_* P' “'N } LIBRARY Michigan State University \~ 1' PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or betore date due. DATE DUE DATE DUE DATE DUE ___J lel | _,_JL__J[_J .25: l l I RP 03 1933 \ fijgfir THE ODNSTRUCT VALIDATION’OF THE MEASUREMENT OF INDIVIDUAL POWER.INWORGKNIZATICNS By Beth Rubin .A DISSERTATION’ Submitted to Michigan State university in partial fulfillment.of the requirements fer the degree of DOCTOR.OF PHILOSOPHY Department of Psychology 1990 ABSTRACT THE W VALIDATICN OF THE MEASUREMENT (F INDIVIDUAL POWER IN ORGANIZATIQIS By Beth Rubin This dissertation sought to define the construct of individual power in organizations, develop a new measure that avoids previous measurement flaws, and provide enpirical evidence of its construct validity. Literature on power within organizations was reviewed, and a definition of the construct of power ms developed. Power was held to be the ability of a social unit to intentionally influence the decisions, actions or psychological processes of another social unit. It can be held by social units of any size, and stems from both authority and dependence. It is specific to different relationships and content issues. Two limitations of previous measures were identified: the lack of content area specificity, and the lack of relationship specificity. Several recurring antecedent variables were identified, incltxling dependence, authority, and control over critical material and informational resources . 'Ihese constructs were integrated to create theoretical links in a nomological network surrounding the construct of individual power. A new treasure of power was developed. This was a 34-item questionnaire, which tapped the ability of one person to intentionally influence six aspects of another's work life. Questionnaire or archival measures of the other constructs in the network were also developed. Surveys were sent to 1200 randomly selected pairs of employees of a large organization. The power of one individual over the other member of each pair was examined; dyads were selected both across and within hierarchical levels. Measures of most of the constructs in the network were administered to both members of each dyad. The structure, independence and reliability of the measures were examined. .A two-stage factor analysis found a single factor underlying the six different content areas of power; the measure was therefore collapsed into a single index. Measures of authority were found to overlap‘with measures of power and dependence. Measures of other constructs were modified in response to these analyses. The empirical relations among the measures of power, dependence and resource controls were compared with the hypothesized relations among the constructs, via correlational and regression analyses. Three of the feur levels of the hypothesized relationships were generally supported. Copyright by BETH RUBIN 1990 For Dr. Julius Wail, 1902 - 1989, to follow in his footsteps. This dissertation could not have been completed without the help of many peOple, only some of whom can be mentioned here. First, I would liketothankthenenbersofmycormuttee, allofwhomsloggedthrough vast quantities of prose in order to provide me with their insight and assistance. Dr. John Wagner gave me the advantage of his knowledge of the literature, greatly improving the theoretical basis of the dissertation. Dr. Kevin Ford helped me focus on the key aspects of this sprawling project, and construct a research project that could actually be accomplished. Dr. Ralph Levine provided expertise in methodological issues, identifying problems and alternatives and enhaming my understanding of these issues. Their time, effort, knowledge and insight are deeply appreciated. Everyaspectofthisdissertationhasbeentomhedandinprovedby the hand of my advisor, Dr. Steve Kozlowski. He saw this complicated process through from the beginning, at a geographical distance and with little assurance of its oonpletion; his insight and focus kept me on track when my inclination was to wander. Steve also gave me emugh room to create my own project, along with excellent advice which invariably proved to be prophetic, helpful and apt. I learned a great deal about all aspects of organizational research from this project, through the guidance of this extremely knowledgable and professional researcher. Steve, my heartfelt thanks. vi I would also like to thank the many people who gave me assistance withtherranytechnicalaspectsofthisproject. ThanksgotoDr. Keith Groty and Sarah Roepke for their kind assistance in developing the sanple of subjects, and to Will Batsford for his cheerful assistance and expertise in arranging for the written materials. My thanks also to Dane McGregor for his expertise and help in putting together the surveys, and to Mary Van Sell for her advise and support at all stages of this project. And my great appreciation for my crew of envelope stuffers: Carol Baum, Dr. Susan Bordo, Hannie Eng, Sheryl Hanson, Dane McGregor, Dr. Michael Miller, Dr. Eileen Oliver and the Oliver Clan, Dr. Bruce Shefrin and Dr. Michael Schramn. Your paper cuts were suffered for a good cause. I would not have corrpleted this project without the support and encouragment of many good friends; they rejoiced with me through the high points, and encouraged and sustain& me through the low points. To mention just a few, my thanks to Bryan Bates, Dr. Mike Miller, Dr. Bill Miller, Dr. Robyn Muncy, Drs. Bernard and Eileen Oliver, Janet Shortall, Steve Snith, and Dr. Ron Wright. Most especially, my deep appreciation to Dr. Mark Karper and Dr. Cliff Donn, who gave me the emotional and practical support that made the completion of this project possible. I would also like to thank my family for their continuing love and support. 'lbmyMom, Father, Dad, Bernice, StanandBarb: mythanks for your belief in me, for asking the question "Mien?” instead of ”If?“ And yes, I'm finally a Doctor. Andmostof all, mythanksgotoDaneScottMcGregor. Althoughhe has my love for so many other reasons, he has my gratitude for all the vii hours of listening, consoling, shoulder massage and chocolate that saw me through this; for believing that I could complete it; and for reminding me that, as important as this project was, it was not nearly as important as that we built, and are still building, between us. viii TABLE OF CONTENTS I. INTRODUCTION......................................................1 The Construct of Power in Organizations...........................2 The Development of Power in Organizations.........................4 The Need for a valid Operationalization of Individual Power.......7 The Process of Construct validation...............................9 Summary..........................................................11 II. mm OF TIE LITWTUREOOOOOOOOOOOOOOOOOO0.0.0.000000000000000013 Key Issms in m mfiw Of tm LiteratureOOOO0.0.0.0.0000000000013 The Construct of Power...........................................14 Intentional Social Influence.........................1S Held by a Social Unit of Any Size....................16 .A Result of Dependence...............................17 A Result of Authority................................18 The Ability vs. the Use of Influence.................l9 A.Relationship-Specific vs. a General Characteristic............................................20 Power: An Issue-Specific vs. a General Characteristic.......22 WOOOOOOOOO0.000000000000000000I00......0.0.0.000000000023 am A Typology of Operationalizations of Power.......................23 S‘ijxtive msures Of moo...0..0000.......OOOOOOOOOOOOOOZS Objective Measures of Potential Power........................25 ijve Hut-es Of W WOOOOOOOOOOOOOOOO0.0.00.00026 WOOOOOOO0.00.00...0..O0..OOOOOOOOOOOOOOOOOOOOOOOO0.0.0.27 Dependence as a Source ofPower...................................28 W muoIOOOOOOOOO00.0.0000...OOOOOOOOOOOOOOOO0.0.0.0.28 M8011 (1962)I0.0.00...OOOOOIOOOOOOOOOOOOOOOOO0.0.0.000029 JaCObS (1974)00.00.000.000...O...O.CO...0.0.0.00000000000033 Pfeffer & Salancik (1974)................................36 Provan, Beyer & Kruytbosch (1980)........................44 OOOOOOOOOOOOOOOOOO0....0..0..00.....00000000000000000054 Summary cmtml we: aid-cal cmtjmernieSOOOOOOOO0.0.0.00000000000056 (continued) ix TABLE OF (INTENTS (cont'd) Crozier (1964)...........................................S7 Hickson, Hinings, Lee, Schneck & Pennings (1971).........64 mriCR (1981)OOOOOOOOOOOOOOOOO0.0...0.00.00.00.0000000079 .WOO...00......0.0.0.0...OO...0.0...00.0.00000000000000089 Integrations of the Dependence-based Theories................91 WOOOIOOOOOOOOO00......00......I0....0.0.0.000000000000103 Authority as a Source of Power..................................105 weber's conception of Authority and Power...................105 Recent Research Holding Authority as a Source Of WOOD...OOOOOOOOOOOOOOOOOOOI..00...00.0.00000000000109 Summary.....................................................120 W: mmmm 0f MOOOOOOOOOOO...0.0.0.000000000000122 W: mmmmts Of MOOOOOO0.0...00.000.00.00000000139 Summary: Operationalizing Power................................142 Subjective Measures of Power................................l43 Objective Measures of Potential Power.......................145 Objective Measures of Enacted Power.........................l47 General Problems in the Measurement of Power................149 Summary.....................................................151 III. Theoretical Links in a meological Net of Power...............152 W 11‘ tm Mlmical MOOOOOOOOOOO0.0.0000000000000000153 mrIOOOOOOOOOO0......COO...OOOOOOOOOOOOOOOOOOOO00.0.0.0...153 .0.0.00.0.0.0000...O..00....0...0.0.0.000000000000153 Dependence Relative control over Critical Material Resources...........154 Relative control over Critical Informational Resources......156 Amrity.0....OOOOOOOOOOOOOO0.00.00.00.000.0.00000000000000157 Other Antecedents of Power..................................158 WOOOOOOOOOOOOOOOO0.0.00.0.0...0.0...0.0.0.0000000000000158 WOOOOOOOOOOOOOOCOOOOOOOOOOOOOOOOOOOOOO00.00.00.00000000160 Relationships Among the Antecedent Constructs and Power.........161 Direct Links Among Antecedent.Constructs....................l62 Indirect Links Among Antecedent.Cbnstructs..................171 Other Variables that are Related Directly With Poorer........174 WOOOOOOOOOOOOOOOOOOOOOOOOOOO0......0......00.0.00000000000176 Iv. WOOOOOOOO0......0......O.0..O.00....0.0.0.00000000000178 amjxtSOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.0.000000000000178 (continued) TABLE OF CONTENTS (cont'd) Procedure.OOOOOOOOOOOIOOOOOOOO0.0.0.0...O.00......00.0.000000000179 MeasureSOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOIOOOO0.0.0.183 Measuring Power.............................................184 A Topic-Specific Measure of Power.......................184 A.Generic Measure 0f Power...O...O...0.0.000000000000000188 A.aneevaluative Symbol of Power........................189 Measuring Other‘variables in the meological Network........190 Authority.OOOOOOOOOOOOOOOOO0.0.0....00.0.000000000000000190 Dependence.OOOOOOOOOOOOOOOO00.00000...0.00.00.00.0000000192 Relative Control over Critical Material Resources.......193 Relative Control over Critical Informational Resources..195 Tenure and Gender.......................................197 Need for Dominance and Need for Affiliation.............197 Intent to Turnover......................................198 Analyses...IOOOOOOOOOOOOOOOOOOOOO.000......0.0.0.00000000000199 Examination of the Measurement Model....................199 Examination of the Theoretical Model....................200 ‘v. RESULmS...0.000.000.0000.0000000000000000000.0...0.0...00.000000202 Part 1: Establishing the Measurement Model.....................202 Measures of Power...........................................203 Measures of Other variables.................................207 Dependence..............................................207 AuthoritYOOOOO0.0.0.000...00.0..0....0.000.000.000000000209 Relative Control over Critical Resources................211 Part 2: Revisions of the Model and Hypotheses..................217 Part 3: Empirical Evaluation of the Revised Model..............220 Testing the Proposed Relations..............................225 Primary Level Relations: Among Power and Dependence.......................................225 Relation 10......00.000000...0.00.00.00.00.0.0.0....225 Secondary Level Relations: Among Individual Differences and Power................................225 Relation 2a.OOOOOOOOOOODOOOO0.0000000000000000......225 Rslation ZbOOOOOOO0.000.000.00000.0.00.0000000000000225 Balation 2C...OOOOOOOOOOOOOOOOO...0.0.0.000000000000227 Tertiary Level Relation: .Among Resource Controls, Dependence and Power.......................227 Balation 33.00.0000...0....0.0.0.00.0...0.0.00.00000227 Relation 3b....00000000000OOOOOOOOOOOOOOOOOOOOO0.00.227 Relation 3c.........................................227 Balation 3d.0.00.00.000.000...00......0.0.0000000000228 (continued) TABLE OF (IDNI'ENI‘S (cont'd) mlatim 3eOOOOOOOOO0.0...0....0.0.00000000000000000228 mlatim 3f.0.0.000...0.0.00.0...0.00.00.00.00000000228 mlaum 390......0.0.0.000...O...0.0.0.000000000000228 mlatim 3h.0.0000000000000000000000000.0.0.00000000228 SWOO...O0.0CO...O0.0.0....00...00.000.000.000000000228 Feurth.Level Relations: Dependence as a Mediator.......229 mlation “00......0.0.0.0.0...0.00.00.00.0000000000231 Elation 4b.OOOOOOOOIOOCOOOOOIO0.0.0.000000000000000231 mlation “OOOOCOOOOOOOOOOOOOOOOO0.0.00.000000000000231 WOOOOOOOOOOOOOOO00.0.0.0...0.0.0....00.00.000.00000000232 VI. DIwSSImOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO00......0.00.00.00.0234 mum Of mirj-cal RalauOnShiPSOeeeeeeeeeeeeeeoeoeeeeee000000234 ”mid-a]. mat-10:18 Of tile St‘flYOOOOI.OOOOOOOOOOOOOOOOOOO0.00.238 Implications for Theoretical and Measurement Development........245 Implications for Measurement of Constructs Related with Power.......................................24S Implications for Theoretical Development....................248 Implications for Model Development..........................251 Mm1®000000000000OOOOOOOOOOOOOOOOOOOOO0.0.0.00.00.00.0000000000254 AppendixA: AppendixB: Appendix C: Appendix D: Appendix E: Appendix F: Appendix G: Appendix H: Appendix I: Appendix J: Supervisor Power Holder Questionnaire..............254 Supervisor Power Recipient Questionnaire...........258 Subordinate Power Holder Questionnaire.............262 Subordinate Power Recipient Questionnaire..........266 Oo-Wbrker Power Holder Questionnaire...............270 Go-Wbrker Power Recipient Questionnaire............274 A Topic-Specific Measure of Power..................278 A Generic Measure of Power.........................281 A Nth-Evaluative Symbol of Power...................282 meim AumoritYOOOO0.00.0.0....0.00.00.00.00000283 (Continued) xii TABLE CF oomms (cont'd) Apmndix L: Relative Control over Critical ”term WOOOOOOOOOOOOOOOOOOOO0.0.0.00.00000000000000285 Appendix M: Relative Control over Critical Information.........28‘7 Appendix 0: Need for Dominance and Need for Affiliation........289 Appendix P: Intention to Turnover290 mm Q: Retestim 0f figures...O..00.0.0.000000000000000291 LIST OF WOCOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.00.00.00.000000000300 LIST OF TABLES Table Name Page 1 Definitions, MeasuresandPredictors of Power...............123 2 Relations Among Antecedent Constructs in a leical mt Of MrOOOOOOOOOOOOOOOOOOOOOO0.0.0.000000164 3 mmt Of um Comma.OOOOOOOOOOOOOO0.0.0.0000000000185 4 Factor Loadings from Oonfirmatory Factor Analysis of an Mic mum m msure..00......0.0.0.0000000000204 S Inter-Factor Correlations of Topic-Specific Power Measure. . . 205 6 Second-Order Factor Loadings for Topic-Specific Power muteOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.00000000000000000000ZOS 7 Correlations AmongPowerMeasuresZOB 8 Rotated Factor Loadings for Dependence Measures.............210 9 OorrelationsAmongPower, Dependenceand Autl‘lority FacmrSOOOOOOOOO0.000000000000000000000000......212 (continued) Table 10 11 12 13 14 15 16 ‘17 18 Arl A92 A73 A95 LIST OF TABLES (Continued) Name Page Rotated Secondrorder Factor Loadings for Power, Dependence and Authority Factors..........................213 Rotated Factor Loadings for Resource control Items..........215 Cronbach's Alpha Coefficients of all mluplyitan ms‘lreSOOO0.00000000000000000000.000.000.00216 Revised Hypothesized Relations Among Antecedent Constructs in a Nomological Net of Power..................219 List Of Relatiom mm mas‘lreSOCOOOOOOOOO0.0.0.0.000000000221 Agreement Across Perceivers on All Measures.................224 Correlations Among Power and Related variables..............226 Hierarchical Regression of Resource Controls and mm on mt..0...OOOOOOOOOOOOOOOOOOOOOOO0.0.0.0...230 Hierarchical Regression of Individual Resource mntrols 3rd mme on MOOOOOOOOOOOOIOO00.0.0.0...230 Reliability Analaysis of Topic-Specific Power me ”WSt.OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO0.00.292 Reliability Analysis of Perceived Authority We Prrtest.0.00.00000000000000000000000.0.00.00.00.0295 Reliability Analysis of Perceived Dependence mm mmst.0.00.00...OOOOOOOOOOOOOOOOOOOOOOO00.0.000295 Reliability Analysis of the Relative Control over Critical Material Resources Measure Pre-test.........297 Reliability Analysis of the Relative control over Critical Informational Resources Measure Pre-test.........297 LIST OF FIGURES Eggs; Name Fae 1 A W1” Of m MSWSOCOOOOOOOOO00.0.0000000000000000024 2 DirectLimcagesAmongAntecedentsina Mlmcal mt Of MOOOOOOOOOOOOOOOOOOO00.0.00000000000163 3 Revised Linkages Among Antecedents in a MIMI mt Of MOOOOOOOOOOOOOOOOOOOO0.0...00.00.00.218 I . INTK'JDIXII‘ICN This dissertation defines the construct of individual power in organizations, and describes the construct validation of its measurement. First, it reviews the literature on power within organizations, focusing on research that sought to identify the sources of power. Problems in the existing Operationalizations of the construct are also identified in this review. From this review, a definition of the construct is developed, requirements for its assessment are specified, and recurring antecedent constructs are identified. Next, the antecedent variables are integrated, to create theoretical links in a nomological network surrounding the construct of individual power. A new measure of power is presented, and the construct validity of this measure is examined by operationalizing the other variables in the nomological net, and comparing the relations among these operationalizations to their theoretical relationships . This section introduces the construct of power in organizations, and provides an overview of the conceptual and measurement issues that will be explored in this dissertation. ~The concept of power is described, and the major theories of its etiology are introduced. The need for a valid Operationalization of power at the individual level is described, and the requirements of construct validation are delineated. The Construct of Power in Organizations Manyresearchersandtheoristshaveexaminedtheconceptofpover in organizations. The approaches have ranged from macro to macro analyses, and have spanrned thirty years (Enerson, 1962; McClelland, 1969: Zald, 1970: Pfeffer, 1981: Blair, Roberts & McKechnie, 1985; Rose Ragins & Sundstrom, 1989). Although researchers from different perspectives have used different definitions of power (Astley and Sachdeva, 1984: Allen & Porter, 1983: Bacharach & Lawler, 1980: Provan, Beyer & Kruytbosch, 1980: Pfeffer, 1981), there is a degree of definitional consensus within the organizational behavior literature. This study will be limited to research that takes a structuralist perspective on organizational power . The literature that approaches the topic from social constructionist or collective bargaining viewpoints will not be examined. Allen and Porter's (1983) compendium of key articles on the topic held that most authors ”view power as an ability, capacity or potential for influence, arnd influence as a process of producing behavioral and psychological...effects" (p. 2). Pfeffer's (1981) comprehensive review holds that "it is generally agreed that power characterizes relationships among social actors” (p. 3). Power is usually viewed as intentional influence, and as a characteristic of social actors at any level of analysis: it can be held by individuals, groups, departments of entire organizations. Despite the consistency in the core of power definitions, the boundaries of the power construct are not precisely set. Many articles onthetopicdonotdefiretheconstructwhentheyconductresearchon the causes of power: the definition and boundaries of the construct must be inferred by the measures used to operationalize it (e.g., Salancik & Pfeffer, 1974) or from the models developed to predict it (e.g., Blackburn, 1981). Several authors have noted that most people believe that they know'what power is, until they are asked to define it (Bierstedt, 1950: Allen a Porter, 1983). unfortunately, this leads to divergent views of its content and limitations. Bacharach and.Lawler (1980) describe power as "primitive" term (p. 14), which acts primarily as an heuristic to focus attention on a phenomenon. Other researchers who acknowledge the lack of agreement on construct definition are not quite as pessimistic (Pfeffer, 1981: Allen & Porter, 1983: Provan, Beyer & Kruytbosch, 1980). Key differences in construct definitions include the requirement for the successful use of power: some researchers hold that power must be effectively used to change behavior in order for it to exist (c.f., Pfeffer & Salancik, 1974), while others hold that it does not require this. Another definitional issue is whether power is a characteristic of a relationship, and hence differs across different relationships, or whether it is a personal or subunit characteristic that is the same in all relationships. A third inconsistency is whether power is different across different types of behaviors and issues, or whether a social actor has a given amount of power across all issues he or she might desire to influence. Different definitions tend to be prevalent within different academic approaches to the concept of power. The sociological perspective generally involves the view that originated with.Weber (1946), and was adapted and expanded by Dahl (1957): power is only power when it is successfully used to affect behavior, and if it changes theoutcome fromnwhatwuild haveoccurredhad itnotbeen used. The social psychological perspective (c.f., Enerson, 1962: Allen & Porter, 1983) tennis to include both behavioral and psychological changesasthevalidtargetsofpower, anddoesnotrequirethe intention to influence in order for power to exist. Industrial relations analyses (c.f., Bacharach & Iawler, 1980) tend to focus on power as it relates to the escalation and resolution of conflict between parties: their definitions generally distinguish the ability to influence another from the use of that ability, and involve analyses of the costs of using power in specific situations. Organizational researchers have used a variety of definitions , with their different construct boundaries. However, there is often a distinction between the theoretical definition of the construct and its measurement. Many researchers have taken a specific theoretical approach.and interpreted their findings in that light, but used measures that tapped a different construct. Therefore, a systematic review of the empirical and theoretical literature on organizational power may identify common characteristics and boundaries. The Development of Power in Organizations Different authors have proposed a wide array of variables to be related to power: these include different sources of power (Bacharach & Iawler, 1980: Morgan, 1986: Hickson, Hinings, Lee, Schneck & Pennings, 1971; Kanter, 1979; Ketter, 1977; Salancik & Pfeffer, 1974; Hambrick, 1981: Blau & Alba, 1982: Enerson, 1964: Mechanic, 1963: Blackburn, 1981) , bases of power (French 8. Raven, 1959: Bacharach & Iawler, 1980) . variables that moderate the use arnd acquisition of power (Kipnis, 1976: Wilkinson & Kipnis, 1978: McClelland & Burnham, 1976: McClelland & Watson, 1973: Mowday, 1978: Pfeffer, 1981; House, 1988; Rose Ragins & Sundstrom, 1989) , tactics that are used by power holders (Kipnis, Schmidt & Wilkinson, 1980; Dunne, Stahl & Melhart, 1978; Schilit & Locke, 1982), effects of using power (Bachman, 1968: Bachman, Smith & Slesinger, 1966; Burke & Wilcox, 1971; Busch, 1980: Cope, 1972: Lord, 1977: Martin & Hunt, 1980: Shetty, 1978: Slocum, 1970), arnd so on. A substantial portion of this literature deals with how people deve10p power. Much of this work adapts corcepts first developed in theoretical and empirical research performed at larger levels of aggregation (Hickson, Hinings, Lee, Schrneck & Pennings, 1971: Hinings, Hickson, Pennings & Schneck, 1974; Pfeffer s. Salancik, 1974; Salarcik & Pfeffer, 1974: Provan, Beyer, & Kruytbosch, 1980: Pfeffer, 1981). Several researchers have proffered theoretical processes to explain how individuals develop power (MeChanic, 1962: Kipnis , 1972; Blackburn, 1981: Pfeffer, 1981: Manz & Gioia, 1983; Jones, 1984: House, 1989), and in recent years, researchers have begun empirically testing these theories (Forrbrun, 1983; Haltbrick, 1981; Blau arnd Alba, 1982). thpredominantviewoftheresearchonthesourcesof powerholds that power stems from dependence upon another (Emerson, 1962) . Two different approaches have been taken within this literature; the first focused on the ability to cOpe with critical contingencies as a cause of dependence, and the second focused on controlling material resources as a cause of dependence. Within the last ten years, these perspectives have been integrated by a number of researchers . The first of these two dependence-oriented views holds that organizations are essentially decision—making bodies (Thompson, 1967). Problems that stop the flow of work nust be dealt with: unplanned, unforseeable problems, or critical contingencies, require information toresolvethem. Peopleorgroupsdependuponthosethathave sufficient information to deal with these problems; therefore, those social units will be powerful (Hickson, Hinings, Iee, Schneck & Pennings, 1971). The second of these views focuses on the material factors that are required for organizational operation, such as money and services purchased with it. People or groups depend upon those who control the flow of material resources that they need to do their work; therefore, those social units will be powerful (Salancik & Pfeffer, 1974). Integrations of these approaches view both the ability to deal with critical contingencies and the control over material resources as differext means of producing dependelce; both are resources that will produce dependence if they are critical for the performance of work. Being dependent upon another gives that person the ability to influexce decisions and behavior. A second perspective on the development of power in organizations focuses on authority as a key source of individual power. Weber's (1946) classic analysis of modern organizations held that rational- legal authority is the predominant basis of the control over behavior in organizations: this has been empirically supported (Tannexbaun, 1968) . Recent researchers have formally presented authority as a source of power which is related to, but Operates independently of, other sources (Astley 8. Sachdeva, 1984: Bacharach & Iawler, 1980: Hanbrick, 1981; WI, 1978). For exauple, Allen and Porter (1983) held that organizational superiors hold more power than their subordinates due to their formal authority, which provides them with the ability to demand compliance with influence attempts ”in addition to” their ability to control rewards and punishments (p. 4). The Need for a Valid Operationalization of Individual Power Salancik and Pfeffer (1974) noted that "power is a concept with more intuitive appeal than empirical precision" (p. 456) . The problem of measuring the construct is highlighted by the recent increase in enpirical analyses of individual power; every sandy operationalized the construct differently. Some studies used control graphs (Tamnenbaum, 1968; Tannenbaum & Kahn, 1957) or influence networks (Brass, 1985: Panbrun, 1981) , while others used questionnaire indices of perceptions (Blau and Alba, 1982) or standardized interview questions (Hanbrick, 1981) . Sane measures tap power over others in general (e.g., “how much influencedoesXhaveovervtnatgoesonaroundhere"), whileothers were topic-specific, tapping influence over decisions in specific areas of work (Blau & Alba, 1982; Hanbrick, 1981). This inconsistency raises the issue of external validity of findings. 'Ihe inconsistency in Operationalization clearly reflects the inadequacy of the definition of the construct. In order to apply firndings from each individual study to a general knowledge base, both tile construct's definition arnd its measurement should be standardized: at the very least, the two must be consistent within each study. The difficulty in measuring power has been described in research that focused on the power of subunits or entire organizations (Provan, Beyer & Kruytbosch, 1980: Pfeffer & Salancik, 1974; Hinings, Hickson, Pennings S. Schneck, 1974). These operationalizations ranged from the allocation of budgets (Provan, Beyer & Kruytbosch, 1980: Salarncik & Pfeffer, 1974) to individual perceptions of the amount of power held by a group (Hinings, Hickson, Pennings & Schneck, 1974; Pfeffer & Salancik, 1974). Studies involving the measurement of individual power have generally followed the techniques developed in the subunit studies, using sirngle item measures (Bachman, 1968; Bachman, Smith & Slesinger, 1966; Fonbrun, 1983; Mowday, 1978) or parallel measures assessing the annunt of power held over different decision areas; these areas were different for every study (Blau & Alba, 1982; Hanbrick, 1981) . Despite researchers who hold that power is issue-specific, there are no consistent ildices that tap power over different issues, either on the imividual or subunit level. In addition to this problem, although many researchers specify that Mr is theoretically relationship specific, most of them empirically r"'easurethepowerofanin'ndividual orgroupoverallothersinthe depertinent or organization. There are few relationship-specific thefigures of subunit power, ard no existing relationship-specific "easures of individual power. Thus, even when the construct specifies dis-unions across power relations, the enpirical results can not "game them. This lack of consistency in measurement reflects and exacerbates the lack of consistency in construct definition. The insensitivity of existing measures to central aspects of the construct indicate that some empirical analyses may have questionable validity due to inadequate constant validity of measures. Thus, there is a clear need for measures of individual power that accurately reflect the constant as it is defined; in particular, the relationship-specific and issue- specific nature of the constant have not been tapped. The Process of Constant Validation According to Cronbach and Meehl (1955), "constant validation takes place when an investigator believes that his instrument reflects a particular constant, to which are attached certain meanings. The prOpOsed interpretation generates smcific testable hypotheses , which are a means of confirming or disconfirming the claim" (p. 65) . The first step in construct validation is thus to specify the construct, and examine the extent to which the proposed measures actually reflect its inportant aspects. Mere intercorrelations among measures do not erAsure valid measurement, if all of the measures fail to tap key chalacteristics of tine constant itself. Once this has been done, empirical tests of the resulting hl’thctheses may provide evidence of the construct validity of the rlieua_8ures. In order to develop these hypotheses, Cronbach and Meehl r W the development of a "mmological network", which is an I. j‘t‘al‘lerlocking system of laws which constitute a theory" (p. 65) . 