctheses-c16-73016 (legacy record id)
Legacy Identifier
3018059.pdf
Dmrecord
73016
Document Type
Dissertation
Rights
Cantrell, Steven Mitchel
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Tags
business administration, management
education, administration