10 The laws, or rules of existence of a nomological network, may specify the relations among theoretical constructs, observable factors , or between observable factors and theoretical constants. A nomological nnet mnust include some observable factors in order to be validated, but it is nnot required that the same operation be used in order to measure a single construct; ”if their positions in the nomological net tie them to the same constant variable“ (p. 66) , we may accept that two distinct procedures measure the same thing. The constant validation of the measurement of power may therefore be accomplished by the careful specification of the construct, ensuring consistency between key aspects of the construct and its measures, and tile examination of a nomological network surrounding the construct. T1113 requires the specification of theoretical concepts , links between true theoretical corncepts, links between observable factors, and links between the observables and the theoreticals. This paper specifies theoretical concepts that relate with individual power in organizations, and develops theoretical links in a Mlogical network surrounding the construct of power. It creates links between the theoretical constants and Specific observable factors, assesses them empirically, and empirically examines the links a""Qmag the observables. Because of the amount of research and the consistency in the key constructs that have been used to explain the development of power, m literature is used to deve10p the theoretical aspects of the l”tht'r-Qlogical net surrounding power. This requires a review of the 1 j‘har-ature, and the specification of constants and their relationships ‘1') power and one another. The next step is to specify the relations 11 between these constants and observable variables; this is done by operationalizinng all constants in the nomological net. The last step is to examine the relations amonng the observed variables. Smnmary This section introduced the concept of power in organizations, and introduced the connsistent and inconsistent aspects in the definition of true constant. It then briefly outlined the major theories of the It identified the need to etiology of the power of individuals . develop a valid Operationalization of individual power, and described The steps taken to specify the tt'ie process of construct validation. limits of the construct of individual power, and to assess the constant validity of its measurement, were described. Chapter II contains a review of the research on the etiology of the power in organizations. This review focuses on several issues: definition of the constant of power; the measurement of the construct of power; and the identification of consistent sources of power. The literature is organized into the major perspectives taken to the caligation of organizational power. The research is summarized within e'a|"~'=lnoftl'nethreekeyissues: theconstructandits limitsare SE>e-cified, problems in prior measures are identified, and several direct and indirect sources of power are identified. Chapter III develops the theoretical links of the nomological net M individual power in organizations. It integrates the sources of N identified in the literature review, along with two other v. a*‘1‘s‘1ables which previous research has found to relate to power and its 12 sources, into the theoretical aspects of a nomological nnet around individual power in organizations. The constructs in the net, and the relations among all of these constants, are specified. Three individual characteristics which have previously been differentially related with power are also described. Specific hypotheses of the size and direction of the relationships among all the constants are presented. Chapter IV presents the means by which observable data were collected. The subject pepulation is described, and measures of all constants are introduced. The theoretical fidelity of these measures to the constants under examination is described, and the annalytic Strategy used to provide empirical support for the constant validity Of the measure of power is presented. Chapter V presents the results Of data analysis, and Chapter VI discusses the implications of these results. Conclusions are drawn regarding the measurement and sources of individual power in organizations , and directions for future researchare discussed. II. REVIEW OF THE LITERATURE This section reviews the literature on the etiology of individual power in organizations. It begins with a synopsis of the major issues that will be addressed in this chapter, a description of the key elements of the definition of power, and the introduction of a typology Of its measurement that will be used in the review. The research that taas focused on the etiology of power is then reviewed, and the definitions and measures of the construct of power in organizations are described. This results in the specification of the boundaries of the power construct. Two direct sources of individual power in organizations, and two indirect sources, are identified. Last, problems in the existing measures of individual power are delineated. Key Issues in the Review of the Literature This review of the literature on the etiology of power focuses on t'1'1-‘l':ee key issues. First and most important, the construct of iridividual power in organizationns is carefully defined. Second, the review identifies and describes causal constructs that have consistently been related to power in organizationns, and integrates the N11 where the sources are essentially identical. The review is o I~§enized according to the basic theories of the causes of power; the f - Ms are summarized within each approach, as well as across them. 13 14 The last key issue addressed in this chapter is the Operationalization of the construct of individual power in organizations. The inconsistency in the approaches taken to measurement, and the fact that key aspects of the construct have not been represented in its measures, both lead to the questionable validity of those measures. Therefore, the measurement techniques used in the research reviewed, their problems and advantages are discussed. In order to facilitate these analyses, the basic perspective taken to the construct of power is described prior to the review of the 1 iterature. The definitional aspects that are widely accepted are described, followed by the basic points of inconsistency. A typology of the different approaches taken to the measurement of power is then presented. The Construct of Power 'I‘hereareseveralaspectsofttcconstructofpoweronwhichthere is general agreement among researchers. Power involves the ability of It may be held by Glue social unit to intentionally influence another. a social unit of any size, and is closely related to dependence and away. There are also several issues in the definition of the construct on Wehthereisdisagreement, thatwillbeexamunedinthisreview. F.3~t‘st is the distinction between power as the ability to influence, as 0% to power as the actual successful use of influence. Second is N issue of relationship-specificity: power as a characteristic of Ex social unit's relationships with other social units, versus power 15 as a general characteristic of a social actor; in the latter view, the social actor has the same amount of power over all others. Third is the concept of issue-specificity. Power may be specific to a given area or issue, which may be broad in nature (e.g., influencing how an employee behaves at work) or may be limited to only one, confined area (e.g., influencing how an employee answers telephones) . The alternative view is that a social unit has a given amount of power, which may be used to influence any and all behaviors. Power: Intentional Social Influence Virtually all organizational power researchers describe power as, or in terms of, influence. Power has been defined as "influence over peeple, processes, and/or things" (McCall, 1979, p. 187). Dahl (1957) defined power as a relationship among social actors in which one social actor can get another to do something that he/ she would not otherwise liave done. Weber defined power as "the probability that one actor thhin a social relationship will be in a position to carry out his own “11 despite resistance, regardless of the basis on which this probability rests” (Cf. Dornbusch and Scott, 1975, p. 31). ’lb Etzioni, Donner was "an actor's ability to induce or influence another actor to carryouthisdirectivesoranyothernormshesupports" (Etzioni, 1961; cf. Deseler, 1986, p. 292). However, social influence is a much broader construct than power. PO): example, it includes the ”mere presence“ effect of social f"E‘Qilitation, where nothing more than the presence of another person cé‘tl increase the drive level, and through that the performance level, 0 E a person (Zajonc, 1965) . Similarly, it includes the phen1omenon of 16 conformity: tre impact of an individual or group upon the beliefs, attitudes or behaviors of another. This includes both informational social influence, as was found in Asch's classic experiment on the perception of lines (1951) , and normative social influence, such as the innoactofgroupnormsontheattitwdesof tnosewnoaspireto menbership (Neweomb, 1943). These types of influence are not always intentional, but nonetl'eless, in trese situations the behavior and attitudes of peOple are changed by others. Power is generally viewed as social influence that is intended (Crozier, 1964; Hinings, Hickson, lee, Pennings & Schneck, 1971: Jacobs, 1974; Mowday, 1978; Hambrick, 1981; Pfeffer, 1981: Schilit s. Locke, 1982) . Power is thus a one particular type of social influence; it is a subset of the construct of social influence. This conceptual distinction is reflected in the wording of the operationalizations of power. Although power is often measured with questionnaires that ask about "influence," the phrasing calls for the respondent to describe influence that is knowingly used, by phrases such as "how much say do you have" and "I comply with my supervisor." Thus, altnough the distinction between power and influence is not always specified, it is implicit in the Operational izations and the specific type of influence that is described as power. Power: Held by a Social Unit of Any Size Tiereiswidespreadaoceptanceoftheviewthatpowercanbeteld b? a social unit of any size, including individuals, groups, divisions or entire organizations. One may describe the power of tne marketing d‘Epmanrtmwentovertheaccountingdepartment, orthepowerofanyone 17 division president over her peers. A social unit of any size may told power over any other social unit at any other level of analysis, and each relationship will be characterized by different amounts of power. For example, the private secretary to the president of the marketing division of a large corporation has a given amount of power over the executive for whom he works: he has a different amount of power over tle other private secretaries for tne other division heads: and he has yet a different amount of power over the entire division as a whole, due to his role as a gatekeeper to the president. Researchers have developed theories that were specifically postulated to hold across levels (Weber, 1968: Hambrick, 1981: Crozier, 1964) . Many researchers borrowed causal processes that were developed at one level, and applied them to tre causation of power at another level (Blackburn, 1981: Provan, Beyer & Kruytbosch, 1980: Astley & Sachdeva, 1984: Manz & Gioia, 1983): this multi-level approach has held for both theoretical and empirical analyses. This stwdy will focus on the power of individuals: it will examine the power held by individuals over other individuals. Much of the literature on the etiology of power has focused on larger levels of analysis: this literature will be included in the literature review, and, where reasonable, applied to the power of individuals. Power: A Result of Dependence A significant amount of the literature on the etiology of power in Oli‘ganizations tolds that it stems from dependence (Pierson, 1962: crazier, 1964: Pfeffer & Salancik, 1974; Jacobs, 1974: Hickson, Hiriings, Lee, Schneck 8 Pennings, 1971; Kotter, 1977: Mechanic, 1962: 18 Blackburn, 1981: Bacharach & Iawler, 1980; Provan, Beyer & Kruytbosch, 1980: Blau & Alba, 1982: Manz & Gioia, 1983: Fombrun, 1983; Astley & Sachdeva, 1984). An organization that depends upon the united way Agency in order tolget its working budget allocation.will be easily influenced by the Agency: the Agency will have power over it. ILikewise, a manager who depends upon her secretary in order to keep her office in order will be influenced by the secretary: the latter will have power over her. Dependence is generally conceptualized as needing or desiring an outcome that another social unit controls: the outcome may be psychological, such as being held in high esteem, physical, such as awarding a raise or granting funds for a special project, or informational, such as predicting changes in the market demand.for a product. It is increased if there are no other sources of the outcome, and if the outcome is highly valued or necessary for the accomplishment of‘work. Power: A Result of Authority Authority is one means by which people acquire the ability to intentionally influence others, which is particularly relevant in organizational settings. It is generally viewed as the ability to direct.behavior, and expect.compliance with such.direction. Many leesearchers have viewed authority as synonymous with formal hiaerarchical position; others specify that authority itself is a belief ill the appropriateness of following directives, which is engendered by hieeerarchical subordination. Several researchers have held that 19 auttority is a distinct source of power (Bacharach & Iawler, 1980: Jones, 1980: Fonnbrun, 1983: Hambrick, 1981; Astley a. Sactdeva, 1984). For example, a supervisor may direct her subordinate to do perform her work in a particular manner, due to the formal autlority she holds; a co-worker will not be able to direct such behavior, because he does not hold the authority to direct compliance with such requests. Power: The Ability vs. the Use of Influence Power has been defined as "the ability to get things done tne way one wants them to be done” (Salancik & Pfeffer, 1977, p. 4), and ”the capacity to effect (or affect) organizational outcomes" (Mintzberg, 1983, p. 4). These views specify that power is an ability, which "in most definitions of the concept,... does not have to be exercised to exist. Wrong (1968) emphasized that altl'ough power is usually defined as the capacity to control others, there is a distinct difference between the capacity to control and actual performance of control” (Provan, Beyer & Kruytbosch, 1980, p. 207). The alternative view of power is that "unused potential is not power, because power implies successful use of the potential” (Bacharach & Iawler, 1980, p. 12); this is the view taken by Dahl, based upon Weber's definition of power as the probability that a social actor can carry out their will, despite any resistance from others that might be offered. Kipnis described his book on those who hold power as being "about people who are trying to influence others" (Kipnis, 1976, P- I): the basic approach was to describe situations in which different influence tactics would be used, and the successive use of more meg-cive tactics until compliance was achieved. 20 This conceptualization of power as successful use of influence is evident in many measures of the construct. I-Iinings, Hickson, Pennings andSchneckmeasuredperceivedpowerastreamountof influenceeach department had on problems in various areas (1974, p. 44). Hambrick measured perceived power as the amount of influence each of several people generally had in affecting the outcomes of four Specific types of decisions (1981, p. 275) . Blau and Alba measured power as "tle amount of influence exercised by a staff member" in specific, central areas (1982, p. 367) . Fombrun measured power by asking employees whom trey went to when they wanted to influence a major decision (1983) . These definitions and Operationalizations all involve using the ability to influence another. Theoretically, however, many of the researchers using these methods defined power as the ability to influence others. For example, Pfeffer (1981) specified that power did not require the use of influence, by holding that power is "a force, a store of potential influence through which events can be affected[:] politics involves tlose activities or behaviors through which power is developed and used in organizational settings." (Pfeffer, 1981, p. 7). Thus, there is disagreement both amonng studies and within stndies onwhettcrtheconstrnctofpowerreqtdresthesuccessfuluseof influence, or whether power is limited to the ability to influence, regardless of the decision to use it and regardless of its success when uSed. Power: A Relationship-Specific vs. a General Characteristic T'he early work of researchers of power in organizations focused on the power of an organizational actor over all others in the system — 21 (Weber, 1968; Crozier, 1964) . The development of the concept of charismaasasourceofautlorityandponerwasbasedontnenotion that this individual characteristic gave one greater ability to influence or direct all others (Weber, 1968; Mechannic, 1962: Blackburn, 1981) Most researchers, however, describe power as a variable that is mm by a social unit, but characterizes relationships among units (Pfeffer, 1981; Allen a Porter, 1983: Hickson, Hinings, Lee, Schneck & Pennings, 1971) . This means that it is not a generic characteristic of a social unit, which is the same in all relationships and all situations. Rather, each social unit has a different amount of power over each other social unit with whom she/he has a relationship. For example, a secretary may have a great deal of power over the specific managers for whom he works, but have less power over other secretaries at tle same level; the former set of individuals depend upon him in order to get their work done, while the latter do not med him for tl'eir performance. Despitethecotmonacceptanceofthisaspectoftneconstruct, the overwhelming majority of measures of power do not differentiate among tle power existent in different relationships. Most perceptual measuresinquiredabouttleamountof influenceleldbyanindividual or department: "over whom“ was not specified. Some measures asked respondents to list the most powerful individuals in a department: the 001mm of one person having different amounnts of power over different c><>--workers was not considered. Some measures examined tl'e outcomes of ountested decisions , and measured the power of organizational actors by identifying who got tne most assignments of critical resources: again, 22 trey measured who had tne most power relative to all otters, but not relative to any other Specific social actor. Therefore, on this issue there is inconsistency within as well as between research analyses of power. The primary inconsistency on this aspect lies between tre operationalizations of tie construct and the definitions of the construct; nearly all of the tests of tne causes of power were based on non-relationship specific measures. Power: Ann Issue-Specific vs. a General Characteristic Some power researchers view power as a general characteristic of a relationnship. In this view, the ability to influence behavior is considered to exist across issues: whatever is desired by the power holder will be done (Bacharach & Iawler, 1980; Crozier, 1964) . Several other researchers view the construct as being issue or topic specific (Hannorick, 1981: Hinings, Hickson, Pennings & Schneck, 1974: Pfeffer, 1981: Salancik & Pfeffer, 1974: Bacharach & Iawler, 1980) . This means that a social unit may have more power over decisionns or behavior in one area or issue, and less power over behavior or decisions in another. For example, a supervisor may have power over one aspect of a subordinate's work life, such as the scneduling of his/her work, and have little power over tne individuals With whom he/sl'e socializes at work. Some researchers even connsider the number of different issues over which a social actor has power as an indicator of the amounnt of power that they hold (Hinings, Hickson, Patnnings & Schneck, 1974) . 23 However, measures of power in organizations generally focus on one or two specific issues: if multiple issues are considered, they are generally all of one type, such as the ability to make decisions over different content areas . Thus , even if the theoretical definition of tie construct holds that power may vary across issues, most of the empirical testing of its relationships does not include that characteristic. Therefore, there is inconsistency both within and across studies . m The preceeding section identified several aspects of the construct of power in organizations. Some of these were described as being widely accepted among researchers, and generally descriptive of the construct. Other aspects , however , varied across researchers and/ or within research studies. These ideas will be the targets of focus in the review of the literature: both the theoretical definitions and the operational definitions of the construct will be examined in an attempt to eliminate tne ambiguity and produce a single, consistent definition of individual peer in organizationns . A Typology of Operationalizations of Peer There have been many different approaches taken to the measurement ofhowmmchpeerasocial unithas. Intneircomparisonof measures of peer, Provan et al. (1980) described the "capacity to control Others" and he "actual performance of tne control of otters” as Calceptually distinct constructs (p. 207) . This distinction is 24 reflected in the two basic approaches to measuring power: tre attempt tomeasure "enacted” power, orpeerthathasbeenusedtoachieve influence: and the assessment of "potential" peer, or tle amount of the ability to influence others that is held by a social unit (Wrong, 1968: Provan, Beyer & Kruytbosch, 1980). These two concepts have each been Operationalized subjectively, by qnestionnaires or interviews, or objectively, such as by hierarchical level or distribution of funds among subgroups. The relationship between tl'e conceptual and measurement approaches was modeled by Provan, Beyer & Kruytbosch (1980): an adaptation of this model is presented in Figure l. Conceptual ization Potential Power Enacted Peer . I Eg. , multiple-item questionnaires subjective | l Type of Measurement Eg. , University Eg. , distribution of objective dept. representation funds to different on key committees departments (adapted from Provan, Beyer & Icruytbosch, 1980, p. 208) Figure 1. A Typology of Peer Measures. 25 Subjective Measures of Power Subjective measures of power tap into how'much power a social unit is perceived by themselves or others as having. This is an amalgam of power previously enacted (and perceived by the self or others) and the perception of the ability to influence: thus, perceptual measures are seen as assessing both enacted and potential power (Pfeffer, 1981: Provan, et al., 1980). This approadh directly asks organizational members whom they perceive to have power or influence, or how much influence a social unit has. The technique has been widely used (Perrow, 1970: Hinings, Hickson, Pennings & Schneck, 1974: Salancik & Pfeffer, 1974: Pfeffer & Salancik, 1974; Bacharach & Aiken, 1976: Provan, et al., 1980: Hambrick, 1981: Mowday,1978: Blau & Alba, 1982: finlkinson & Kipnis, 1978; Bachman, Smith & Slesinger, 1966: Fombrun, 1983; Smith & Tannenbaum, 1963: Roussell, 1974). Most of these perceptual measures use‘Likertrtype ratings, although occasionally rankings of the various social actors in the organization are used (Hinings, et al., 1974). These items have been used with both structured interview and written questionnaire formats. Objective Measures of Potential Power Objective measures that aim at identifying potential power include factors such as university departments' representation on key committees (Salancik & Pfeffer, 1974: Hills & Mahoney, 1978), level in the organizational hierarchy (Bacharach & Aiken, 1976: Hinings, et al., 1974), formal authority for deciding specific issues (Blau & 26 Schoenberr, 1971: Hinings, et al., 1974; Blau & Alba, 1982), amount of financial dependence, and national reputation (Pfeffer & Leong, 1977; Provan, et al., 1980). These are used as indices of the ability to influence others, rather than as indices of influence that has already been used. Most«of these measures attempt to tap formal authority, dependence, or potential causes of dependence such as control over the flow'of key resources. Objective Measures of Enacted Power Measures of enacted power examine the consequences of power use "as these become manifest in decisions made within the organization. Presumably, power is used to affect choices made within social systems. Then, one way of assessing the distribution of power is to see which social actors benefit, and to what extent in contested decisions within organizations" (Pfeffer, 1981, p. 49). This measurement approach is most consistent.with the view of power that requires its successful use. This technique assumes that power which is held has been used, and that decisions therefore reflect the power of the various decision! making entities. It allows for the use of archival data as indices of power. These measures include university departments' representation on key committees (Pfeffer, & Salancik, 1974) and the distribution of funds to various organizations or departments (Provan, et al., 1980: Pfeffer & Salancik, 1984). As is clear, several of the same indices have been used as both objective measures of enacted power and objective measures of potential power. 27 Sum—Marl The preceeding sections presented a brief description of the construct.of individual power in organizations. Power is viewed as intentional social influence; it stems from authority and dependence, and may be held by a social actor at any level of analysis. It may be an ability or demonstrated influence; it may be specific to different relationships or general influence within a system, and it may be specific to different issues, or general influence across issues. The basic approaches to measuring power have focused on either measuring the outcomes of its use, or measuring its potential for use. Both objective and subjective indices have been used. Subjective indices are viewed as assessing both previous use of power and potential power for future use, while objective measures are generally viewed as representing either past use or potential use; the objective indices themselves, however, overlap. The next sections will review the research on the development of power. First the literature that views power as stemming from dependence will be reviewed; this will be examined in two sections, each.taking a different approach to the basis of dependence. This will be followed by a review'of the literature that focuses on authority as a source of power. . The recurring elements of these views will be identified, and a detailed description of the construct of individual power in organizations will be delineated. Last, limitations of the existing operationalizations of the construct will be summarized. 28 DependenceasaSourceofPower Thissectionreviewstheliteratureinoneofthedanimnt perspectives of the sources of power in organizations. This perspective holds that social actors have power over others who are dependent upon them (Buerson, 1962); there are two major views of the causes of this dependence. The first position holds that dependence results from control over resources, and the second holds that dependence results from the ability to cope with uncertainty or deal with critical contingencies . Several researchers have noted the interrelations between the two perspectives (Pondy, 1977; Pfeffer, 1981) , and have carbined elements of both views into more unified perspectives (Astley & Sachdeva, 1984; Fon'brun, 1983; Blackburn, 1981: Abdel-Halim, 1983). Resource Control Several researchers of organizational power view it as stetming from control over resources (Emerson, 1962; Pfeffer & Salancik, 1974; Jacobs, 1974: Kotter, 1978; Provan, Beyer, & Kruytbosch, 1980). Researchers who hold this view focus on the dependence that results from actually having or controlling the flow of resources that other social actors need or desire, and controlling them so coupletely that few alternative sources exist for obtaining what is desired (Pfeffer, 1981, p. 99) . These resources can be material goods (such as money) , information (such as technical expertise), or any other inputs into the work process . 29 Emerson (1964) The first researcher to directly propose dependence as the major cause of powerwas Emerson (1962). Hedefinedpoweras theamommt of resistanceonthemrtofActorBwhichcanbeovercmebyActorA. Power, in his formulation, is equal to and caused by dependence. Dependence is caused by an individual's motivational investment in goals controlled by another, and by the availability of those goals elsewhere. According to this view, power is relationship-specific. In every relationship, both members have a certain amount of power over the other; relationships characterized by imbalance are viewed as unstable. In such relationships, the actors will tend to stabilize their power relationship by the use of several "balancing operations”. In order to test these hypotheses, Enerson (1964) conducted a set of laboratory studies. The first of these was a post-hoc analysis of a previously performed study on the relationship between conformity to group norms and status insecurity. The initial study involved twenty Boy Scout patrol groups who participated in a laboratory study, in which they were asked to judge the distance between points of light; a false group norm was cowmwmicated, and deviation from accurate judgementstowardthatfalsescorewestakenasameasureof conformity. Results indicated that: scouts who had greater commitment to their scout group conformed more than those with less commitment; scoutswhowerereportedashavingveryhighstatusconformedtothe groupnorm: andscoutswmowerereportedashavingverylowstatus conformed to the group norm. 30 lie-analysis of these results, taking into account the source of the status perceptions , showed that those high status individuals who conformedtothegroupmrmwerethosewithhigh status outsidetheir own group, as well as within their group. Those with high status from one source only, either from within their group or from outside of it, did not conform: those with high status inside but not outside their group conformed the least. This was interpreted as evidence supporting the theory of power, based upon the notion that conformity to the group norm was evidence that the individual had little power because he was dependent upon the group. In-group status was hypothesized as increasing the power of the individual. Those group members with low status both inside and outside the group had little power; they conformed the most. Those who had high status within their group, but not outside of it, were powerful, and did not need to conform to the group norm. Those who had high status outside their group but not within were not dependent upon thegroupforhighstatus; itdidnothavepoweroverthem, andhence they did not conform. However, those who had high status both within and outside of the group conforned to the group norm; this was hypothesizedtobeduetothegroup's increasing "thedependenceofthe otherwise independent member, through granting him the special rewards of high status (e.g., prestige, salary increase, etc.). By doing so, the group can keep and control its valued members" (Brerson, 1964, p. 291). Asecondstudywas perfomedtofuruieracaminethesehypotheses. It involved a laboratory game, where individuals could form coalitions with other members of their triads in order to win multiple rounds. 31 Gamescoreswerecreatedbyachanceroll ofadie, multipliedbya weighting factor that was different for each person at each trial; power was measured as the size of each person's weighting score at that trial. Cumlated points were used as an index of a previous power advantage. Thepointswonbythewinnersofeachromdcouldbedivided however they chose: it was hypothesized that the drive to balance the power relationships would result in coalitions that reduced the overall variance in the accumulated scores. For this study, dependence (and hence, power in any given round) was held constant; only those rounds in which all three players had the same weighting score, and thus all wereviablepartnersinawinningcoalition,wereusedinthe analysis. Motivational investment in the winning was assumed to be high for all players. The availability of the goals elsewhere was indicated by the number of coalitions available to a member which would grant that person a winning score; again, this was held constant by examining only situations in which all players had the same weighting score, and all three potential coalitions would allow the members to win. The results supported the hypothesis that coalitions would form in such a way as to equalize the cumulative outcomes among all members. When one member had lower cmulative points, and a coalition of any members would allow a win, the coalition usually involved the individual with the lowest points. This was assumed to give that low- Scored individual more power, because the result would boost the higher "lender's lead over the middle-scored (non-coalition member) player. Therefore, it was hypothesized that the winnings of such coalitions 32 would be divided in the favor of the ”more powerful", lower-scored member. This was not fully supported by the results; half of these situations resulted in an equal division of winnings. There were several methodological limitations of this set of studies. The first problem is that they were used to support a theory of power, withouteverdirectly measuringtheconstruct. Powerwas used as an explanatory variable in the first study, in order to make sense of confusing results; all analyses were post hoc. Further, it was assumed that only in-group status confers power within the group, as opposed to out-group status; this notion had no theoretical validation. In addition, it was assumed that conformity to a perceived group norm corresponded to a lack of power, and resulted from dependence, again without justification. Inthesecondstudy, powerwasasswmedtoexistineachtriad, and be determined by both the relative weight assigned to each individual player in that round, and the need for well-off players to boost their scores vis-a-vis their closest competitor. Thus, while culmlative power was assumed to be represented in cumulative winnings, the player with the lowest cumulative winnings was prOposed to have greater power in any given trial of equal weights. On the one hand, Eherson used whatisinessenceamanipulationoftheamountofreswrcesthatthe individual controls (e.g., points he/she could win) as an index of her/his power; ontheotherhand, powerwasseenastheabilitytodeny anotheraccesstothoseresources. Thetwowerenotdistinguished either conceptually or experimentally; again, no direct measure of the construct of power was used. 33 These studies provided some support for the effects of motivational investment on the dependence-power relationship, and the external availability of desired outcomes on dependence. They assumed that dependence caused power, and considered the manipulation of one to indicate change in the other. This initial formulation of the causes of power did not involve an organizational context in any way; it was a purely social psychological approach to the determination of individual power. Jacobs (1974) Jacobs (1974) performed a theoretical analysis and expansion of Brerson's (1962) thesis that power is caused by dependence, focusing on the power of entire organizations over one another. He defined the construct of power as the deliberate overcoming of resistance in another in order to achieve one's goals. It was differentiated from the construct of control, which was defined as influencing others' behavior, regard less of intentionality and consciousness , and regardless of the resistance of the recipient. In this view, there must be resistance to the influence attempt in order for power to exist; compliance with legitimate requests from a person with authority isnotconsideredtobetheuseof power. Further, theinfluence attempt must be successful in order for it to be considered power. Like merson, Jacobs held that power was relationship-specific, and both parties in all relationships have some amount of power; unlike Emerson, he viewed unbalanced power relations as potentially stable. Jacobs integrated Blau's (1964) notion that compliance with directives or desires is a means of paying social debts caused by the Sharing of resources. Control was posited to result from either 34 dependenceorcoercion; theformerwasseenasbeingttemostcommon, and caused by controlling resources that others need. Both more and lessdependentactorsmayusecoerciontogetcmpliamewithdemands, while only less dependent actors may use the others' dependence to engender compliance. Jacobs then applied these theories to the control of organizations. Organizations were presented as entities embedded in enviroments; the control of organizations stems from their exchange relationships with other entities in that environment. The concepts of "motivational investment in goals" was replaced by the term ”substitutability" of the goals received, which was viewed as including both its availability from other sources and the organization' 8 willingness to do without it. Therefore, it included the 'essentiality" of the resource (p. 50) . From this, Jacobs posited two components of dependence, resource essentiality and resource availability. This proposition became the basis of a model of measuring organizational control. The model held that an organization's technology and environment determine which resources are the most critical; these exist at points of dependence. Points of dependence upon the environment occur at input acquisition, output disposal, capital acquisition, acquisition of production factors, and the aoquisitim of the labor force. The criticality of these points can be ranked by the availability of the resource elsewhere (e.g., the ezdstence and availability of other purchasers of the output), and the essentialness of the resource (e.g., the need for the input in order to maintain organizational functioning) . The ranking of each point of 35 dependence was proposed as the multiplicative function of the essentialress of the resource (which he recommended be rated by boundary-spanning individuals both inside the organization and in the environment, and be averaged across all raters) and tle number of competitive suppliers of the resource. Points of dependence which are essential and have little substitutability were considered to be ”problematic“. Several hypotheses were then proposed. The first was that organizations can be controlled by those who control their most problematic dependencies; that is, by those upon whom they are most dependent. The second dealt with organizational dependencies that exist at the disposition of output: the more fractionated the buyers of an organization's outputs, the less control they have over the organization. The third hypothesis dealt with coercion as a source of control. It held that when there are great differerees in values among the enviromental actors that the organization is most and least dependent on, the least dominant will resort to using coercive action occasionally; it is an alternative method for influencing the organization. The last hypothesis applied this notion to a specific type of environmental actor: when the values of regulatory agencies and those of actors that control problematic dependencies do not conflict, an organization will experience infrequent problems with the regulatory agencies. Jacobs' analysis focused on the control of entire organizations, although he adapted work at the individual level of analysis. He presented no empirical data, but his work expanded on the view that 36 power is based on dependence; he formally specified the control of critical and non-substitutable resources as the basis for dependence. This work added more limitations to the concept of power; however, after distinguishing the constructs of power and control, he proposed a theory of control. His notion of control is close to that which would later be considered power, with the only distinction that control does not require intentional use. Pfeffer & Salancik (1974) Pfeffer and Salancik (1974) also followed Emerson's lead that same year by conducting a set of empirical studies in organizations. They published two related studies, the first focusing on the measurement of power in organizations and its relationship with decision outcomes (Pfeffer & Salancik, 1974) , and the second attempting to identify the causes and usage of power (Salancik s. Pfeffer, 1974) . These studies examired the power held by organizational departments over one another: they investigated the power of different departments in a large University. They defined the construct of power as ”the ability to influenceoutcomes, clengingwhatmighthavebeenintreabsenceofthe use of power” (Pfeffer & Salancik, 1974, p. 148) . The two studies added substantially to the empirical knowledge base, and also added a degree of confusion about construct definition and measurement. In the first of these pieces, Pfefffer and Salancik (1974) found that University budget allocation decisions would result from the use of power, as well as from what they called 'universalistic,” or objective, criteria; an example of these is the number of Instructional Units (IU's) per faculty member, which is an index of the department's work load. This study used regression analysis to identify the 37 criteria used in making decisions, by identifying those criteria that accounted for the most variance in the decisions made. The dependent variable was the proportional allocation of discretionary University fundstoeachof 29departmentsoverthirteenyears; that is, the outcome of the budget allocation decisions over time. Several measures of departmental power were used. A single item subjective measure was used in an interview format; department heads wereaskedtoratehowmuchpowereachdepartrenthadwithinthe University. Power was described to respondents as the ability to influence decisions to achieve a desired outcome. The response scale used a 7-point Likert-type format, anchored with "a great deal of power” and "very little power". A "don't W response was also allowed. The other measures of departrent power used archival data, in an attempt to minimize response bias and allow for unobtrusive data collection; these indices were objective measures of enacted power. They included the proportion of departrental representatives on the University budget committee , the departtental representation on the University research board, the departmental representation on their respective college executive committees , and their representation on all committees. They found evidence of convergent validity among both subjective and objective measures, as most of the correlations among ttese variables were quite high; they ranged from .16 to .80, with most intercorrelations between .5 and .66. Regression analyses identified the universalistic criterion of departtental 10’s as the largest predictor of budget allocation, followed by the power indices of research board representation and then 38 by all committee representation. This held true when using all data points for each of 13 years, and when averaging archival data over the 13 years and using those averages to predict budget decisions; correlations were also significant after partialling out the variable of department size. Thus, there was evidence that both universalistic criteria and power were used to make budget decision, the former having more impact than the latter. The authors also attempted to test their definition of power itself, by comparing the power of the departments with "what would have happened” had the departments not exercised their power; that is, they examined the effect of departmental power on the relationship between teaching load (IU's) and budget allocation, and on the relationship between number of faculty and budget allocation. They hypothesized negative correlations between department power and both the IU-budget correlation and the faculty-budget correlation. The results were not definitive; their strongest indices of power (subjective responses and research board membership) resulted in non-significant correlations, while another strong index of power (total committee representation) found significant negative correlations (r = -.38 and -.24, respectively). This last analysis was the only one to directly test the definition of power used in the study; however, it did not clearly support the existence of the construct as theoretically specified. The preceeding analyses, however, all supported the strong relationships among the indices of power and the outcome of power. In the second article in this sequence, Salancik and Pfeffer (1974) held the basic thesis that power accrues to departments that have 39 control over critical resources. The greater the ability of a subunit to provide resources, and the more "critical, important or valued" (p. 455) the resources, the greater power the unit will have. In the University setting , seven important departmental resources were identified: number of graduate students; number of undergraduate students; national rank or prestige; public visibility; amount of outside grants or contracts; administrative and service contributions to the university; and business and professional contacts. The relative importance of these was determined by directly asking department heads to rank the effect each should have in determining budget decisions (with the exception of the last factor) , and by asking them to rank them in order of their importance to the overall Operation of tie University. The data collection site was the same in both studies, and the measures overlapped. The research committee representation measure of power, the overall committee representation measure, and the subjective measure of power described above were used in this study. The amount of control over the seven critical resources were measured by each department head's perceptions of his/her own department's control; the response scale was a 5-point Likert-type scale, with anchors of "much more than average" to ”much less than average”. Four of tte departrental resources were also measured by archival data. Statistical analysis of the relation among resources (measured archivally) and power indices found that amount of outside grants acquired was more strongly correlated with power than other resources, and the undergraduate IU's had a very weak relationship with departlent 40 power. National rank and graduate student IU's were also correlated with most power indices. Regression analysis found that the best predictor of two indices of power was the prOportion of faculty supported by outside grants, followed by the proportion of graduate IU's taught. Both the correlations arnd the regressions indicated stronger relationships between resource controls and sole indices than others; the research board representation index and the subjective index were much more strongly related with resource controls than the total commitee representation index. Analyses of department heads ' reports of their departrents ' access to resources (as opposed to the preceeding use of archival data) confirmed the finding that departmental access to outside grants and contracts had the strongest correlation with two of the power indices. Interestingly, there were very different patterns of correlations among the research board representation and subjective indices, as opposed to the total committee representation index; with the latter index, all correlations were lower, but different resources were more important. For example, business contacts and public image arnd visibility, which were unrelated to the research board representation index, were strongly related to this index. These results showed that different resources were differentially related with different indices of power; this pointwasnotdescribedinthetext. rTherextsetofanalysesfocusedontheuseofpowertodetermine the flow of critical resources. The authors hypothesized that scarce and critical organizational resources would be acquired by the use of power. The resources examined in the empirical analyses in this part 41 of the study were nearly all research- related. They included various types of internal research grants , which were ranked by department reads in terms of their scarcity annd their criticality. None of the resources which the preceeding section found to be related to any of tleindicesofpowerwereusedintheseanalyses. Zero-order correlations among the four resources and the measures of power founnd that resources ranked as more scarce and critical had stronger correlations with power indices; the research board representation index had correlations roughly twice as large as the other power measures (e.g., r's of .90, .85, .74 and .31). The authors then partialled out the effects of "objective criteria,” such as tre national rank of the departnent, on the power—resource relationships . The research board representation index maintained its high correlations with three out of the four resources, inncluding the two that were "critical annd scarce" . The other indices of power had non- significant partial correlations with the critical annd scarce resources, and had negative correlations with the least scarce resource. Regression analysis was used to attempt to predict the distribution of resources by tie inndices of departmental power. They founnd that the research board representation measure of power predicted allocation of these resources, as did the universalistic criterial of the proportion of graduate IU's and the prOportion of teaching faculty. The latter two measures are both indicative of the need for research support. Taken together, these two studies add both information and some confusion to the knowledge of organizational power. They supported the contention that power, along with objective criteria, may be used to 42 affect organizational budget allocations. However, defining the constructofpowerintermnsofrequiringtheuseof influencetochange behavior from what it otherwise would have been, was problematic; there wasequivocal evidenceofwhethertheuseof influencechanngedthe allocation of funds. The second study found that posession of critical andscarceresources, inthiscasetheaccesstooutsidefundsand having many graduate students, was related to the power of a departnent. It also found that only one index of power was related to the use of power in allocating resources, after universalitic variables were partialled out. These articles demonstrated the difficulties with ambiguity and/ or inconsisteicy in the measurement of the construct of power. The construct was described in the body of the research as the use of influence that produces results different from what would have been doneordecided, hadpowernotbeenused. Thiswas nothowthe construct was measured, lowever. Tm types of measures of power were used , departrental representation on various committees and subjective perceptions. The subjective measure inquired about tl'e ability of the department to affect decisions. The committee representation indices of power were all objective measures of enacted power. As such, tne latter measures tapped the outcores of decision-making processes concerning several different issues (representation on a variety of committees) , and their different patterns of relationship with other variables indicated that trey did not all measure the same thing. The researchers were actually measuring the outcomes of past decisions in which power EX have been used. It is impossible from these to identify what departnental representation would have been 43 like, had power not been used. Therefore none of the measures actually tapped the construct as it was described. Further, in the attempt to statistically identify how decisions mighthavebeenmnadewithouttheuseofpower, theresults failedto provide solid evidence that such power existed; out of the six correlations used to identify the relation between the amount of power and the effect of need (IU's and size of teaching faculty) on fund allocation, only three were statistically significant. Therefore, the empirical work in these studies explored the relationship between resource controls and the ability to intentionally innfluence decisions, rather than the construct that the authors formally defined as power. The analyses in the second study dealinng with the criticality and scarcity of resources that were associated with power also denonstrated ambiguity in construct measurement. The first part of this study identified specific resources that were found to be critical and scarce in Universities; however, only those resources directly pertaining to internal finnanncial support for research were used in these analyses. These were not among those identified as most scarce or critical. There was no evidence that those particular resources were truly critical or scarce for all departnents; the assessment of criticality and scarcity was performed by ranking these resources against one anotner, rather than by any non-relational index. These specific resources correlated extremely highly with the research board representation index, and support for hypotheses was mainly found in tnese correlations. This allows a possible alternative explanation for the findings : those departnents with greater representation on the research board had more control over the 44 distribution of several of the research funds, which may have allowed them to allocate more of the scarce funds to themselves. In addition, thosedepartmentswithnnormsthatsupportafocusonresearchmayhave been more likely to seek out finnanncial support for research. Or, sone departnents may exercise power in the specific area of obtaining research funding; this may have been previously enacted by getting departnental members to sit of the research board, and may later be enacted by acquiring more research funding. The ambiguity of interpretation resulted from the fact that only research funding resources were examined, and that only the research board power index had significant partial correlations with them. This indicates that direct attention to the issue-specificity of power measures may be entrenely important in internal and external validity of findings. Provan, Beyer & Kruytbosch (1980) Provan, Beyer and Kruytbcsch (1980) coducted an enpirical analysis of the impact of resource control on the power held by organizations over oe annotrer. They used Emerson's dependence-based view as a basis of all discussion; power was seen as the innverse of, and caused by, dependence. This paper contained a rather lengthy description of the problems of defining and measuring power. It held that power is the capacitytocontrol others, anddoesnotneedtobeexercisedinorder to exist. The authors did not insist upon limiting power to intentionally used influence or successfully used influence. They described it as relationship-specific, and focused on one issue of influence. 45 The authors applied Salancik and Pfeffer's notion that "organizations will be powerful relative to others to the extent that they (1) controlresourcesneededbyothersand (2) canreducetheir dependencies on others for resources“ (p. 200) . They also viewed organizations as embedded in networks of relationships which involve mutual dependencies . Organizations were seen as depedent on their resource suppliers, particularly for resources that are critical for their operation; ties to the environment as well as ties among specific organizations affect these dependencies . To study these issues, the researchers examined the power held by social service agecies over tne United Way organization which granted them funds, and the power of the United Way over the agencies; this was doe with attention to the overall environment in which those relationships existed. Thus, the focus of this study was a set of dyadic power relationships between each social service agency and the United Way; power was viewed as relationship-specific, although each relationship was affected by the others as well as the overall environment. Essentially, the social service agencies were dependent upon the United Way for their financial support. Monies were acquired through Unnited Way donations from the community, with the understanding that they would be used to support vital human services. Therefore, the United Way was depedent upon the social service agencies to successfully perform their functions and maintain community good will. Several hypotheses were tested. The first set focused on social service agency links with the environment. It was hypothesized that an agency would be more powerful relative to the United Way if it had more 46 links with the community, was better established, and provided more intensive services; all three were viewed as alloving an agency greater control over the resource of community support, critical to He United Way. Agency poer, relative to the United Way, was also hypothesized to be greater if an agency had more linkages withother social service agencies, allowing greater control over material and informational resonrcesneededbytheagency; andifthereweregreatercomunity demand for the agecy's services, which was presented as a unniversalistic criterion for greater agecy influence. The latter hypothesis denonstrated some ambiguity in the definition of power. As Pfeffer and Salancik (1974) found, decision ontcomes may well be determined to some extent by rational criteria; that does not mean that power itself, which the authors hold to be caused by dependence, will be affected by the fact that an agency has impartial reasons that it ought to receive funds. mltiple measures were used for all variables. Hoever, there was sone indication of questionable construct validity for sone measures. Three of the measures of agency-community links were the number of full- time and part-time erployees, the number of agency volunteers who worked at least eight hours a month, and the number of clients served by the agency. However, as the authors noted, these may just as apprOpriately be seen as indices of agency size. In addition, the first three indices assume that enployees and clients have positive feelings toward, and interest in, the agencies with which they interact; tne research on worker burnout, however, has shown that human service organizations are often staffed by employees with negative feelings toward their enployers and clients (Maslach, 1982; Maslach, 47 1978; Maslach 5. Jackson, 1981; Maslach s. Pines, 1977; Pines s. Kafry, 1978; Golembiewski & MUnzenrider, 1981; Szilagyi & Sims, 1975). The degree of agency establishment was measured by the number of yearstl'eagecyhadbeeninOperation, andbythenumberofyearsthe agecy had been affiliated with the United Way. However, community acceptance and support may not increase with the years of operation; many social service agecies Operate for years with a very low profile and little public recognition. Great attention was paid to the measurement of power; despite this , several of the measures were clearly problematic. The distinction between enacted and potential power was drawn. It was noted that subjective measures tend to tap both previously enacted power and its potential for future use, while objective measures were held to measure either potential or enacted power. Both have specific limitations: subjective measures may a1 lov for response biases, and the objective enacted measures fail to assess power that exists, but has not been used in any particular situation or decision. The latter may also indicatethatpowerhasbeenused,whenithasnotbeen. Inorderto fully measure all aspects of the construct of power, both subjective and objective measures, tapping both potential and enacted power, were used. In this study, power was viewed as being specific to different relationships, and it was measured that way. All indices focused on the power of each specific agency over the United Way. First, two perceptual measures of the power mm by organizations over one another were used. The first was a three-item scale, consisting of three questions abont the amounnt of influece each agency reld over the 48 United Way's decisions to allocate funds. This was administered to each agency's executive director, asking him [her to describe the control held by his/her own agecy. The questions asked tre executives to indicate their amount of agreement with statements, "My agency could readily obtain funding from United Way: a.) to cover increases in client needs; b) for new programs,; c) to cover increased administrative expenses" (p. 224) . The response format was a 5-point Likert-type scale, with anchors of ”strongly disagree” to ”strongly agree.“ This measure had an Alpha coefficient of .70. The secod measure of perceived power was administered to United Way staff, rather than to individual agency representatives; each agecy was described by a staff member who worked with that agency. This was a single- item measure, which asked tte "extent to which each agency is able to successfully influece United Way in order to meet its budget needs" (p. 224) . The response scale was a 5-point Likert- type format with anchors of ”virtually no influece“ to "a great deal of influence.“ However, the results provided no congruece between the responses to trese different itens, across organizations. The researchers found a correlation of -.08 between agecy directors' and United Way staff's perceptionsoftrepcweroftheagencies. Thismaybeduetodifferent perceptions across raters, or it may be due to the different measures used. In particular, tne measure used for the agency directors' rating of theirowngronps' powerdidnotusethe “howmuchinfluencedoesthis agency have” wording, which most subjective power measures do. It asked for the stregth of agreement with statements that indicated high 49 agecypower; itislikelythatagecydirectorsrespodedtothiscue by over-estimating their own agency's power. It is impossible to determinethisfromtlereportedresults, asresponsestothisscale were standardized and a mean of 0.0 therefore resulted; this cannot meaningfullybeconparedtothereportedmeanof 3.120nthesingle- item question administered to the Unnited Way staff. However, this would be consistent with the findings of Pfeffer and Salancik (1974) , wtcreportedthatdepartnentheadstededtooverinflatetneimportance ofthosepowersonrcesthattheirdepartments reldinabundance. Furtner support for this lies in the the interrelations among the agency director measure and all other variables; it had far lower intercorrelations with all other measures of poer, as well as with its sonrces. Although it was not directly addressed by the authors, neither of tl'ese subjective measures of power were general in nature. Both questionnaires specifically tapped the agencies' potential for intentional influece over decision making by the United Way, and even targeted one specific type of decision: fund allocations. Both perceived poer measures actually tapped a power over a specific issue, rather than overall influece. T‘l'estudyalsousedtwoobjectivemeasuresofenactedpower: the percentage of an agecy's budget requests that were met by tte United Way, averagedoverthreeyears; andthepercentageincreaseinan agecy's funding by the United Way, again averaged over three years. Theformermeasureassmedthatanagencythathaspoerwilluseiton a regular basis to get more funding. It also assumes that no norms exist for the United Way that differ among agencies, in terms of the 50 proportion of their funding request that will be granted. For example, it may be that Agecy A generally overinflates their budget estimates by 15%, while Agecy B does so by 30%. The second measure was based on these same assumptions, as well as the notion that agecies will desire increased funding every year. Tierewasnoevidecetosupportthisconjecture, andtheresults showed nonsignificant correlations between these two objective measures (r a .14) . Bach measure also had different correlation patterns with predictive variables. The first of these measures seemed more valid; its intercorrelations with sources of poer were more consistent, and four of the twelve were significant at the p < .10 level. The second of the two measures correlated in the opposite direction as predicted for ten out of twelve relationships . Like the subjective measures, these indices all focus on one specific type of ontcome of power: fund allocation decisions. The first of these two indices is very close to the index used by Pfeffer and Salancik (1974) to validate their measures of power; it, too, was an objective measure of the outcome of budget decisions which did not imply that an annual increase in funding was indicative of power. Pfeffer and Salancik' 3 measure was not relationship-specific; it merely identified those groups which had received greater funds. Provan, Beyer and Kruytbosch's measure was relationship-specific, and identified tte extent to which each group's formal funding request was granted. Thelastindexofpowerusedinthisstudywasviewedasan objective measure of potential power. This was an attempt to measure the "netdependence' oftheagencyontheUnitedWay. Oonsonantwith 51 Emerson, thisindextookintoaccoonttnecoceptthatbothpartiesin a relationship have a degree of poer over one another; it included a measureofboththedependeceof eachagecyupontheUnitedWay, and oftnedependeceoftheUnitedWayuponthatagency. Theindexof agecy dependence was the percentage of an agecy's total budget that wasfundedbytheUnitedWay, whiletl'eindexof UnitedWaydependece was the percentage of the United Way's total fund allocations that were giventothatagency. Thenetdependeceindexforeachagencywas formed by subtracting tne United Way dependece from the agecy dependence. The strength of this measure lies in its fidelity with the relationship-specific and reciprocal nature of the construct, as described by Emerson (1962) , Jacobs (1974), and these researchers. The attempt to formally include the reciprocal nature of the poer relationships, by including indices of both parties' dependece upon one another, is novel. The specific measures which formed the index continued the focus on poer over the specific issue of fund allocation decisions, and the use of percentage of funds allocated allowed for the easy conputation of the ”net dependence". Unfortinately, the title of ”net dependence” is misleading, because the construct of net dependence was not tapped. TheindenofagencydepedecedoesindeedmeasureUhitedWay control over a resource that is critical to each agency; there were few other sources of financial support, and it was essential for the functioning of the agecies. Hoever, the index of United Way dependenceuponeachagencydidncttapitscontroloveranyresource that is critical for tne United Way. The hypotheses that underlie the 52 idea of mutual depedece specified that to United Way needed positive community perceptions of its activities, in order to raise funds. This is the critical resource that the agecies provided for tie United Way: itisthesourceof UnitedWaydependece. Measuringthepercentageof funding that wennt to the agency did not tap United Way dependece. The measurenent problems were reflected in the enpirical results. The intercorrelations among poer measures provided little evidece of convergent validity. The strongest intercorrelations were obtained between the single item perceptual measure (administered to the United Way enployees) and several of the other indices; it had the only correlations that were significant at p < .05. There were only two of trose: its correlation with the net dependence measure of potential power (r =- .26), and with the percentage of budget granted measure of enacted poer (r a .42). This perceptual measure and the percentage of budget granted index both had the fewest problems with their construct validity; they seem to be the most valid of any of the poer measures. Their patterns of intercorrelations with the predictive variables were extrenely similar; nine out of twelve were were within .15 of one another. All other measures showed little consistency, as tteir correlations with predictive variables were either in the opposite direction most of the time (agecy directors' perceptions and the percentage increase of funding), or were entrenely high (net dependece) . Provan, Beyer and Kruytbosch factor analyzed the independent variables, in an attenpt to identify underlying causal mechannisms that accounted for variance in poer indices. The found two factors with eigenvalues of greater than 1.0. The first consisted of the number of 53 employees, the number of clients, a perceptual measure of board member influenceinthecommunity, andtwomeasuresofboardmember interlocks. This was called “links with the community”: however, most of its components were also indices of size of agency and size of board. Thesecondfactorconsistedofameasureofboardmember prestige, the number of years serving the community, and the years in the United Way; the latter variable clearly anchored the factor, with a factor loading of .95. This was called ”establishment in the community"; its component elements reflect both status and years in the human service domain. These factors, along with the other independent variables, were subject to regression annalysis in an attempt to predict the various measures of power. The power measures with both the highest zero-order correlations and with the largest R (.77) was the net dependence measure of potential power. The independent variables with significant beta weights were the “links with the community" factor and the cost per client index. This indicated that larger, more labor-intensive agencies depeded upon the United Way for most of their funding, compared to the proportion of the United Way's funding that they received. Thefirstofthetwepowerindiceswithahighdegreeofconstruct and convergent validity was the single item measure of perceived power. This measure had a significant zero-order correlation with the two factors identified by the factor analysis, “links with the community" and establishment in the community (r = .27 and .28, respectively). Whether the first of these is interpreted as agency size or number of community links, these results support the hypothesis 54 that there is a relationship between embeddedness in a community, with the resulting inncrease in potential communication clnannels and access to resources, and power. However, regression analysis found no significant beta weights in the attenpt to predict this index of power; the R was .34. Theenactedpowerindexofthepercentageofbndgetrequestgranted had significant zero-order correlations with two predictors : the number of interagency program (r = .28) , and client growth (r a - .47). Tlese same predictors had significant beta weights, with an R of .58. Those agencies that received a larger proportion of their requested budget were those that were offered programs jointly with other agencies. On the other hand, agencies that served more clients were likely to receive a smaller proportion of their requested budgets. This particular objective criterion for fund allocation did not appear to be used. Thus, the analyses supported the basic contentions that budget decisions were related to factors that cause power. These factors included establishment in the community, size, and links with both the community and other agencies; these were all factors which produce greater visibility for the agencies, which was a critical resource for the United Way. E‘EEEX The enpirical studies of the resource-dependence view of organizational power suggest and support the notion that control over critical, scarce resources leads to power. However, the empirical works are often hampered by problems with construct definition and 55 measurement. The construct of power was not always connpletely defined, nnor measured in a way consistent with the definition given. Regardless of the formal definition of tre construct, most of tie perceptual measures focused on intentionally used influence; these measures showed tte greatest degree of construct, convergent and discriminant validity. Althoughtheconstrnetofpoerwasnotdefinedasbeingissue- specific, Pfeffer and Salancik (1974) measured it with tte outcomes of decision making on different issues, and got different results depending on which issue was focal. Provan, Beyer and Kruytbosch focused on oe specific issue in all measures: control over fund allocation decisions. All researchers described power as relationship- specific. mly tne measures used by Provan, Beyer and Kruytbosch actually tapped this aspect of the construct. There was inconsistency of measurenent within as well as across studies. mst measures were validated by comparing then to one anotler, or to issue-specific outcomes; their congruence with the construct definition was nnot always used to validate trem. Some variables, such as the outcome of budget decisions, were consideredtreresultofpowerinonesmdy,andameasureof powerin annother. Several causes of power and results of power were confused with the measurement of power itself, and tlen correlated with other causes and results. In such cases, thecorrelations between “power" and both its ”causes" and "effects" were higher than tte correlations among measures of power: outcome measures often related quite strogly to outcomes of power use, particularly wden both measures dealt with outcomes of decisions made about the same specific issue (e.g., fund allocation decisions). Likewise, potential power measures tended to be 56 more strogly related with mneasures of critical resources that granted, but were not the outcomes of, power (e.g., agency size and community links). The theoretical work was more consistent. The treoretical causes of power were clearly identified: control over resources which are both essential for the conduct of work and non-substitutable create dependence: dependence grants the resource-controller power. Control overresourcesmeanstheabilitytoacquirethen, iftheydonotexist within tne organization, or the ability to determine their flow if they do. Critical resources include material factors such as mnonney, and non- material factors such as access to communication channels and communnity support. Despite the measurenent and construct definition problems , these basic theses are supported by the empirical findings. Organizational decisions about tie distribution of funds were found to be predicted by the power of organizational subunits and by other organnizations. Such power was empirically related to the control over critical resources; ttese resources included status, money, links with the community and links with other organizations at the same level. Control over Critical Contingencies The second major approach taken to the development of depedence in organizations is that of control over critical contingencies and the ability to cope with uncertainty (Crozier, 1964: Hickson, Hinings, Lee, Schneck s. Pennings, 1971; Hinings, Hickson, Pennings & Schneck, 1974; Hanbrick, 1981) . This view is based on the conceptualization of 57 organizations as decision-making systems that cope with uncertainty through a variety of procedures (Thompson, 1967) . It holds that power accrues to those social units that “cope with uncertainties, or ' strategic contingencies' facing the organization" , because other social units are depedent upon them to do so (Hambrick, 1981, p. 253) . Crozier (1964) Crozier (1964) first applied the concept of coping with uncertainty to tie causation of power in organizations . He performed an in-depth analysis of two French organizations, a "Clerical Agency" and an entity he called tne "Industrial Moncpoly"; Pfeffer (1981) later identified this as a tobacco-processing organization. After a detailed examination of these organizations , he proposed a specific theory of the functioning of organizations in general , which he applied to the functioning of cultural systems. Of relevance here are his findings regarding the etiology of intra-organizational power . T'ne Monopoly consisted of thirty processing plants, a central office , a research center , warehouses and other production support locations. The research was conducted in the processing plants, which were located all around the nation. The organization was established and adminnistered by the French government; due to this, it had several unusual organizational characteristics. First, its sole purpose was to produce a product; its sales were controlled by ancther arm of tie government. Prices were set by the state, and, as a monopoly, there was nno corpetition; therefore, there was little pressure to improve production techniques, expand markets or increase efficiency. The organization was essentially isolated from its environment. 58 Nearly all of the decision making was centralized. Work procedures were extremely routinized. Personnel policies were determined by the central office, and allowed little lee-way. The organization was nmionized, and both promotions and salary were conpletely based upon seniority; salaries were standardized nation-wide. It was nearly innpossible to be fired, and selection was formally specified to give preference to family menbers of incumbents. Selection of the different types of employees was performed independently, and the interaction among the groups was mninninmal. There were several groups of employees. The first and most numerous were the production workers , who were primarily working-class women. The "maintenance workers" formed the second group, who performed the function of maintenance engineers. Tl'ese were highly skilled trades people, nearly all men. The next group consisted of the shop foreman, who performed white collar office work and first line supervision; they were generally young men with college degrees. Each plant also had a few higher level administrative officers, doing work such as accounting and personnel. Each plant had a single technical eginneer, who oversaw all maintenance and repair problems, and therefore directed the work of the maintenance workers. Iast, each plant had an assistant director, in charge of Operations, and a director, responsible for the entire plant. Crozier nnoted that there was one single powerful group in this organization, and its power stemned from the type of work the enployees did. This power imbalance existed on both the step floor level and at the managenennt level; consistently, the maintenance workers and technical egineer were the dominant group in nearly all plants. 59 Although the assistant director was formally the superior of tie technnical engineer, the two operated as peers: each could have some powerovertheother, andinsomeplants, theegineerhadnearly conplete control over decisions that were made by the assistant director. an the shOp floor level, the maintenance workers were identified as by far the most powerful group. Trey were extrenely cohesive in their attitudes and behavior patternns , they worked in the ways and at a pace with which tley were comfortable, and trey regularly made key decisions affecting all workers, such as when to shut down machinery. In contrast, tne shop forenen had little power. Trey acted as timekeepers and accounting officers, ensured that all workers had supplies and maintenance services, and decided seniority cases: trey had nc ability to discipline or reward employees, or to make decisions about substantive issues . Trey viewed tremselves more like production workers than like managenent, and reported that the maintenance workers undercut their authority: this was reported from Crozier ' 8 direct observations, as well . The production workers were the least powerful of any group. Crozier identified the cause of the maintenance workers' power. In this organization, protected from tne environment and run by rigid and detailed rules guiding all decisions, there was only one source of uncertainty: machinery break-dam. T'lere were machines at all levels ofmcdernizationandconplexity, andtheywerepronetobreakdowndue tottenatureofthematerial processed, theirage, andsometimesdue toimprOpermaintennance. Onlyonegroupof enployeesknewhowtodeal 60 with this uncertainty: trey therefore had a great deal of power. [Tlhe peOple who are in charge of maintenance and repair are the only ones who can cOpe with machine stoppage. They cannot be overseen by anyone in the shop. No one can understand.what they are doing and check on them... Production.workers are displeased by the consequences of a machine stOppage. It.disrupts their work: it is likely to make it necessary for them to work harder to compensate for lost time... It is not surprising, therefore, that the behavior of the maintenance man...has a tremendous importance for production workers, and they try to please him and he to influence them.... The contrast between the power wielded by the maintenance men and the lack of influence of the supervisors explains the advantage that the former have over the latter. Supervisors cannot check on maintenance...[or] reprimand the mechanics who work in his shop. There is likely to be a perpetual fight for control, and the supervisors will usually be the losers. (Crozier, 1964, p. 109). Crozier adopted Dahl's definition of power: "the power of a Person A.over a person B is the ability of A,to obtain that B do something he would not have done otherwise" (p. 157). This holds that power is the ability of a social unit to change the behavior of another. It is, by definition, the successful use of social influence, and is consistently described as being intentionally used. It is relationship specific: indeed, it ”refers to a kind of relationship that is neither unidimensional nor predictable ' (p. 145). His descriptions of the power of the maintenance workers focused on both foreman and production workers acceding to their requests and decisions, rather than disputing them. The'description of the use of power indicated that it.was seen as existing across all types of decision issues, behaviors and feelings. 61 Crozier also described the ways in which the maintenance workers kept their information restricted from others. First, the skills necessarytofixsomanydifferentmachineswerelearnedthrogh experiece with machines: it was not possible to send forenen through a training course and thus give them the innformation needed. No other employees were allowed to involve thennselves in any way with machine maintenance, regardless of their abilities or knowledge. ”The one unforgivable sin of a machine Operator is to 'fool around' with her machine... no explanation is ever given“ (p. 153). Second, all written plans, blueprints, directions or other such information were consistently “lost". All information was only available through experience with the machines, which only maintenance workers had. Tl'epowerofthemaintenancestaffextendedupwardtothetechnical engineer, who was responsible for, and understood, the work of the maintenance crew. He protected them in any disagreements that moved up into the administrative ranks , and they granted him power vis-a-vis the assistant director and director. This situation changed when a plant undertook re-organization , which might involve acquisition of new buildings or major equipment. In such situations, the decisions regarding the changes were made only byttedirectorandassistantdirectorn ‘theyttenhadmorepowerthan thetechnicalegineer, becausetheyhadtheinformationandformal autlority to deal with re greatest area of uncertainty . In this analysis, power resulted from the ability to deal with uncertainty that affected others' work. In this organization, hierarchical autl'ority did not allow one any ability to deal with uncertainty, as there was generally only one type of uncertainty. In 62 order to increase their power, plant directors and assistant directors soughttoaddanotnersourceofuncertainty, overwhichtheywouldhave sore measure of control: expansion and re-organization of the plant. This became the primary goal of assistant directors, wl'o wanted greater influece. The directors were described as having sone control over ”uncertainties": that is, they could decide how the rules would be administered, they directly controlled tne supervisory staff, and they had a great deal of status. With formal authority to direct, but no resources or information to control, tte supervisors had little power. In addition, tne maintenance workers actively worked to weaken the authority of the supervisors. Because their power was ”covert” , rather than "legitimate“ , they ”constantly attackled the supervisors] and deprecatted] tteir efforts,... making it impossible for then to assert their authority” (p. 154). This research was exploratory in nature, and therefore used case analysis: Crozier referred to it as the ”clinical method" (p. 4), consisting of an in-depth analysis of one or two particular organizations. Therefore, few statistics were reported, and all that were presented were descriptive. Relations among groups and individuals were described in qualitative terms. The findings, however, were based on UK) successive stndies: one involved site visits to twenty of thirty plants within tre organnization, and extensive, Open- ended interviews with both managenent and unnionn members. The secod consisted of in-depth examinations of three sites , involving observations , a questionnaire administered in interviews to a large 63 prOportion.of the employees, and group discussions of the preliminary findings. Because of the exploratory nature of the research, the hypotheses evolved from the case analysis: they were not tested empirically. All of crozier's interpretations of his findings centered around the concept of coping with uncertainty as the key cause of power. He noted that in this particular organization, security and constancy were the major rewards of job incumbency: thus, reducing uncertainty should be far more powerful a consideration in this sample than in other, less predictable and less mechanistic organizations. However, this did not affect his interpretation of the causes of power. Other interpretations of the findings were possible. For example, it was held that the maintenance workers had a great deal of power because machine stoppages were the only sources of uncertainty: supervisors' inability to any decisions or control any resources, rewards or punishments was viewed in terms of the lack of 'uncertainty in those areas. An alternative, resource-control interpretation would hold that the maintenance engineers were the only ones who controlled critical information. Supervisors held no control over the acquisition or flow'of critical resources that.would normally be under their jurisdiction, such as money, promotions or use of modern machinery. Thus it may not.be the ability to deal with uncertainty, per se, that is vital: however, this was the only interpretation taken by Crozier. Another source of power that was supported by his description, but not identified as such, was formal authority. It.was described as the key competing source of power, at both the Shep floor and upper 64 managenent levels. Technical engineers were in continual conflict with assistant directors, because while the former held key information to cope with uncertainty , the latter had hierarchical auttority over them. As a result, decisions were usually made together, by mutual agreement. At the slop floor level, the formal authority of the supervisors was a sufficient threat to the power of the maintenance workers that they continually undercut them, leading to their demoralization and the perception of the production workers that the supervisors were not really in charge. Thus, witnout otter resources or uncertainties to control , the hierarchical authority was not sufficient to overcome the ability to cOpe with uncertainty: it was, however, its chief competitor in the creation of power. Hickson, Hinings, Lee, Schneck & Pennings (1971) Hickson, Hinings, Lee, Schneck & Pennings (1971) integrated Crozier' s uncertainty-coping viewpoint with Lawrence and Lorsch' 3 view of organizations as systems of interrelated departments, and with Thompson' 8 view of organizations as decision-making entities. They produced a formal theory of tie causation of power, focusing on "the division of labor as the ultimate source of intraorganizational power, and power [being] explained by variables that are elements of each subunnit's task, its functioning, and its links with other subunits” (p. 34). i This theory used the Emersonian (1962) perSpective that power characterizes relationships among social units. Power was defined as ”the determination of the behavior of one social unit by another" (p. 36): that is, the use of intentional influece, specifically aimed at changing behavior. 65 Power was held to exist in three dimensions: its weight, scope and donainn. Tle weight Of power is similar in concept to its amount: it was defined as ”the degree to which B affects the the probability of A behaving in a certain way" (p. 36) . Therefore, this definition of powerdidnotrequiresuccessful influece. Thescopewasviewedas tre "range of behaviors" Of each social unit that may be determined by the power holding unit. This specified that power could exist over different issues. The autnors went farther to specify the view that different amounts of power are likely to exist over different issues: In publisned research such distinctions are rarely made. Power consists of sweeping undifferentiated perceptions Of respondents when asked to rank individuals or classes Of persons...on influence. Yet...if it is taken for granted that, say, marketing has the most to do with finance matters,...then the validity Of forcing respondents to generalize single opinions across an unnstated range of possibilities is questionable. To avoid these generalized Opinions, data collected over a range of decision topics or issnes are desirable." (Hickson, Hinings, Lee, Schneck & Pennings, 1971, p.37). Thedonainofpowerwas seenas thenumberof social unitswhose behavior is determined. Autnority was ”regarded as that part Of power which is legitimate or normatively expected by some selection of role definers" (p. 36) . Subunit autnority was conceived of as the ”formally specified range of activities trey are Officially required to undertake and, therefore, to decide upon” (p. 36) . This theory held that power is caused by depedence, rather than the ability to cope with uncertainty. “It is not tre uncertainty, but what is done about it that confers power" (Hinings, Hickson, Pennings & Schneck, 1974, p. 22) . A social unit becones depedent upon another 66 if: the unit is able to OOpe with uncertainty that stOps the work process of other subunits: the first unit is central in the flow Of work: and the coping abilities of the subunit are not substitutable. The actual source of power is the ability to cope with uncertainty, but this only accords power to the focal subunnit if other subunits cannot cope with the uncertainty, and if they are dependent upon the ability Of the focal unit to solve problems in order to perform their functions. Centrality Of workflow was viewed as having two subsets: the imnediacy of the inmpact of the unit's work on the functioning of otner subunits, both in terms Of time and severity Of impact, and the pervasiveness of links to Otter subunits (i.e., the number of other subunits the focal unit has ties to). The theory held that the more extensive and immediate the impact of a subunit's performance on other units, the more that its ability to cope with critical contingencies will be a source of power. The theory also held that the less substitutable the activities of a subunit are, the more power it would have. Substitutability was viewed as "the ability Of the organization to Obtain alternative performance for tie activities” (p. 40): the direct interpretation of this was the ability to find other units or agents, inside or outside the organization, wlo could accomplish the subunit's tasks. Tl'e indirect interpretation Of this moved from the subunit to the individual level: it was the ability to replace the individual incumbents Of the department. 67 These variables were integrated into a model of tie causation Of power. we more variable was presented, and used as an integrating construct: control over strategic contingencies. ”A contingency is a requirenent of the activities of one subunit which is affected by the activities Of another subunit" (p. 43): strategic contingencies are tlose that affect power, due to their centrality and nnonsubstitutability. The model thus used control Of strategic contingencies to be the direct precursor to power. "It represents organizational interdependence: subunits control contingencies for one another' 3 activities and draw power from the dependencies thereby created" (p. 43) . The effectiveness of OOping, together with the amount of uncertainty Of departmental inputs, caused control of contingencies. Centrality of workflows had a positive causal impact on the strategic nature of these contingencies: substitutability of activities had a negative causal impact on their strategic nature . The routinization of activities was presented as having two effects. The first was increasing the proportion of routine task behaviors in order to prevent uncertainty , thereby OOping with uncertainty by preventing its occurrece: the example given was to institute plannned maintenance of machinery in Crozier's Mocpoly, tiereby minimizing breakdowns. The second was routinizing by increasing information and absorption of unnplanned changes. This is a way to make activities more substitutable, by making the means Of coping more regulated: the example given was to create task instructions that specify how to repair machines. This does not prevent tne machine from breaking down, but increases the amounnt of 68 information available for now to fix it wren it does, allowing otters to substitute for the work of the maintenance department. Most of to same autrors conducted an enpirical innvestigation of these causal hypotheses some years later (Hinings, Hickson, Pennings & Schneck, 1974) . Trey examined the power of the four departments in each of seven small manufacturing organizations. The research explored the relationships among departmental power and the extent to which the subunit coped with uncertainty in inputs, the immediacy Of the effect of its Operation on the functioning Of otter subunits, the substitutability Of its operation, the routinization of its activities , and its linnks with tie work of other subunits. Data were collected in three steps. First a set of exploratory interviews were conducted with the chief executive Officers of several Of the companies: all variables in the theory were included in the discussion, and the specific subconponents of several variables were identified during this process. Second, a set of semi-structured interviewswerecondnctedwitheachcmandtneheadofeachoftle28 departments included in the study: these interviews tapped most variables. Trey asked respondents to describe only their own departnent. Iast, a questionnaire including measures of all variables was administered to to department heads and CED's. All department heads responded with descriptions Of their own departnents and each of tie other three departments. Through the conbinnation Of interviews and questionnaires, multiple measures were used for each variable: all innvolved perceptions of to samedepartmentheadsandcm'sineachconpany. All questionnaire data were examined for concurrece across department heads. Although 69 no inter-rater reliability indices were reported , the authors noted that "there was substantial agreenent between respondents both within and between organizations", on both to ordering of tie subunits in their organnizations and on the overall score (between one and five) on each variable. Substitutability, "the ability of the organization to Obtain alternative performance for the activities Of a subunit,” (p. 26) was measured in the interviews by questions focusing on the difficulty innvolved in the actual substitution Of others for tie work performed by the subunit, and by questions focusing on the extent to which tasks could be performed by other departments or agecies . Substitutability was also measured by a questionnaire item inquiring about the ease Of replacing a closed department. Two aspects Of workflow centrality were posited as existing: workflow pervasiveness and workf low immediacy. Workflow pervasiveness was defined as "the degree to which the workflows of a subunit are linnked to tie workflows of other subunits." This was measured by an interview question describing where subunit inputs came from and where subunit outputs went to: these descriptions included every subunit, parent companies, suppliers, custoners, governnment agencies, etc. This wasusedtocreatea "matrixof interconnections" amongsubunits, and five measures of departmental connectedness. These linnks were highly intercorrelated (r's ranging from .62 to .94): it was held that subunits that were pervasive teded to be pervasive in all ways and with all groups. Two questionnaire items were also used, inquiring about to perceptions Of ”how far the flow of work" of one's own 70 departnent, andother departnents, were connected to tnework Of each of the other apartments. Workflow immediacy was coceived of as "the speed and severity with which the workflow of a subunnit affects the final outputs of the organization" (p. 27) . This was measured by classifying each departmental output by the length of time it would take for it to have an effect on to organization's output of finished products. Tno questionnaire items were also used to tap workflow severity and mrkflow speed. The first measured the size of the impact a department's closing would have on the "shipping Of finnished goods from true plant" (p. 43): tie second measured the length Of time it would take for a departnent's closing to affect the shipping Of parts. Uncertainty was defined as "a lack of information about future events, so that alternatives and their outcomes are unpredictable" (p. 27) . The first measure of uncertainty used a scale to categorize the annount Of "unpatterned variability" of ten different organizational inputs: the score for each subunit was created by summing them. COping with uncertainty was defined as "effectively dealing with uricertainties". Three techniques for dealing with uncertainty were sShecified and measured. Coping by prevention dealt with "activities mien reduced the probabilities of variations occurring in the inputs of the organization: for example, advertising aimed at offsetting a pQQsible fall in sales," while coping by information included ~§Qtivities which provided forewarning of probable variations in the LITE-Ems Of tie organization ." The last technique, coping by a‘bfiorprtion, was defined as "activities which Offset the effects Of N variations in the inputs in te organization" (p. 29) . These k 71 three different means Of OOpinng with uncertainty were used to describe each of to ten organizational inputs that were previously identified as important. Coping by absorptionwas given adouble weight. Inordertotakeintoaccounttheextentoftneenisting uncertainty, the measures Of coping and Of uncertainty were conbined. Both the unpatterned variability scores and the OOping tectmiques scores were divided into quartiles, which together prodnced a set Of 16 categories. This measure actually takes into account tre amount of mertainty that exists to be coped with: however, it only counts the number of coping techniques witnout taking their effectiveness into account. The last independent variable, routinization, was defined as "the Process Of rationalization and proceduralization" (p. 30). This \rariable was measured by questionnaire items innquiring about the amount of work in each department that fol lowed a set procedure, and how much eachdepartmentdidthesameworkinthesamewayeveryday. Power was defined as "the determination of the behavior of one 8Ocial unit by another" (p. 30). The dimensions of "weight" and " Scope" were measured by both interviews and questionnaires: the "domain" Ofpoerwasneldconstantastleoterthree subunits ineach Organization. Tl'epowerofeachsubnmitoverall otherswasthe dependent variable in the study. Thus, despite the description Of E""Qtuen: as being relationship-specific, the power of each department over a1 1 others in to organization was actnally measured. The first measure was subjective. It asked each department head in time organization to respond to a questionnaire consisting of parallel Netions about the amount or influece held by each Of the four ¥ 72 departments over problems in each of 17 specific issues; these issues load been identified in the preliminary interviews as ”recognized annd z-ecurrent problem areas in the organizations" (p. 30). These included issues such as marketing strategies , product packaging, operating performance of equipment, obtaining raw materials, overall capital and tar-capital budget, annd personnel training and development. T're issues were “allocated” to the different functional areas; for ee-cample, product packaging was considered a problem within the “marketing function, while obtaining raw materials was allocated to the Production function. described. The technique used for this allocation was not It was not determined by the formnal authority of each department for decisions on the issue; this variable was measured separately, using perceptions of tne CED's, as an index of power. It is therefore assumed to be based on the researchers ' determination of each issue's logical situation. This single question, repeated for 17 issues annd four departments, Vvas used to create three different inndices. The response category was a five point Likert-type scale with anchors of "very great influence" to "little influece." These responses were used to construct tie power weight scores; these were the means for each subunit for each of ti‘ne 17 issues, created by averaging the perceived amount of influece a'cross all four department heads. I These responses were also used to create a power scope score, by QQ‘l-lnntingthenmberof issuesonwhichasubunitwasratedatorabove % mean score for all subunits. An index called ”net scope deviation" NS also created, which provided a combined power index. It was ab Qduced by subtracting the number of issues that existing in a k 73 subunit's ”own" task area on which the subunit was rated as having less than the mean power weight, from the number of issues outside the subunit's taskareaonwhichitwasratedashavingmorethanthemean power weight. These inndices were all produced by the same data, annd were strongly intercorrelated. The weight and scope scores were intercorrelated at r a .93, aswerethenetscopedeviationandthescopescores. Bothof trelattertwowerebasedonthefrequecyofhavinggreaterpowerthan the mean; the more often that occurred (i.e., the greater the scope), the more likely a subunnit would have a positive net scope deviation. Weight and net scope deviation were correlated at r s .81. The mean weight score for each of the four types of departments were presented for each of the 17 issues. These results showed that, despite the strong intercorrelations between the weight annd scope inndices, every department had varied amounts of influece across decision-making issues. The overall most powerful department had the least innfluece over some issues, and the least powerful department had the most innfluence over others. This provides empirical evidence for the issue specificity of power. One limitation of all inndices produced from these data is that they focused on power over the resolution ofgproblenm. While this fits the strategic contingencies approach to the causation of power, it does not fit the definition of the construct itself. Power was defined as the ability to determine others' behaviors: tneir behavior is not limited to how they resolve problem. This measure described different specific problems, and measured power over each issue. It trerefore allowed the specification of power over different issues. Even with 74 the specification of 17 different issues, however, the resultant indices still did not tap the breadth of issues included in the construct; treyonlyexaminedtheamomntandscopeof influeceover how problem are solved. Power measures were also created from data gathered in the structured interviews. First, each department head was asked which departments played a role in each of four stages of decision making about each of the 17 problem areas. These stages were: raising the issue; providing information or services; deciding on the course of action; and carrying out the action. They reported "overwhelming" consistecy in responses on these item (p. 31) by all department heads. All four indices constructed from these questions were considered ”participation power" as opposed to perceived power; all inndices, however, resulted from the perceptions of the department heads. The questions asked for perceptions of "who raises the issue or initiates discussion“, etc. (p. 43). The perceived power data gathering questions all asked about the power of the department over decisions made in different areas; tre ”participation power” data gathering questions all asked about ”who" played which roles in decision making. This may account for some of the differences among the results: the latter questions focused attention on the individual who happens to be the department head, while the former questions focused attention on the entire department. Any lower level employee whose actions or information might influece decisions would not be taken into account in the latter questions, but would in the former. 75 The authors hold that ”participation in stages of the decision process is frequently tantamount to exercising power" (9. 34): therefore, basing power inndices on this participation was considered ”actual" rather than potential power. This implies that the department teads' perceptions of who participates in decision making are objective: the data gathering techniques, however, are subjective. It is likely that reports of who participated in decision making were innfluenced by perceptions of who was more powerful. This study did not measure, as Pfeffer annd Salancik (1974) did, tie outcomes of decisions; there was no objective criterion to validate perceptions of influece over decision making against actual results of decision making. In addition, the notion that participation in decision making measures actual power does not fit the definition of the construct of power. An executive may regularly present information or attempt to influece proceedings, without determining the decision or tie subsequent actions taken. In actuality, only the "choosing action" question taps the construct of peer as defined. All other questions tap ways of exerting influece on the decisions made annd actions undertaken, by raising the issue, presenting information, or carrying out tre decisions. Therefore, despite the consideration of decision participation descriptions as ”actual" power, these indices are also subjective measures of power, based on both potential to innfluence annd on previously enacted peer. The first index formed from this data was was called the involvement score: it was formed by counting the number of decision stages in which the department head participated, for each of the 17 issues. The stage of deciding on action received a double weight. The 76 secondinndex, anotherscopeindex,wascreatedbycomtingthenumber of issues in which the subunit participated in decision making at any stage. The third index was called the "net involvement deviation", which was formed by subtracting the number of decision stages in any issue withinthefunctionalareaof thesubunitinwhichtteunitdidnot participate, from the number of decision stages in any issue outside of the unit's function in which it did participate. Again, tne choosing action stage was weighted doubly. The last index was also called tne "net scope deviation", which was formed in the same manner as the net involvement deviation score; this index, however, counted the number of issues inside versus outside the functional area in which the unit participated at any stage. These latter two indices were considered appropriate because they tapped the extent to which a subunit gains peer over issues that are formally not its concern, by controlling critical contingencies. However, these indices are farther from the construct of power itself ; the ability to determine others' behavior is not limited to influence over only those decisions that are not logically a part of the formal function of a department. Ibgically, the more decisions a social unit can innfluece, the more ability they will have to determine others' behavior: this is particularly true in small, tightly coordinated organizations. The intercorrelations among these measures were all over .87, with the exception of net scope deviation: its median intercorrelation was .6. That measure also had far lower correlations with the questionnaire measures of power than the other interview-based indices. This was surprising, as the number of decision stages a 77 subunnit is involved in is logically connected to the number of issues it is involved in. The strongest intercorrelations across data gathering formats held for the perceived power weight; it correlated at .92 with tte participation involvenent score, at .86 with the participation scOpe score, and at .73 with the net involvement deviation score. After that, the perceived power scope score had the strogest cross-format correlations. The last two indices were considered to be aspects of "position power"; that is, tne power attached to tie formal designation of the subunits and their heads. Position scope was tl'e number of the 17 issues that the department had formal authority over; this index was taken from the interviews with the CED's. The last measure was tne position status , which was measured by dc formal hierarchical position of the head of the departnent. Vice presidents were scored as one, and the number of levels of subordination below that were cnmnulated. Neither of these measures tap power as it is defined; rather, they assess a source of the ability to determine the behavior of others. Having more decision-making authority in a department, and having greater formal authority in a subunit leader, may well grant the subunit more power. Using these indices as measures of power, however, lacks construct validity. The position status item had relatively moderate to low intercorrelations with other indices of power; the strogest was .68, with position scope. These were followed by its relationships with decision participation scope (.67) and decision participation involvemnt (.66) . Thus, it seem that the higher the hierarchical level of the department read, the more decisions he or sne participated 78 in. This did nnot mean that his/her department was necessarily perceived as having more power, or had any impact on decisions outside of his/ner functional area; these correlations were far lower. The position scope index had its strongest inter-correlations with indices of decision making involvement. It is logically consistent that having formal authority to make many decisions correlates strongly with the amounnt of involvement with all stages and issues of decision making. This index also correlated strongly with the perceived power weight score; it may be that more powerful departments were given more specific issues to decide, or having more issues to decide made a subunit powerful. All indepedent variables were strongly correlated with most power measures, with the exception of routinization and workflow pervasiveness. The patterns of intercorrelations were consistent for the weight, scope and scepe deviation indices of perceived power, and for tne involvement, some and net involvement deviation participation indices. The other indices had some strong intercorrelations with predictive variables, but they were inconsistent and often smaller than the other scores. The variable with the strongest relationship with power measures was cOping with uncertainty. This was followed by the relationships with workflow immediacy, non-substitutability and pervasiveness. To furtter examine this order, partial correlations were reported with the weight of perceived power index. These analyses indicated that the partial correlations for coping with uncertainty, holding workflow immediacy and non-substitutability constant, were over twice as strong (.66) as the partial correlations of the other two variables with 79 power: workflow immediacy had a partial correlation of .28, while the last variable had a partial correlation of .24. Thus, the findings supported the hypotheses of relations among the elements of the strategic contingencies theory of the causes of power. They also provided for detailed measures of all elements in the model. The measurement of power was undertaken with greater precision than had previously existed. mltiple issues were identified, and different amounts of power were found to exist across them. However, the relationship—specific nature of the construct was not tapped. Unlike the self report problems found by Provan, Beyer and Kruytbosch (1980) , this research found that department heads' reports of their own departments' power were fairly consistent with others' evaluations of their power. Each department head showed a slight, consistent bias toward overrepresenting their departments' power, but were otherwise consistent with other measures. Hambrick (1981) Herbrick (1981) empirically examined the sources of power, using an adaptation of the strategic-contingencies approach of Hinings , Hickson and their associates. He presented and tested several links of a cross-levels model of power in organizations, and found support for the hypothesis that characteristics of the environment moderate the relationships between functional area, scannning behaviors, and executive power. Like Provan, Beyer and Kruytbosch (1980) , this conceptualization sought to integrate a view of the causation of power in organizations with environmental considerations; where the former researchers focused on a resource-control view of power, Hambrick used the critical 80 contingencies approach. This research developed an overall theory of the way in which environments affect tie development of power and used this theory to deve10p specific hypotheses, as opposed to tne analysis of the specific environment of a given organization, and drawing hypotheses from that case. Theconstructofponerwasneverdefinedinthiswork. Its causation was discussed, but not the limits and content of the connstruct itself. I-lambrick presented a tneoretical model of tie development of individual power in organizations: some of its links were empirically tested. This model began and ended with tne power of individuals, which indicated the temporal nature of power; power affects mnany factors, which in turn affect the power of those individuals. The model consisted of three types of variables: power, tl'e sources of critical contingencies, and the bases of power. There were two sources of critical contingencies: the organization' 3 strategy and its environment. Strategy was defined as "a pattern in a stream of important decisions” (p. 255), and used the typology developed by Miles and Snow (1978) to characterize these patterns. This involved describing organizations in termns of their recurrent decision patterns; three types of decision patternns were identified. "Prospectors compete primnarily by engaging in substantial new product/ market development but pay relatively little attention to efficiency. They prosper by stimulating and meeting new market Opportunities. Defenders maintain a position in a mnarket by offering low prices, excellent service, high quality, or prompt delivery. Tney engage in little new product deve10pment. Analyzers have some of tie characteristics of each of tie two extreme strategies. They make product/ market changes more selectively and slowly than Prospectors and pay less 81 attention to efficiency and stability than Defenders.” (Hambrick, 1981, p. 256). Theenviromentwasviewedintermsof its impingingontle functioning of tie organnization. Following Hickson et al. (1971) and Katz and Kahn (1966), tl'e effects of tie environment on the acquisition of inputs, the throughput process, or the disposition of outputs were considered. The model held that individual power affects the organizational strategy, and the strategy and environment affect one anotler. The environment limits which strategies will be effective, and encourages some strategies over others; however, powerful individuals decide which strategies will be followed, by making the decisions that, over time, constitute the strategy. The strategy may change tie environment of the organization, by changing its technology or outputs. Three bases of power were identified: formal authority, individual OOping behaviors, and personal characteristics. Individual power directly determined the formal authority of the individual, which was indicated by the person's heirarchical level. According to this view, more powerful individuals will be promoted to positions of greater authority. This, in turn, increases tieir future power. Themodel heldthatbothtlestrategyandtleenwironmenthave causal effects on the behaviors which would be used to cape with critical contingencies. Two types of coping behaviors were described: thefunctionalareainwhichtlepersonworked, andtl'euseof environmental seaming. The former was seen as an inclusive index of caping behavior because it determines how well individuals are 82 "situated to cope" (p. 254) with critical contingencies of different types. Environmental scanning was viewed as a specific type of coping behavior: it allows individuals to gain information about changes in tneenvironnment, andthusrednceuncertaintyinordertoresolve critical problems. Four different areas of the environment were specified as being relevant for scanning: tie output environment, consisting of "external product/market treds or events“; the throughput environnment, which includes "external developments bearing on the processing or delivery of products/services": the administrative environnnnent, defined as "external developments bearing on tne determination of roles and relationships in the organization"; and the regulatory environment, including "government regulations, taxes, sanctions, accreditations, litigation, etc." (p. 257). The underlying notion that allows the specification of functional area as a means of cOping with uncertainty is that one's formal position has a strong effect on access to information that allows one to cope with uncertainty, and therefore deve10p poer. The corresponding notion that underlies the use of scanning behaviors in different environmental areas as a means of cOping with uncertainty is that tnere are individual differences that determine the acquisition of information that allows one to cape with uncertainty, and therefore deve10p poer. Specific hypotheses were developed from the model's proposed links between tie strategy used, the organizational environment, and the coping behaviors used by individuals: tress were empirically tested. Tne model also proposed that the three bases of poer, formal 83 autncrity, caping behaviors and personal characteristics, all caused an increase in individual power. The linnks between heirarchical authority andpower, andbetweencopingandpower, formedtnebasisof empirically tested hypotheses. Twosetsofhypothesesweretested. Thefirstsetexaminedthe effect of environment as a moderator of the poer sources-poer relationships. Specifically, in organizations operating in environments which impinge primarily on outputs, both working in an output function and scannning output environnments would be positively related with poer. In organizations Operating in throughput-focused environments , both working in throughput functions and scanning throughput-focused environnments would be related with power. The second set of hypotheses dealt with tne moderating effect of the organization's strategy on the power sources-power relationships. It was hypothesized that the relationship between output-oriented caping (both working in output functions and performing output-focussed scanning) and poer would be stronger in organizations that follow a Prospector strategy than in those following a Defender strategy. Similarly, thronghput—oriented cOping was hypothesized to be more strongly related with power for individuals who worked in Defeder organizations than for those in Prospector organizations . The environment in which the organizations existed were measured by selecting organizations in specific industries; tie dominnant environmental requirement of each industry was characterized qualitatively, from reviews of the literature and expert Opinion. Hospitals were identified as existing in a throughput-focused environment, requiring efficiency and cost cutting: insurance firms 84 were identified as existing in an output-focused environment, requiring new product/markets; and colleges were identified as existing in an ewironmentthatdidnothaveoneparticulardemand, andsoactedasa control for the other two entrees. Organizations within each of the three industries were selected for inclusion in the study based on their strategy. Organizational strategy was measured by an analysis of “quantifiable, published data” from annnual inndustry source books, and by categorizations by panels of industry experts. The coping behavior of functional area was viewed conceptually in terms of its effect on individual behaviors: tnerefore, it was measured by asking executives to rank order the role or roles that were the most simnilar to their job duties and responsibilities. Eight roles were described, which corresponded to the functions of accounting/ finance, exteer relations, general managenent, marketing, personnel, process improvement, product/ service development, and operations. This index was taken as a measure of the ability to gather information about events in the environment, and thus cope with critical contingecies. Hoever, having major responsibilities and duties in an area measures far more than the ability to gather information. It also taps the tasks oe performns and the material resources that one controls. This variable was assumed to measure only one aspect out of tie many that are part of an individual's work role. Environmental scanning was measured by three sel f-rated questionnaires, combined into one index. Tte first asked each person how frequently they learned of events or trends in each of 20 different areas of tie environment, which were specific subsectors of the four 85 aspects Of the environment deemed relevant to the study : throughput, output, administrative and regulatory. The second measure asked them to rate the extent to which they made a point of staying abreast on events in each of the same twenty categories, and the last measure askedthemtodescribethepercentagesoftimethattneyspenton scanning behaviors which were allocated to each of the four major areas. Power was measured subjectively. An index was formed by combining four questions, each Of which inquired about the amt of influece held by different executives over different kinds Of decisions. The decision areas were "determined a priori to be of major consequence to the organization” (p. 262) . These decision areas were: product/ market additions , major capital expenditures , annual Operating budget allocations, and organizational redesign. The response format was a 7- point Likert-type scale, anchored with "no influence", "moderate influece" and "total influece." For each organization, each respondent was asked to rate the power Of all Of the executives in that organization who participated in the study. Like Pfeffer and Salancik's (1974) measure, this index referred to general, as Opposed to relationshOp-specific, power. The responses of all executives in each organization were averaged. Because unreported correlation annalysis found strong correlations between the power of each individual across decision areas, a poer score was fornned by averaging the responses Of all raters and for all decision areas. Internal consistecy reliability assessments within each industry found Cronbach alphas of over .80 for tne single power score. 86 This combination Of scores across decision categories may have produced large internal consistecy reliability, but it may also have masked differences in the existence and causation of power across decision areas in different industries. In all other analyses, neirarchical level was partialled out Of all correlations due to the size Of its relationship with power: it was not partialled out Of tnese correlations. It may well have accounted for tie high inter- correlations among power over different decision areas; different causal patternns for different decision areas may have been masked by this. NO annalysis Of power over different decision areas was performed. In order to validate the power index, it was compared to an external variable that the model held ought to be strogly related: heirarchical level was used. The college and insurance industry samples both had correlations Of .71 between these variables, while in the hospital sample tne two were correlated at r = .49. Examination Of the hypothesized relationships between environment and tie development of power found mixed support. In tne output- focused industry, both working in an output function and seaming externnal product/market trends were correlated with power, even wwhenn hierarchical level was partialled out. For executives in tie throughput-focused industries, tnere were significantly smaller partial correlations between each Of these OOping behaviors and poer (eg., correlations Of -.35, -.29 and -.04) , supporting the hypotheses. In the throughput-focused industry, both having work responsibilities in a throughput function and scanning tte throughput sector had significant partial correlations. The output-focused 87 industry showed significantly smaller partial correlations between functional area and power, although the thronghput-scannning relationships were fairly close. The hypotheses regarding the moderating effect Of organizational strategy on tie poer sources-poer relationships were less strongly supported. Again, partial correlations were used, which control led for all sources of poer except the one directly related with power. Heirarchical level was consistently strongly correlated with poer, for all strategic approaches and all industries: tie partial correlations ranged from .57 to .88. Working in output functions, regardless Of industry or strategy, was related to loer power levels. For l'cspitals and colleges, the negative correlation was significantly stronger for organizations that used a Defeder strategy than those with a Prospector strategy. Thus, even trough the correlations were in the Opposite direction as hypothesized, tne differeces between them were in the hypothesized direction. Working in a throughput function had a stronger partial correlation with power in Defender than in Prospector organizations in two industries. The scanning behavior-power relationships provided stronger support for the hypotneses. Output seeming was more strongly related with power for individuals in Prospector firms than for those in Defender firms, across all three industries; only the rcspital and insurance industries, however, provided statistically significant differences in partial correltions. Throughput scanning was more strogly related with power for those who worked in Defender organizations than for 88 those in Prospectors; the only industry in which this difference was significant, however, was the insurance industry. These results indicated a sharp difference between the two ”coping behaviors". the interpretation of these findings is that the formal job duties of an encunbent do not determine their actual coping with uncertainty. The job may provide an opportnnnity to gather information, but does nnot actually relate with power. The actual performance of behaviors that acquire information allowing one to cope with critical contingencies is apparently less related with the position itself than with the individual in that position. Another interpretation of these findings is based upon the different factors that functional area actually measures. The functional area of work involves not only information available, but also the physical resources controlled by the executive, and the centrality of the individual's role in the organization. Throughput areas include accounting/ finance, process improvement and operations; these tend to be more central, particularly in organizations following a Defender strategy. All three involve controlling the flow of material resources which are vital to several other departments (eg., money for operating budgets and finished outputs for sales). This may well explain the positive relation between work in the throughput function and power, particularly in Defender organizations. The output functions include marketing and prodnct develognent; these do not involve much control over material resources that others need to do their work. This may explain the nnegative correlations between work in this function and power. These functions are more central in the work flow in Prospector organizations; this may explain 89 the smaller negative correlations between work in this function and power in Prospector organizations . This interpretation is supported by the finding that working in the general management function was consistently strogly related with high power, and working in external relations (i.e., public relations) was consistently strogly related with low power. The more central function, which involves directing and coordinating the behavior of subordinates in general, was positively correlated with power; the more tangential function, wlnich involves directing the communication with individuals outside the organization, and little control over material resources, was negatively related with power. Power was measured by averaged ratings of innfluence of organizational decision-making in four areas: therefore, its measurement was general rather than relationship specific, and inch different decision issues without differentiating among then. The single best predictor of power was the hierarchical level of the executive. Coping with uncertainty by generating information that was critical for each type of organization had significant, independent effects on eecutives' power. The organizations' strategy and environment created different conditions, which made different types of information innportant. These variables were significantly related with the power of individuals, which was measured as the general ability to innfluenoe different kinds of decisions. m The preceeding articles deve10ped the perspective that power results from dependence caused by the ability to cope with critical 90 contingencies. These studies evolved a theory and enpirical tests of aspects of the theory at both the submit and individual levels: despite several problematic Operationalizations, the relations among the key constructs have generally been supported. The research identified and supported the essential positions of the critical contingencies thesis. These hold that when there is uncertainty in organizations, and it affects the work performance of othersocial units, theywillbedependentuponthosewhoseworkor activities allows them to cope with the problems. This grants power to the problem-solving unnits. This is particularly true if no others are able to deal with the problems, and if the problem-solving capacity of the power holding unnit is closely tied with the the work of others. These positions were supported by Crozier's case analysis, and correlational relationships found annong the variables. The articles characterized by less precision in measurement had less consistent support for the basic relationships between cOping with critical contingencies and power; those with greater precision found consistent support for the relationships enbedded in the theory. Within articles, the measures with less reliability and construct validity seemed to provide inconsistent support for the theory, while more reliable and valid measures provided greater support. The construct of power was defined as intentional influece or control. It involved control over behaviors, but was often measured as control over decision-making: other aspects of the control of behavior were nnot tapped, with the exception of the general perceptual measures which allow the inclusion of all aspects of control. Power was generally defined as being relationship-specific, as it was based on 91 dependence: however, itwasnotmeasuredinsnchawayastoallovthe description.of eadh social unit's power over one another. Rather, power over all other social actors was tapped. Hinings and.his associates defined and measured power as being issue-specific. They found evidence to support the contention that power varies across issues, even though their measure was limited to issues of decision.making. Integrations of the Dependence-based Theories Several organizational power researchers have noted that the two perspectives of resource control and critical contingencies are interrelated. Bondy (1977) considered them to be aspects of one another. Salancik and Pfeffer (1974) presented control over critical resources as a means of controlling uncertainty. Pfeffer later reversed the focal concept.when he noted that “uncertainty coping capability can be defined as a critical resource in the organization“ (1981, p. 109). In recent years, the components of these theories have been explicitely integrated (Blau & Alba, 1982; chbrun, 1983: Astley & Sachdeva, 1984; Manz & Gioia, 1983). This section will first review the theoretical integrations of the theories; neither of the formal integrations presented any empirical data. It will then present the research.which.has empirically examined combinations of the two dependence-based views. Most of these articles are characterized by a 1ack.of formal specification of the construct of power; they tend to focus on its causation to the exclusion of construct definition. 92 Manz and Gioia (1983) produced a theoretical process model of power, in which they integrated the two nnain theories of the causation of dependece. The ability to deal with critical contingencies was viewed as an informational resource; control over any type of critical resourcewasviewedasthecauseofpower. Theyfocusedonthe individual level of analysis, and described the causation of power, the outcomesof itsuse, andthefeedbackof itsuse intofuturepoer. According to Manz and Gioia, power is "the ability or potential to influence others' (p. 461). In contrast, control was defined as ”the actual process of exercising influence over others" (p. 464) . Two types of control were distinguished: control over resources, which serves as a basis of power, and control over social units, which results from power. The former was referred to as resource control, and the latter as personal control. The authors briefly reviewed both the critical contingencies and the resource control research, and concluded that an important element of a given subunit's power is its ability to make other subunnits dependent on it for the absorption of mcertainty , the provision of scarce resources, or the satisfaction of other needs... A conmnon thread which runns through the usual conceptualizations of power is the existence of somedependenceonthepowerholderbecauseofa unique control over needed resources (i.e., information, peOple, funds, other key resources. Control of snch resources, therefore, serves as a basis for the possession of organizational power. (Manz & Gioia, 1983, p. 462). Themodel createdbyManzandGioiaheldthatthebasesofpower are controls over critical resources. This meanns ”access to information, people, [and] other critical and scarce intraorganizational resources . " These create dependence when the 93 resources are nonsubstitutable. This type of control grants power, which is the ability to influece others. The use of this ability is enacted through what the authors call ”bases of control:" here they referred to personal, rather than resource control. These bases were identified as the five French and Raven (1959) "types of power” (p. 464): that is, raard, coercive, legitimate, expert or referent power. Use of these bases leads to personal control, or the process of influecing others' behavior. It was hypothesized that the way in which personal control is enacted determines the relationship between the controlled and the controller. This relationship included subordinate attitudinal and behavioral outcomes, such as job involvement and satisfaction. The use of control was also presented as having a feedback effect on the amount of power held. It was posited that it is possible to relocate both resource and personal control downward, to individuals at lower heirarchical levels. This would result in increased satisfaction and commitment of employees, as well as increasing their power. This model was developed and presented as a theoretical piece: no test of the model or its component parts was suggested. Its definition of power is consistent with those previously used: the ability to influece others. It was described as being relationship-specific, and basedupondependence: theissuesorareasit includedwerenot addressed. Astley and Sacl'deva (1984) focused on the subunit level of analysis in their integration of the critical contingecies and resource control views of power. Their theoretical synthesis held that there are three sources of power that stem from the structure of the organization: 94 authority from hierarchical position, control over both informational and material resources, and network centrality. These factors cause dependence, which results in power. The two key points of this synthesis involved (1) the specification that ”the capacity for dealing with... uncertainty can be regarded as a valuable resource" (p. 106); and (2) tne nnotion that network centrality acts as a source of power in that tne ”degree of access to multiple resource flows will affect a subunit's ability to generate dependencies throngh dyadic resource exchange" (p. 106) . Network centrality was also postulated as a source of power, in that it makes a subunit ”functionally indispensable”: that is, other units require its outputs in order for then to do their work. In the first of these functions, network centrality increases a subunit' 8 poer by allowing it more Opportunities to generate relationships with other subunits where it offers resources that are needed: in the second function, centrality increases a subunnit's power by increasing the criticality of its output, and thus the criticality of sons of the resources it controls. Thus, Astley and Sacndeva (1984) suggested that. the ability to cope with uncertainty and other critical contingecies results directly from knowledge or information: this is one type of resource. Whether the resource is informational or physical, control over critical resources creates power, by engedering dependece. This occurs only when both the control and the criticality of the resource are perceived by other social actors. Critical resources are those required by other social actorsinordertOperformtheirtasks, forwhichthereisno substitution, and which no other actor can provide: the more central 95 the social unit is, the more critical are the resources that it controls. Blau and Alba (1982) took aspects of the critical contingencies approach, aswellasaspectsoftneresourcecontrolview, tothestudy of individual poer. while they did not formally integrate tne two theories, trey used aSpects of both in an empirical identification of the causes of power in a residential psychiatric center for children. This organization was characterized by a system of participative decision making, where most organizational decisions were made by committees consisting of employees at all levels of hierarchy and status. In one committee, an individual would be the hierarchical superior to another, and would be that same person's subordinate in the next committee. The researchers examined characteristics of the individual and the subunit to which they belonged, and found that characteristics of both predicted individual power. They defined power as "the amount of influenceexercisedbyastaffmenberinareasthatarecentraland problematic in the clinical treatment of youngsters" (p. 367) . Thus, poer was defined as the amounnt of influece intentionally used, which describes power that has already been enacted rather than the ability toinnfluenceintl'efuture. Theinfluencewasspecifictooeissue: problems and aspects of clinical treatment. Power was measured by subjective self reports. Ten items were used initially, which asked each staff member about the amount of influece that tl'ey themselves held, and the results were factor analyzed. Two factors were found in the questionnaire, one focusing on the tasks of clinical treatment, and the other tie tasks of child care. The six 96 items that loaded on the first factor were summed to create tre measure ofpower. Tieseitemsaskedabouttheamonmtof influenceover decisions to: modify a treatment plan: change a program or scheduled activity: obtain a psychological test: transfer a youngster from one team to another: change the content of school curriculum: and develop discharge plans (p. 368) . The response scale had four categories: "a great deal,“ "sore," "little,” or ”none." The Alpha coefficient for this measure was nnot reported. No validation of this measure was attempted. It specifically targeted one of the two main areas of decision making that were measured: more areas may have existed in the organization. Many of tie indepedent variables were separated to differentially describe access to infornmation about the two identified decision making areas. This measure of power only dealt with one of those areas: it ought to correlate more strongly with access to infornnation about that issue, even without use of the concept of power. Individuals who have access to, and seek, information on specific issues would be more likely to make decisions about those areas. This distinction was indeed fond for two of the three indices of information network centrality regarding child care as opposed to clinical practice. The researchers examined tne individual characteristics of status, as defined by civil service classification of one's occupation: the amounnt of daily interaction with patients, measured as a large or small amount: the degree of participation in decision making, measured by the number of types of decision making meetings attended regularly: the centrality in networks of working relations, measured by counnting the nunber of col leagues throughout the hospital who indicated contacting 97 tne individual weekly regarding clinical practice, and the same index of interaction regarding child care: the "dependence" upon colleagues outside of oe's work unit, measured by counnting the number of employees outside the unit tnat the individual reported contacting weekly regarding clinical practice, and tie sanne index for "dependence” regarding child care: and the integration into one's unit, measured by counting the number of colleagues inside one's work unit who indicated contacting the individual on clinical issues, and on child care issues. It must be noted trat tie questions from which the network centrality , dependence and integration indices were created did not measure different concepts: all indices were based on the same two questions, asking who a person regularly contacted to discuss clinical or child care matters . This question did not tap dependence or even one-way flow of comunication: by asking who contacts whom, it did not tapwhogave informationtowhom. Therewasnomeasureoftle importance of the infornmation discussed, or who provided information for whom. Therefore, despite its title, tne "dependence" index did not measure dependece as it is conceived of in the power literature. Instead, it measured inter-relatedness in communnication networks outsideone'sownunit: Hambrickmightviewthisasanindexofcoping by seaming tne department's environ-went. Thus the centrality in communication flows that are critical for the performance of work, and an index of centrality in environmental communication flows, were examined as predictors of individual poer in organizations . These were consistent with the critical contingencies view of power. Mentership in groups which made decisions regarding the work processes and allocation of all resources was also examined as a 98 predictor of individual power: this was consistent with the resource control view of power in organizations. The subunnit characteristics examined included the mean amounnt of daily interaction with patients: the relative degree of network centrality: the relative level of integration within a work unit for both clinical practice and child care: and the relative level of ”dependence” , or integration into other work units . Both zero-order correlation and regression analyses were perfornned. Consonant with Hambrick (1981) , they found that organizational status predicted individual power . Consonant with Salancik and Pfeffer (1974) , but on the individual level, they fond that membership in organizational decision-making committees predicted individual power. Integration in communication networks outside one ' 3 unit, focusing of issues of child care, predicted power. Integration in communication networks within the unit, focusing on issues of clinical practice, also predicted poer. Iast, for nearly seven out of eleven work unnits, mere membership in tnat unnit predicted individual power: in fact, the effect of work unit was the strongest predictor of power. Results also found that the mean amounnt of interaction with clients, and the relative inclusion in communication networks outside one's own unit, were the most important factors in determining a subunit's effect on individual power. Tne relative integration of an employee in the comunication network within the unit was the next most important factor. Thus the results supported elements of both theories of power. The greater the interaction with other members of other units, both by 99 individuals and by entire units, the greater the individual power: this may have been because interaction allows the more opportunities to develop enchange relationships of both materials and information. Membership in decision-making groups also predicted power, as did formal occupational status . The construct of power was vaguely defined in this study, and its measurement was limited. Power was not held to be relationship- specific: the nature and degree of its influece was unnspecified. The cocept of intentionality of influence was unclear. Formally, power was considered to be influece in a specific work area. Its measurement was limnited to the ability to affect decisions in a specific work area. Deciding about child care was not considered to be power, althongh it certainly may nave involved influencing the betavior of others. Thus, the measure was issue-specific, and confined to a very narrow issue. Fonbrun (1983) also performed an enpirical analyses of a combination of the critical contingecies and resource control approaches to the poor of individuals within organizations . I-le focused on ”attributions of poer" (p. 494), which he did not define, but described as the ability to influece peers and superiors. He identified two ”domains“ of poer as being relevant in the organization he examined, which were research laboratories. These domains distinguisned among types of decisions that were made, and resulted in the identification of two types of poor, administrative and technical. The former was viewed as involving "the allocation of organizatioal resources (budgets or personnel) , " while the latter 100 consisted of influece over "decisions relevant to the critical tasks of the organization, its technical core" (p. 496). Thus, inhisdistinctionof types of poer, Fonbrundividedthe domains of influence into the logical outcomes of the two different dependence-based tneories of power. Power to allocate organizatioal resonances is sinmilar to Emerson's (1962) , Pfeffer and Salancik's (1974), and Provan, Beyer and Kruytbosch's (1980) definitions and measures of the construct of power. Those researchers found evidece linking such poer to control over critical resources. Power to deternmine the way in which critical tasks get performed is similar to the manner in which Crozier (1964) , Hinings, Hickson, Penmings and Schneck (1974). Hambrick (1981) and Blau and Alba (1982) defined and measured the construct of power. Those researchers all found evidece linking the access to and the acquisition of information needed to solve critical contingecies with power. In this article, poer was not formally defined. It was not described as relationship specific, as requiring intentional usage, or as requiring successful usage. It was defined as issue-specific: in fact, formal hypotheses were developed regarding the different relationships that ought to exist between different power issues and antecedent variables. Although all of the hypotheses were not supported, different causal variables were found to relate to poer over the different types of issues. Fonbrunn hypothesized that an individual's administrative poer was determined by her formal position, because it alloed control over resource allocation. Two aspects of fornmal position were identified: hierarchical level and subunit menbership. The subunit to which an 101 individual is assigned was used as a measure of the formal context in whichanindividual existed: itwasheldtobeanindicatorofthe authority resulting from menbership in specific groups. Fonbrun also held that an individual's technical power was determined by his expertise, because it provided the ability to deal with critical contingencies. It was also held to be determined by the communication networks in which he exists, which were called the innformal context. The author further hypothesized that, for technical employees, informal network menbership would be more important than expertise in determining both technical and administrative poer. Thus, adnminnistrative power was held to be determined by position factors, because of the control over resources they engender. Technnical poer was mm to be determined by informational factors, because of the control over critical innfornmation they engender. Poerwasmeasuredbytheresponsestotwoquestionswhichasked employees to list the most influential people whom they have directly tried to influece when they wisned to affect a major decision: one question was directed toward administrative decisions, and one toord technical decisions. Individual poer was scored by tne number of nominations each person received, for each type of poer. This provided two issue-specific indices of influence over decision making. Itmaybebiased, however,duetothefocusoftheresponses toward individuals whom the employees themselves tried to influece. There may be influential individuals whom others do not attenpt to influece, because of personality, isolation of work procedures, or hierarchical level. No validation of these measures was reported, and no indices of reliability were possible. 102 Hierarchical position was measured by categorizing jobs as managerial or technical, and as occupying oe of three statns levels withineachtype. Ehnpertisewasmeasuredbyfourconponentindices: formal education: years of seniority: professional ism: and information control. None of the questions were included in tie article, and alpha coefficients were not reported. It was noted flat the constituent indices correlated more strongly with one another than with other measures. Formal subunit membership was measured by the "formally defined project structure of the division“ (p. 499): informal group membership was measured by network analysis. Individuals were grouped statistically with those who reported the highest prOportion of joint interactions. Regression analyses were performed separately on each index of power. The results indicated that nearly all of the factors examined significantly related to both kinds of individual poer. Having an administrative job was the best predictor of both power indices: the status of the job was also predictive of administrative power, but not of technical poer. Thus the first two hypotheses were supported. Formal edncation predicted administrative power, but the other conponents of “expertise" did not. Professionalism predicted technical poer, with a negative beta weight: the other ”expertise" indices did not. It is impossible to identify wl'etner the lack of impact of information control on poer was due to measurenent error, as the measures were not described. Iast, centrality in communnication networks was predictive of both kinds of poor. The third hypothesis was not supported, while the fourth was supported. 103 Within only the technical staff, status was predictive of both kinds of power. Formal education alone of the expertise indices predicted technical power: none of those indices predicted administrative power. communication network centrality predicted technical power, but not administrative power. Thus, the last hypothesis was not supported. Fombrun's analysis found that formal assignment to an administrative position was the single strongest predictor of individual power, whether regarding administrative or technical decisions: he attributed this to the control over formal resources and informational contacts such positions allowed (p. 505). centrality in communication networks was also and independently related to power over any type of decision: this variable allowed individuals to gather more information regarding uncertainties and how to deal with them. Hierarchical status was linked with power over administrative issues, but not with power over technical issues. Power over technical issues was linked with perceived professionalism, although administrative power was not. These distinctions support the notion that there are differences among the power over different issues. 5.9M The formal syntheses of the resource control and critical contingencies views all shared the perspective that power is based on dependence: dependence comes from the fact that someone controls material resources, or controls the information needed to cope with uncertainty, which are critical to the performance of work. 104 In this research, tne construct of power was geerally ill- defined. Its measurenent either used single items to tap different issues, or different items to tap a single issue: either way, tie reliability of the indices were suspect. When the construct was defined, it was most often described as being relationship specific: it was not measured in that fashion, however. It was sonetimes viewed as being issue-specific: analysis of different causes of power over different issues received sore suport. Its successful use was not required in either construct definition or measurenent. Validation of measures of poer in this literature was minninmal. However, the tneoretical views of the causes of poer, and the empirical support for them, were consistent. The complete, formal syntheses of the to depedence-based views were not empirically examined. The empirical analyses which included aspects of both theories found support for both views: factors that allow control over critical information and factors trat allow control over critical resources both related with power. Specifically, formal hierarcical superiority , holding a position in management at any level , and membership in decision-making groups were indices of control over resources that empirically related to power. Centrality in both work- oriented and non-work-related communication networks , which afford access to and control over the flow of innformation, were also related with poor. A recurrent finding was that fornmal position, measured as either occupational status or heirarchical superiority , was a strog predictor of individual poer. This was enpirically distinct from the causes of dependence: regression analyses showed indepedent effects. The next 105 section will review research that explains this finding: tress views hold that authority is a different cause of power, distinct from dependece. The subsequent section will summarize tre consistent elenents in all of the research reviewed, resulting in a definition of the connstruct of poor. Recurrent problems in tte operatioalizations of the construct will also be summarized. Authority as a Source of Power Manny researchers have examined the concept of authority as a means of explaining the control of behavior in organizations (Weber, 1968: Pfeffer, 1981: Katz and Kahn, 1978: Simon, 1972). Like power, authority deals with the potential to influece others, but tne concept of authority ”has implicit in it the notion of legitimacy or ethical sanctification" (Peabody cf. Dessler, 1986, p. 292). This section will review both theoretical and empirical research that deals with authority and its relation to poor: tie tneoretical work that develops the relations among the constructs will be presented first, followed by enpirical research that supports tie theory. Weber's Conception of Authority and Power Weber (1968) defined and explored the cocepts of poor, domination of authority. He told tl'at "poor is the probability tnat one actor within a social relationship will be in a position to carry out his own will despite resistance, regardless of tie basis on which their probability rests" (Weber, 1968, p. 53). Domination was ”the 106 probability ttat a command with a given specific content will be obeyed by a given group of persons” (ibid, p. 53). Thus, domination was a specific type of power that held for entire groups: "all conceivable qualities of a person and all conceivable conbinations of circumstances may put him in a position to impose his will in a given situation. The sociological cocept of domination. . .can only mean the probability trat a command will be obeyed“ (p. 53) . Thus, Weber's definnition of power is similar to those used in the research that followed, and was previonsly reviewed. It was relationship specific, and involved any type of social influece that resulted in getting one's wishes met: this is in contrast to domination, which was not specific to the relationships among individuals, and was limited to influence by direct comnand. Both poor and domination, hoover, involve the ability to overcome resistance on the part of the subordinated social unit: this is not partoftheconstructofpoerasithasbeendevelopedintle organizational literature. Weber then explored how domination was brought about. He held tnat all domination implied a level of voluntary conpliance, based on custom, affectnal ties, material interests or ideal motives (p. 213) . According to his theory, there were "two dianmetrically contrasting types of domination, viz., domination by virtne of a constellation of interests... and domination by virtue of authority" (p. 943): authority was defined as "legitimate domination" (p.214) . The former cause of domination was ”based upon influece derived exclusively from tne possession of goods or marketable skills guaranteedinsonewayandactinguponthecodnctofthosedominated, 107 who... are motivated simply by the pursuit of their own interests” (p. 943). Weber described the control of large central banks on the capital market and ”the domination by to breweries over the tavern owners whom they supply with their equipment" (p. 944) as examples of the fornmer: trese are situations wherein poor is caused by dependence stemming from control over nonsubstitutable critical resonances. Thus, inf luece which is derived from superiority in a dependence relationship was not considered to be authority. Weber then classified the "bases of legitimacy", which made domination legitimate. These three kinds of claims to the legitimacy of to demand for compliance were: charismatic grounds, which produced charismatic authority: traditional grounds , which produced traditional authority: and rational grounds , which produced legal authority . The first of these was normative compliance based on sone "extraordinary” qualities of the person that engenders the compliance with requests and even obedience to commands. The second, traditional authority, was based on the belief in to ”sanctity of age-old rules and poors. The masters were designated according to traditional rules and are obeyed because of their traditional status“ (Weber, 1968, p. 226). The last basis of legitimate authority was rational-legal: this is most prevalent in modern organizations . , This type of authority asserted tlat 'obediece is ood to the legally established impersonal order. It extended to the persons exercising the authority of office under it by virtue of the formal legality of their commands and only within the scope of authority of the office“ (p. 216) . Ratioal-legal authority rested on the cannon acceptance by organizational mennbers of the validity of several ideas. First, that 108 any legal norm or rule may be established, either for purposes of expediency or rationality , regardless of individual agreement with the rule; all members are bound to obey such rules. However, this covers all individuals within the boundaries of the rule. Second, the laws are corprised of a consistent set of rules which have been intentionally set, and are administered rationally and in the organization's interest. Third, that individuals in positions of authority are themselves subject to impersonal rules. Fourth, the laws mnust only be obeyed in the role of organizational member; thus, "there is an obligation to obedience only within the sphere of the rationally delimited jurisdiction" of the superior (p. 218) . daedience is nnot owed to a superior as an individual; rather, obedience is owed to the "innpersonal order“ of which both are a part (p. 218) . Weber (1968) later cautionned that "it should be kept clearly in mind that the basis of every authority... is a belief, a belief by virtue of which persons exercising authority are lent prestige. The composition of this belief is seldom altogether simple. In the case of 'legal authority,‘ it is never purely legal. The belief in legality comes to be established and habitual, and this means it is partly traditional...Furthermore, it has a charismatic element..." (p. 263) . Thus, Weber' 3 original analysis did nnot involve the relationship between power and authority; he focused on the subset of power that allowed the commanding of groups. Authority was one means of attaining that domination: controllinng resources that organizations needed in order to function was annother. Authority was defined as legitinate domination; rational-legal authority was the type most prevalent in work organizations. This was based on the shared belief that members 109 ought to comply with rules and the individuals who administer them, insofar as they occupy roles of authority. According to Katz and Kahn's (1978) review and integration of weber's work, influence that is based on authority of the rational- legal type stems from "the universal acceptance of the conceptnof legitimacy“ (p. 330). This means that individuals who enter and remain in an organization follow; and accept as legitimate, its demands and rules; these included the formal laws, legal norms, and social norms. ”Involvement in any social structure thus means that its symbols of authority and its rules, promulgated in the prescribed manner, are accepted as binding“ (Katz and Kahn, 1978, p. 301). This view'holds that individuals in organizations obey the directives of their supervisors (in areas that are relevant to their roles) because of the acceptance of their right to issue such directives. The person who occupies a position of legitimate authority may influence the behavior of others merely by requesting or commanding it. Recent Research Holding Authority as a Source of Power Astley and Sachdeva's (1984) theoretical model of the sources of the powernof subunits prOposed authority as a direct source of power, distinct from dependence and network.centrality. The authors described authority as a characteristic of both individual and subunit power; hierarchical authority was seen as a means of creating power over subordinates, and the distribution of authority throughout a subunit was viewed as a means of creating power over other subunits. This view parallels Tannenbaum's (1968) conception of organizational control as 110 being distributed throughout groups, rather than Hinings, Hickson, Pennings and Schneck's (1974) conception of subunnit authority being the issues formally under the department's purview. Bacharach and Lawler (1980) reviewed the literature on the causation and use of power in organizations . They viewed the concept of power as a heuristic term, which has not been adequately defined and limited, but serves to focus attention on phenomena of interest. These phennomena are distinguished by their 93, or the "the most basic aspects of the phenomenon" (p. 15). The form of power was held to include sanctions and dependence as essential features. The other aspect of the phennomenon is its content; this is the set of elements that are "idiosyncratic and specific to the situation. Dimensions of content vary nnot only in terms of degree but also in terms of presence" (p. 15). Authority and influence were used as examples of content. The form of power was then specified. The first aspect of its form was that power is essentially a characteristic of relationships among actors who are embedded in a larger structure. The authors held that powerisaninteractive phenomenon, andsomustbeexaminedintermsof the way in which key groups or actors interact with one annother . The secod aspect of its form was dependence. Dependence “implies that an actor's outcomes are determined by the interrelationship between his or her behavior and the behavior of others... Dependence is not a constant. It varies across relationships and settings...“ (p. 19) . Two bases of dependence were identified: the availability of other sonrcesofthevaluedoutcomes, andtheamonmtof importanceorneed for the outcomes (of. Emerson, 1962). 111 The last aspect of the'form of power was called the sanctioning aspect; this is the ”actual channges actors can and do make in each other's ontcomes... it is the active conponent of the power relationship.. . [consisting of] manipulations of rewards, punishments, or both“ (pp. 23-24) . The authors identified two differennt probabilities involved in power: the probability that sanctions would be used to overcome resistance, and the probability that the sanctions would be effective. These probabilities are subjective rather than objective. The authors described the potential for power and the use of power as analytically independennt phenomena; social actors with a great deal of power ”may have influence simply because others anticipate that party's wishes and act accordingly” (p. 25) . However, both were viewed as aspects of power. Thus, according to this view, all power is characterized by its relationship-specific nature, dependence based on control over valued outcomes that are nnot otherwise available, and the use of that control in order to overcome resistance and influence others. They then described the content of power; that is , factors that may characterize power in some circumstances. Those factors were influence and authority, which were viewed as creating two different types, or manifestations, of power. » Authority was viewed as the formal right to make decisions which others are bound to follow, and "a prime source of social control“ (p. 30): it is inherently an aspect of hierarchy, held by incumbents of higher organizational levels, and zero-sum in nature. It was seen as unidirectional, and circumscribed in terms of its domain, which is the 112 number of social units over whom the actor holds authority: its scope, which is the range of activities over which the actor rclds authority: and its legitimacy, which is the shared belief in the superior's right to make decisions, and the subordinate's mandate to follow them. It also implied involunntary submission by subordinates: even if they disagree with an order, they must conply when power of an authoritative type is used. In contrast, influence was viewed as being available to organizational merbers at any level of hierarchy, and non-zero-sum in nature. It is nnot formally sanctioned and multi-directional: it can influence benavior upward and laterally, as well as downward. In addition, it was seen as uncircumscribed in terms of domain, scope, and legitimacy. Iast, it implied voluntary submission to the influence. Bacharach and Iawler then distinguished among the "bases” and "sources” of power: the former were viewed as factors that a social actor can control that "enable them to manipulate the behavior of others,” while the latter were the means by which they ”come to control the bases of power" (p. 34). These authors viewed the source of authority as the formal organizational structure: those at higher hierarchical levels in the organization have authority over those below them. Autl'crity is “based in”, or enacted through, coercion, remuneration, norms and )mowledge. The coercive base is the control of pnnishment: remuneration is the control of rewards: the normative base is the control of symbols: and knowledge is the control of information. 113 The sources of influence were identified as an individual's personality, expertise and the opportunity to develop dependence, rather than from the organizational structure. Personality was defined as individual characteristics such as charisma and verbal skills, which provide an individual with more credibility in the kncwledge they are perceived as havinng, and in their ability to effectively invoke norms to influence others. Opportunity was seen as centrality in informal commnication networks and formal production processes: the more central , the more opportunity to create dependencies . Central ity allows influence through the control it provides over information and over slowing the production of outputs: the latter was deemed an aspect of the coercive power base. Expertise allows the control over ncrmns and innformation. These authors ' separation of power into two "types" distinguishes it from all other theoretical and empirical analyses. This separation contains an inherent contradiction between the formns , or constant characteristics, of power and one of its two types. Power's ”form" was described as always involvinng a relationship enbedded in a structure, dependece, and sanctions. However, authority was described as the determination of behavior due to widespread acceptance of superiors ' right to issue directives: that is, nnormative obedience to directives. Authority and innfluence were described as being different in several key ways: some of the distinctions drawn between them resulted in a description of ”authority" that did nnot follow the "form" they established for power. Both authority and influece relationships involve control over rewards, punishments and information: it is the normative aspect of obedience to superiors , however, that distinnguishes 114 authority from influece. Only the obedience to directives issued by superiors is circumscribed in terms of domain, scope and legitimacy. Only this normative obedience innvolves involuntary compliance, and is zero-sum. Thus, many of the characteristics that distinguish 'auttcrity" from ”influece" as types of power are only true of one its bases: the ncrmative right to demand obedience. This base of power, however, does not involve dependence. The control over rewards, punishments and information that are part of hierarchical superiority engender innfluence, not authority: they are not circumscribed in scOpe and legitimacy, they are not zero-sum, and they do not involve involuntary conpliance. They also work because of the dependence that they create. Therefore, the distinction of power into two separate types results in one type which does nnot have all the characteristics that the authors hold as necessary parts of the construct. This separation of types of power does nnot allow for an internally valid theory: however, the separation of authority from influence created by control over resources is consistent with Weber (1968) , Astley and Sacrdeva (1984) and Hambrick (1981) . Several articles that used qualitative descriptions of the development of individual power in organizations l'ave identified authority as one of its sources. These include the direct identification of authority as a cause of power (Jones, 1984) , and the indirect identification of authority as a cause of power (Mechanic, 1962) . 115 JOnes (1984) analyzed an individual case in order to understand the differences and similarities among organizational members in their perceptions of the sources of power held by individuals, the amount of power and its uses. She found that the degree of control over material resources, the authority stemming from formal position incumbency, personal respect, the ability to determine others' perceptions of reality, and control over the flow of information were all identified as sources of power. Although based upon a qualitative analysis of single case, this analysis supports the effects of control over material resources, control over information and formal authority as causes of power. Mechanic (1962) indirectly addressed authority as a source of power, by specifying factors other than hierarchical position that also created it. He defined authority as institutionalized power, and held that individuals holding organizational positions of high rank also hold authority. He used the terms ”power," "influence" and "control" interchangeably, and defined power as "any force that results in behavior that.would not have occurred if the force had not.been present" (p. 351). Therefore, power was seen as a force, rather than a characteristic of a relationship. (Meshanic held that incumbents of positions of high rank.hold "formal power”, which was described as authority: "lower participants recognize the right of higher-ranking participants to exercise power, and yield‘without.difficulty to demands that they regard as legitimate“ (p. 350). His basic thesis was that individuals who»do not have hierarchical authority may also hold power: they may apply the force required to change others' behavior. 116 This power of ”lower participants” stems from others dependence upon the power rolders. Dependence is caused, in this view, by control over access to information, persons and instrumentalities. Information was viewed as including knowledge of the organnization and its norms, procedures and techniques of Operation, as well as knowledge about peOple. Persons were defined as anyone inside or outside tl'e organization "upon whom the organization is in sore way dependent" (p. 352) . Instrumentalities were viewed as including all physical resources, including equipment, machines, and money. Specific power sources available to lower participants included expertise, nnonsubstitutability of task performance, and centrality in communnication and workf low networks . Blackburn (1981) theoretically integrated this article with the critical-contingecies theory of power, to produce a critical- contingencies model of the sources of power of “lower participants“ (LP's) . He, too assumed that incumbents of high ranked positions autonatically held power, due to their formal authority. He then proposed a model which described the control over strategic contingencies as a source of power of LP's. This model proposed that the relative expertise of the LP, the access to persons, innformation and instrmentalities of his! her position, and his/ her effort together allowed the control over strategic contingencies, which gave him/her power. Thus , these qualitative analyses identified authority as a source of power in organizations, distinguisnable from the effect of dependence on power. Three enpirical analyses of power have found support for this conception of the indepedent causal affects of 117 authority and the control over critical resources that results in dependence. Fonbrun's (1983) empirical analysis, described earlier, fond that ”an incumbent's formal position is an overwhelmingly important cue in other enployees' attributions of power” (1983, p. 503). This factor was enpirically distinct from expertise and centrality in communication networks, both of which were also strongly correlated with power. Fonbrun noted that supervisory position may be related to the control over material resources, as well as being a source of authority. Hambrick (1981) found empirical evidence tl'at hierarchically-based authority is a source of power, distinct from the control over critical informational resources . His theoretical model presented formal authority, behaviors designed to cope with uncertainty, and personal characteristics as independent sources of power. He found that hierarchical level had the strongest relationship with the power of executives. Both the functional area in which an executive worked (used as an index of the executive's ability to cope with critical contingecies) and the scanning behaviors they used to cope with uncertainty had signnificant correlations with the power of the executive, when statistically partialling out the effect of formal authority. Blau and Alba (1982) examined the effect of formal position, participation in decision-making groups, and embeddedness in communication networks. They found trat all three claracteristics were independent predictors of tie power of individuals. 118 Hickson, Hinings, Pennings, Schneck and Lee's (1971) theory of power held that authority was "regarded as that part of power which is legitimate or normatively expected by some selection of role definers" (p. 36) . Subunit authority was conceived of as the ”formally specified range of activities they are officially required to undertake and, therefore, to decide upon” (p. 36) . Authority and poer were viewed as separated by a time lag: the hierarchical authority of an individual was presented as the "fossilized impression of" past power rankings. This view of authority as legitimized power differed from the classic Weberian view in that it was limited to the nnormative expectations of a particular set of individuals who acted as ”role definers.” If individuals who were not in that group of role definers accepted as legitimate the right of a social unnit to make decisions, this power would not be considered authority by the authors. For example, if the head of the production department of a small manufacturing firm nas had the final determination of which production equipment would be purchased for the past five years, even though that decision is formally considered to be within the domain of the engineering department, this power is not authority: the role definers who establish formal responsibilities, such as the Chief Executive Officer, did not designate this decision as legitimate by the individual who actually makes it. Yet, all the other organizatioal menbers have come to accept this decision pattern as legitimate. Thus, this view of authority as formally legitimized power does not nwessarily include power that is accepted as legitimate by those whose behavior is determined by the group. 119 This view of authority as the impression of previous power was not supported by the enpirical findings of Hinings, Hickson and their associates (1974) . They used several indices of authority as measures of power: these included the number of issues a subunit was formally designated as responsible for making, and the hierarchical level of the department head. These indices had among the loest and least consistent relationships with all the other measures of power used: the department heads' formal level was the least consistent. This view was also contradicted by Crozier's findings of power in the absence of authority, and individuals holding authority roles who did not have poor. The distinction again turns upon the view of authority as power which has been accepted as legitimate by "role definers." These are generally not the social units who are affected by tnat ”authority." Thus, despite the role definers' acceptance of the "authority , " or right to direct behavior , those whose behavior is directed may not accept that authority, or may find authority where the role definers did not allow it: Crozier's (1964) supervisors and maintenance workers are examples of each of these. Authority in such situations can not be a type of poer, for no power exists. The notion that the authority hierarchy of an organization reflects fossilized power may be true in part. However, it implies that the hierarchy of today accurately reflects the power distributions of yesterday: the case of Crozier's maintenance workers and technical egineers belies this. The empirical work of Hinings, et al. (1974) , Pfeffer and Salancik (1974) , Blau and Alba (1982) and Fonbrun (1983) , indicate that, although trose with hierarchical superiority are more likely to be powerful, there are powerful individuals who do not exist 120 in tre upper echelons of tie organization (cf., nechanic, 1962: Blackburn, 1981) . m This section reviewed the work of several authors , who have identified authority as a source of power. The construct of authority is generally defined as the right to issue directives and be obeyed. The main source of this right within organizations is tte hierarchical superiority of the position held by an incumbent: superiors have tre right to direct the behavior of subordinates, within specified boundaries. This is one source of poer, which has been theoretically and empirically distinguished from dependence: it is a separate mechanism through which individuals gain the ability to determine others' behavior. Conceptually, a distinction can be made between authority as it is defined in this literature (e.g., the mutually accepted right to direct behavior) and the position trat an individual fills. Holding a position of hierarchical superiority grants one the right to direct behavior: according to Weber, so may personal charisma or tradition. Both Weber and Katz and Kahnn (1978) hold that, although rational-legal authority predominate, ”even the industrial enterprise will show all types of authority patterns" (Katz & Kahn, 1978, p. 299) . Although in most organizations trey are extremely closely related, a position of direct hierarchical superiority is not itself authority: there are other sources of autrority besides hierarchical superiority, as exenplified by Crozier's technical egineers. 121 The position itself also includes a numter of tasks and responsibilities. It may include tne right to decide how resources are allocated to subordinates, or to set policy for the communication of information: either of these may engender dependence, a distinct source of power over subordinates. Astley and Sacrdeva (1984) speCifically address this issue when they write, ”the superior's right to issue commands is enacted by a legitimized procedure that is considered by both superior and subordinate as correct. Thus, ...obedience is due a superior more because of institutionalized privilege tnan because it is instrumental in fulfilling tre dependent subordinate's needs" (p. 106) . Aside from depedence engendered from control over resources required to acconplish onne's job, most subordinates are dependent upon their hierarchical superior for sanctification of the position that the subordinate folds. Thus, in a sense, nearly all positions of hierarchical superiority create depedent subordinates: the mainteance workers and technical engineers described by Crozier were unique in this respect, because due to the conplete formalization of tte selection, pronotion and reward systems, this dependence did not exist. It may be tlat the dependence of subordinates on superiors for job incu'rbency and rewards does not exist, or is lesser, in unionized organizations. However, individuals are often dependent upon individuals other tlan their supervisor, whether for information to solve problems or for materials to acconplish their work. The constructs are related, but not symmetrical: most situations of hierarchical superiority involve somedegreeofdependence, butthere isoftendependencewithoutany hierarchical superiority . 122 Summary: The construct.of Power This literature review has undertaken a thorough examination of the construct.of power as it has been defined and as it has been operationalized. As was noted, several aspects of the construct are generally agreed upon by most researchers, while several aspects are subject to significant inconsistency across authors and within studies. When both the definitions and the measurement.of the construct are examined together, however, a number of factors become evident. Some of the commonly accepted aspects of the construct are generally not evident in its measurement. Some of the issues of disagreement prove to have important empirical and theoretical implications: others prove to add significant error to the research process, or to be less relevant to the exploration of the antecedents of power. A brief summary of the different theoretical and Operational definitions of power used by researchers is presented in Table 1. This table includes the formal definitions of the construct of power used in each article, the characteristics of power as it was described in the theoretical discussion, the operationalizations of the construct, the predictive variables for which data were presented, and the measures of those variables. The result of the review and comparison of the research is the following description.of consistent elements of the construct. For aspects of the construct on which there was no consensus, those AU'H-DRS metallic (1962) Emerson (1964) Crozier (1964) weber (1968) Hinings, Hickson, Pennings, Schneck &Iee (1971 & 1974) 123 Table 1 Definitions, Measures and Predictors of Power CCNSTRLXIT DEFINITICN C(NSTRLCT MEASUREMENT Any force that results in - behavior that would not have occurred if the force had not been present - a force to change - influence on behavior - successful influence Amount of resistance of Actor B 1) Lack of conformity to that can.be overcome by Actor A. group norm - relationship specific 2) weighting factor of score - equal to dependence of single game round - influence over any behavior 3) Cumulative score in game (lowest score=highest power) Ability of Person A to get - case analysis Person B to do what would not have done in absence of the use of power - successful influence - relationship specific - influence over behavior, decision making, psychological outcomes - intentionally used Probability that one actor in a - social relationship will be in a position to carry out his ‘will despite resistance - relationship specific - influence by any means Determination of the behavior Amt of influence over each of one social unit by another of 17 issues (subj, 17 its) - influence over any behavior 1) amt of influenceflissue - intentionally used 2) # of issues above avg amt - has 3 dimensions: ”weight,” 3) t of not "own" issues the amt of influence: 'sOOpe,' above avg amt - t of ”own” range of behaviors influenced: issues below avg amt. & ”domain”, # of social units influenced Participation in 4 stages of decision making in each.of 17 issues (subj, 68 items) AUTI-DRS Mechanic (1962) Emerson (1964) Crozier ( 1964) (1968) Hinings, et. al. (1971 & 1974) 124 Table 1 (continued) . LEVEL OF PREDICTIVE VARIABIES MEASURES OF PREDICTORS ANALYSIS Indiv Authority: formal poer ~— Control over information Control over persons Control over instrumen- talities: physical resources Dependence Indiv Resource control 1) High in-group status 2) Ability to fornm winning coalition Availability of goals 1) High status from ontside elsewhere group 2) Number of winning coali- tions available Subunit Ability to com with - (case analysis) & Indiv uncertainty Indiv Authority - Possession of goods or skills Subunit Ability to cope with Combination of uncer- uncertainty: - Uncertainty - Coping Workf low centrality: tainty measure 1) s. coping scores 1) Unpatterned variability of each of 10 org'l inputs (subj, 10 items) 2) aunt of uncertainties (subj, 1 item) # of strategies for each of 10 org inputs (subj, 30 items) - workflow pervasiveness 1) t of linnks of dept inpnts/ outputs w/ other units & actors 125 Table 1 (continued) . AUTI-DRS C(NSTRIXIT DEFINITICN Hinings, et al. (1971 & 1974) Jacobs Deliberate overcoming of (1974) resistance in another to achieve goals - successful influence - overcoming resistance - intentionally used - influence over any behavior Pfeffer & Ability to influence outcomes, Salancik changing what might have been (1974 s in the absence of the use of 1974) power - relationship specific - successful influence - influence over any behavior COWSTRIXZT MEASUREMENT 4) t of decision stages of participation 5) t of decision issues of participation 6) t of stages of not "own“ issues participated in - t of stages of ”own issues not participated in 7) t of ”own” issues partici- pated in - # of "own” issues not participated in 8) t of issues assigned to dept by CEO 9) hierarchical level of dept head -— (theoretical research) 1) 1 item subjective index: amount of power held: powers ability to influence 2) Dept. representation on budget committee 3) Dept. representation on research committee 4) Dept. representation on College executive committee 5) Dept. representation on all committees AUTHORS LEVEL‘OF ANALESIS whims. et al. (1971 s 1974) Jacobs ( 19 74) Org'l Pfeffer & Subunit Salancik ( 1974 s. 1974) 126 Table 1 (continued) . PREDICTIVE VARIABLES - workflow immediacy Substitutability Routinization Resource control Coersion resource control Criticality of resource Scarcity of resource MEASURES OF PREDICTORS 2) aunt of work connec- tions of om dept (subj, 1 item) 3) aunt of work connec- tions of other depts (subj, 1 item) 1) Speed of effect of each dept output on org ontput (subj) 2) speed of effect of dept closing on org (subj, 1 item) 3) severity of effect of dept closing on org (subj, 1 item) 1) Difficulty of replacing unit members (subjective, 5 items) & task performance (subjective, 3 items) 2) ease of replacing dept (subj, 1 item) Routineess of work (subj, 2 items) (Proposed) rating of essentialness of org'l resource X t of suppliers 1) External grant support 2) t of grad students taught 3) National rank 4) Internal research grants subjective rating (1 item) subjective rating (1 item) 127 Table 1 (continued) . AUIT-IORS C(NSTRUCT DEFINITICN (INSTRUCT MEASUREMENT Provan, Capacity to control others 1) 3 item subjective scale: Beyer & - an ability, need not be used agency ability to obtain Kruyt- - relationship specific funding from United Way bosch 2) 1 item subjective index: (1980) amt of innfluence (1980) of org over U.W. 3) % of agecy budget request met by U.W. 4) % of increase in agency budget from U.W. 5) % of agency budget funded by U.W - % of U.W. budget allocated to agency Bacharach Has 3 permanent aspects: - & Iawler - is a characteristic of relation- (1980) ships among actors in a larger structure: - dependence of one actor on another - is used poer by manipulating rewards & punishments Has 2 manifestations: - authority: inf luece - relationship specific - intentionally used Hambrick - influece Amt of innfluece over (1981) - feeds back into itself decisions (subj, 4 its) - intentionally used (1980) 128 Table 1 (continued) . LEVEL OF PREDICTIVE VARIABLES MEASURES OF PREDICTORS ANALYSIS Org'l Links with community 1) # enployees (resource access) 2) # volunteers 3) 3 clients served 4) board prestige: 8 members in Social Register & % in wealthy homes 5) board influece rating (subjective 1 item) Linnks with other agen- 1) t & size Of joint cies & United Way projects (resource access) 2) # agecy board links 3) i UW-agency board links Establishment Of agency 1) it years Operating 2) # years in UW Program effectiveness % increase in clients served Intensity Of services 5 cost per client Bacharach Indiv Sources of power : - & Iawler (1980) - organizational struc- ture: individual person- ality: expertise: Oppor- tunity to gain critical innformation or control critical work process Bases Of poer: coersion, control Of punishment: remunerat- tion, control Of rewards: normative, control Of symbols: knowledge, control Of information Hambrick Subunit & Poer Bases: (1981) Indiv - formal authority Hierarchical level Of dept. head - coping betaviors External scanning: 1) Prequecy Of learning Of events in 20 areas Of envt. (subj, 20 items) 2) Interest in info abt 20 areas Of envt (subj, 20 items) 129 Table 1 (continued) . AUTHORS CCNSTRUCT DEFINITICN Hambrick -- (1981' cont.) Blackburn - influence (1981) Blau & Amount Of influence exercised Alba by a person over central, (1982) problem areas - influece over a specific issue - intentionally used CCNSTRWT MEASUREMENT Amount Of innfluence held over decisions (subj, 6 items) 130 Table 1 (continued). AUTHORS LEVEL.OF PREDICTIVE VARIABLES ANALYSIS Hambrick Coping behaviors, (1981, cont. cont.) Sources Of critical Contingencies: - Org strategy - Org environment Blackburn Indiv Expertise (1981) Access to persons Access to information Access to instrumental- ities Control Of critical con- tingencies Blau & Indiv 8 Job Status Alba Subunit (1982) Decision making Interaction'w'clients Network centrality ”Dependence“ Integration in unit MEASURES OF PREDICTORS 3) 8 Of scanning time spent on each Of 4 areas (subj, 1 item) Functional area: ranking Of functional roles (subj, 1 item) Prospector, Defender or Analyzer (interview: expert categorization: & qualitative analysis Of org'l statistics) Industry chars (qualitative expert descriptions) Civil Service classificatn # Of decisionrmaking meets ings attended (subj) Amount (subj, 1 item) # Of others who contact himlher weekly abt work 1) t of others outside group s/he contacts abt clinical issues: 2) 9 Of others outside group s/he contacts abt about child care issues 1) t Of others inside group who contact himvher abt clinical issues 2) t Of others inside group who contact him/her abt child care issues AUTHORS Manz & Gioia (1983) Fombrun (1983) Astley & Sachdeva (1984) JOnes (1984) 131 Table 1 (continued). CONSTRUCT DEFINITION CONSTRUCT MEASUREMENT Ability or potential to - influence Others. - intentionally used Ability to influence peers and # Of times listed as influen- superiors. tial in 2 decision areas - has 2 domains: (subj, 2 items) technical, influence over decision making abt technical issues: & administrative, influence over decisions abt allocation Of resources - influence - - intentionally used - influence over behaviors & attitudes - can be relocated to lower levels Of org'l hierarchy - feeds back into itself Ability to influence behavior - & attitudes Manz & Gioia (1983) Fumbrun (1983) 132 Table 1 (continued). LEVEL.OF PREDICTIVE VARIABLES MEASURES OF PREDICTORS ANALXSIS Indiv Control over critical - resources: info, materials, money, peOple. Non-substitutability Of resources Indiv Formal position 1) managerial/technical job classification 2) status level Of position 3) formal subunit Informal position Group Of people wV frequent interaction Expertise 1) Formal education 2) Years Of seniority 3) Professionalism 4) Information control Astley & Subunit Hierarchical authority - Sachdeva (1984) JOneS (1984) COntrOl over info'l resources & physical resources Network centrality Indiv control Of material - resources Formal authority Personal respect control Of information flows 133 perspectives that have enpirical support are identified, and tncse which have been contra-indicated by such analyses are discarded. 1. Power was consistently viewed as intentional influence over the decisions, actions or psychological processes Of other social units. When focused at the level Of the individual, this is an organizational incumbent's intentional influence Of another. There was sone consensus on the Objects Of tte intentional influence. Nearly all definitions Of poer involved influence over the behavior Of groups or individuals. Many specified influence over decision making in their definitions Of power , and Others focused on behavior in their treoretical definnition, but Operationalized it as decision making. Only a few authors considered psychological processes to be a subject Of power (Crozier, 1964: Astley & Sacldeva, 1984). Behavior was included in the formal construct definitions by most authors (Emerson, 1962: Mechanic, 1962: Crozier, 1964: Weber, 1968: Hickson, Hinings, Lee, Schneck & Pennings, 1971: Jacobs, 1974: Pfeffer & Salancik, 1974: Bacharach & Iawler, 1980: Astley & Sacldeva, 1984: Jones, 1984) . Operationally, the measurenent Of influece over actual behaviors was measured only occasioally (Blau & Alba, 1982: Pfeffer & Salancik, 1974: Hinings, Hickson, Pennings & Schneck, 1974). The autl'crs who specifically defined poer as involving control over decision making inclnded only Blau and Alba (1982) and Fombrun (1983) . Just as Often, researchers developed a tleoretical view Of power as influece over behaviors, but actually measured influece over decision making (Hinings, Hickson, lee, Schneck & Pennings, 1974: Provan, Beyer & Kruytbosch, 1980: Pfeffer & Salancik, 1974). 134 2. Therewasconsistencyintleviewthatpoercanbeteldbya social unit Of any size, including individuals, subunits or organizations. Measures Of power were used and validated at all levels: the most sOphisticated work, and the measures that utilized all aspects of tl'e construct were only developed at the subnmit level. Hoever, power clearly existed across levels. T‘l'eories Of tne causation Of poer developed at one level were consistently applied to other levels, and tneir basic tenets supported. This proved problenatic for the identification Of a cannon definition of one variable, authority, when taken to tie level Of the subunit. Hinings and associates measured this variable in two ways: as the tOpics Of formally assigned decision responsibility, and as the hierarchical level Of the department head: this was called the "height Of level Of hierarchy tO which subunit is attached" (p. 32) . The former view Of subunit authority was consistently more strongly correlated with other poer measures, and is conceptually more valid as an index Of tne authority Of an entire subunnit. Hoever, at tre level Of individual power, all other predictive variables had a comnon interpretation, and the effect Of control over critical informational resources and the control over critical physical resourwsonttedevelOpnentOfdependence, anditsimpactonpoer, were supported. 3. There was consistent support for tre dependence-based approach to individual poer in organizations. From the original, experimental work on dependence as tie source Of poer (Emerson, 1962), through the specification Of the ability tO OOpe with critical contingecies as a 135 sonrce Of dependence (Blau & Alba, 1982: Manz & Gioia, 1983: Blackburn, 1982: Hambrick, 1981: Crozier, 1964), dependence was supported as a source of tie poet of individuals in organizations. Research on individual control over material resources as a source Of depedence, which then engendered power, also supported for this view (Emerson, 1962: Mechanic, 1962: Weber, 1968: Jacobs, 1974: Salancik & Pfeffer, 1974: Blau & Alba, 1982: Manz & Gioia, 1986: Astley & Sactdeva, 1984: Bacharach 8 Iawler, 1980) . 4. There was consistent support for authority as a sonrce Of individual power (Bacharach & Iawler, 1980: Jones, 1980: Fombrun, 1983: Hambrick, 1981: Astley & Sactdeva, 1984) . This was supported on both theoretical and empirical grounds, and distinguisl'ed from the control over resources through both modes Of research. It was theoretically viewed as tie accepted right Of a hierarchical superior to direct betavior, although it was most Often measured, not as a belief, but as formal hierarchical position. Bacharach and Iawler's (1980) thesis that poer based on authority is a completely separate type Of power than power based on dependence was essentially flawed: it resulted in a type Of ”power" which did not fit their own definition of the construct, nor was it supported by empirical evidece. 5. Power was theoretically viewed as both an ability to influece, and tie successful use of tie ability. Several researchers held tnat these were both viable, conceptually distinct constructs (Wrong, 1968: Provan, Beyer 8. Kruytbosch, 1980) . Many researchers tteoretically 136 identified power as an ability (Crozier, 1964: Pfeffer & Salancik, 1974: Provan, et al., 1980: Manz & Gioia, 1983: Fombrun, 1983: Astley & Sachdeva, 1984: JOnes, 1984). Nearly all of the Operationalizations of power tapped intentional social influence, and most measured the ability to influence as power (Hinings, Hickson, Pennings & Schneck, 1974: Pfeffer 5 Salancik, 1974: Provan, Beyer & Kruytbosch, 1980: Hambrick, 1981: Blau & Alba, 1982: Fombrun, 1983). Those that asked respondents to»directly describe the ”amount Of power“ defined it as ”the ability to” get.what one wanted, determine results, or influence decisions (Pfeffer & Salancik, 1974). The measures did not require that influence be successful, and thus behavior be determined: they simply required that a person or department be perceived as being able to influence another. Even the most direct measure Of the actual use Of power involved all aspects Of the attempt to influence another, rather than just the successful use Of power to determine decisions. Hinings and his associates (1974) used an index Of perceived decision-making involvement, which included a set Of 68 questions regarding the amount Of involvement.different executives were perceived as having in different stages Of decision making over various issues. One possible use of their data.would.have used 17 questions directed at the actual ability to decide what to do, and thus be successful in the use Of power. Despite defining power as influence over behaviors, however, they chose to include all aspects Of influence over decision making as part of their power indices. A number Of researchers theoretically defined power in terms of its successful use, but then measured it as an ability. (Hinings, Hickson, 137 Pennings 8 Schneck, 1974: Pfeffer 8 Salancik, 1974: Provan, Beyer 8 Kruytbosch, 1980). One researCh team empirically investigated the definition Of power as an ability to change behaviors: that is, by its successful use (Pfeffer 8 Salancik, 1974). This was done by statistical analysis of the relation between power and Objective predictors Of decisions. Although the use Of power to affect decisions was supported by all other evidence reported in that article, this analysis did not find strong support for the notion that, when defined as changing behavior, power even existed. Objective measures of enacted power only tap the use Of power that is successful: their use rests on the assumption that existing power will be used. However, these measures Often resulted in low'convergent validity correlations and inconsistent relations with other variables (Provan, Beyer 8 Kruytbosch, 1980: Salancik 8 Pfeffer, 1974). The worst results occurred for measures that involved an unwarranted assumption that power was actually used. Thus, power was most Often and most successfully measured as the ability to intentionally influence another. This is congruent with one Of the recurrent definitions. In this view; it does not necessitate successful use: power still exists if one attempts to influence another, but is unsuccessful. In such a case, the actor may have some power, but he/she does not.have enough to successfully influence another. 6. There was significant consensus on the view Of power as specific to, and based on, relationships (Emerson, 1962: Crozier, 1964: weber, 1968: Pfeffer 8 Salancik, 1974: Provan, Beyer 8 Kruytbosch, 138 1980: Bacharach 8 Iawler, 1980) .' Yet, while many researchers defined the construct this way, it was generally measured in such a way as to be insensitive to power differences across different relationships: generally, the poer Of a social unit vis-a-vis all social units at the same level was measured. The only researcher to investigate the relationship-specific power Of individuals was the experimental work Of Emerson (1962) , in which power relationships were manipnlated, but poer itself was never measured. One article enpirically measured power itself in a manner tnat was sensitive to different power relationships: this work tapped the power Of entire organizations over one another (Provan, Beyer 8 Kruytbosch, 1980) . There was great consensus on the idea that power stemmed from social units' interactions and relationships. This finding is botln consistent and thoroughly embedded in the theoretical view Of the construct. In addition, the causal treories Of power, which were enpirical ly supported, rested on the notion that power differs across relationships. Therefore it is accepted as an integral part of tie construct, which has not been adequately considered in enpirical research. 7. Most power researchers identified the construct as allowing influece over behaviors or decisions: only a few specified that different amounts Of poer may exist over different types of behaviors or decisions (Hambrick, 1981: Hinings, Hickson, Pennings 8 Schneck, 1974: Pfeffer, 1981: Salancik 8 Pfeffer, 1974: Bacharach 8 Iawler, 1980: Fombrun, 1983) . Eyen fewer enpirically measured power over 139 different issues (Hinings, et al., 1974: Fonbrun, 1983): several researchers used different issues in their measures Of power, but used them withont identification Of different power levels on different issues (Hambrick, 1981: Blau 8 Alba, 1982). However, evidece supported the notion tnat different amounnts Of power were held by the same social actor over different issues (Hinings, et al., 1974) , and even that sore Of the relations with antecedent variables differed, depending on the specific issue Of decision making or beravior influenced (Fonbrun, 1983) . Therefore, despite the inconsistency Of the inclusion Of this elenent in treoretical construct def initions , evidece indicates Eat it is a valid conponent Of power. It will therefore be included in this definnition Of the construct Of individual power in organizations. From this review, a definition Of the construct emerges. Individual power will be viewed as tie ability to intentionally influence another: others' benavior and decision making are the main subjects Of this influence. The amount Of this ability is specific to different relationships and content issues. Every influece attenpt neednctbesuccessful forpowertobeheldbyoepartyovertle other. Power stems from dependence and authority. Summary: The Antecedents Of Power The review Of the literature indicates that several variables have beenlinkedtothedevelOpmentOf power. Twotypes navebeen identified and distinguished by several researchers: dependence and its sources, and authority and its sources. Dependence was viewed as 140 needing something that is predominantly available from one specific person or group. It can be caused by having or controlling the informational resources tnat others need in order to solve problems tnat stop their work: that is, information needed to deal with critical contingecies. It can also be caused by controlling the physical resources tl'at others need in order to acconplish their work. Several kinds Of critical information rave been identified. Hambrick (1981) identified awareness Of events in the environment as critical information, while Hinings, et al. (1974) viewed departments' ability tO cope with unpatterned variation Of organizatioal inpnts as tie critical innformation. This coping may be accomplished by routinization Of problem-solving procedures , gathering more innformation, predicting the variation, or dealing with the problem after it has occurred. The kinds Of critical physical resources trat exist have not been subject to the same intensity Of analysis. Salancik and Pfeffer (1974) identified resources that were related to the power Of University departments: these included external grant monies, internal research grants, the number Of graduate students, and the national rank Of the department as non-informational resources. Hinings and his associates identified sets physical inputs into both organizations and subunits which were needed for work: tnese were not listed, but aggregated for themeasurenentof uncertaintyinthominnputs, andother such constructs. None Of these, however, provides a framework for describing critical material resources which is applicable across jobs or industries. 141 Dependence is produced when the resources controlled are important. This is determnined by the ability Of the social unit to substitute another resource or to do without it (Jacobs, 1974: Provan, Beyer 8 Kruytbosch, 1980: Hickson, Hinings, Lee, Pennings 8 Schneck, 1971) . Workflow centrality allows one to control resources trat others are more likely to need (Fombrun, 1983: Manz 8 Gioia, 1983). The actual control over resonroes, whether innformatioal or material, has been viewed as being acconplished in two separate ways. Some researchers focused on the ability to Obtain critical resources as predictive Of power. Salancik and Pfeffer (1974) , Jacobs (1974) and Provan, Beyer and Kruytbosch (1980) focused on the acquisition Of material resources as being key in creating dependence. Likewise, Hinings and his associates (1974) , Hambrick (1981) and Crozier (1964) focused on the acquisition Of information needed to cope with uncertainty as being key in creating dependence: the former viewed environmental scanning as meanns Of acquiring such information, and the latter described tne acquisition of tie knowledge necessary to fix machinery as being acconplished by powerful department members. Thus, acquisition is one means Of controlling resources. The other technique identified in this literature as a meanns Of controlling critical resonrces is by deciding the flow Of a resource. Crozier (1964), Fonbrun (1983) and Blau and Alba (1982) focused on the ability to control where critical information goes as being key in creating dependece: Crozier specified limiting who has access to the information, while the latter two identified centrality in networks Of communication. Pfeffer and Salancik (1974), Provan, Beyer and Kruytbosch (1980) and Fonbrun (1983) also identified that tie control 142 over the flow'of material resources, such as money, was related to power. In summary, researchers have identified two types Of critical resources, and two ways Of controlling them. Materials and information are both types Of resources, which become critical when other social units in an organization require them in order to do their work. They can.be controlled by their acquisition from external sources, or the ability to control where they go and who else has access to them. The second source Of power identified by theoretical and empirical research is authority (Mechanic, 1962: Waber, 1968: Bacharach 8 Lawler, 1980: JOnes, 1980: Hambrick, 1981: Blackburn, 1981: Blau 8 Alba, 1982: chbrun, 1983: Astley 8 Sachdeva, 1984). Authority is the shared perception that someone has the right to issue orders and expect compliance. In work organizations, the most common basis Of this is rational-legal: the formal organizational hierarchy embodies the concept that those higher up may direct the behavior Of those farther down. Formal authority was empirically found to have significant relationships with power, independent Of the effect Of dependence or the control over critical resources (Hambrick, 1981: Fombrun, 1983: Blau 8 Alba, 1982). Summary: Operationalizing Power Many problems are evident in the previous Operationalizations Of the construct Of individual power in organizations. These will be grouped according to the typology described in Figure 1 at the beginning Of this chapter. 143 Subjective Measures Of Power Using perceptual measures Of poer has several coceptual limitations. First, they rely on respondents' inferences about reality: if a social unit's power stems from its covert control over communication flows, others may not be aware Of the influece being used. Another criticism of trese measures is that trey rely upon the organizational members to report honestly their perceptions in an area where there may be a great deal Of positive and/or negative evaluative loading. However, perceptual measures most directly tap tre construct Of power as it has been developed: tie capacity tO influence others, rather than the enaction Of that ability, or the formal control over resources tlat "ought to” enable one to control others. In the research reviewed, most of these measures used a questionnaire format to tap perceptions Of poer. Of those, several used single items tapping the amount Of overall poer held by a social unit (Salancik 8 Pfeffer, 1974: Pfeffer 8 Salancik, 1974: Provan, Beyer s. Kruytbosch, 1980) . To the extent that multiple item measures Of ' perceived power have been used, trey have been parallel itena that inquire about the amount Of poer held over different functional areas or types of decisions (Blau and Alba, 1982: Hambrick, 1981: Hinings, Hickson, Pennings and Schneck, 1974: provan, Beyer and Kruytbosch, 1980: Fonbrun, 1983) . Nearly all Of ttem used parallel items that used the "howmuchinfluencedoesthisunithave' wording, andmostused Likert-type ratings . 144 When multiple items were used to measure power over specific content areas of work or decision-making, they have inquired about content areas that are Specific to the organization under study. Therefore, each study using such questions has used a different measure. None of the questionnaires is applicable across different types of work. Statistical assessments of multiple item measures indicate that they generally have high internal consistency reliability. Inter-rater reliability has also been generally high, both for single item and multiple item measures, particularly for members of the same group or organization. Pfeffer and Salancik found "enormous consistency” in the responses among university department chairs (Pfeffer, 1981, p. 59) . Hinings, Hickson, Pennings and Schneck (1974) found that the perceptions of the power of four departments "tallied overwhelmingly" across department heads (p. 31). Of course, inter-rater reliability could not be assessed unless different raters described the power held by the same social units; this was not often the case. On the other hand, there may be far less consistency in perceptions across organizationns . The lack of consistency across raters and measures fournd by Provan, Beyer and Kruytbosch (1980) may be attributable to bias caused by the wording of the questionnaires, or to atmedifferenceintheperceptionsofpowertratexistsacross organizational bounndaries . The most conservative estimate would be that perceptions of power are consistent within a given group, but may not be shared across organizations. It also seems tl'at self-report measuresofpowertendtowerdsleniencyerror. 145 Objective Measures of Potential Power Theoretically, objective measures of potential power are non- subjective indices of the ability to influence others, rather than the enaction of that ability. They have been proffered as preferable to subjective measures because they are not open to response bias. HOwever, there is substantial confusion in their interpretation, which makes their use in the assessment of either the sources or the results of power problematic. Pfeffer and Salancik (1974) used several objective indices of potential power; this measurement technique led to a number of problems in the interpretation of results, which were previously described. Departmental representation on key committees was viewed as both enacted power and potential power. It was an index of enacted power because it reflected the use of power in the decisions to appoint representatives to these committees. It was an index of potential power because it measured the degree of control held by the department over deciding the flow'of critical resources, such as research grants. These measures were linked to a number of other variables, which were presented as either causes or results of power. These included the allocation of University funds and the distribution.of internal research grants. lather variables included the number of graduate students and preportional funding by external grants. The former were held to be the outcomes of other decisions where power was used: the latter were held to be sources of power. It is clear that the number of graduate students is determined to a large extent by the internal and external funding; one of the external grants that was an.”outcome" of power was dedicated to graduate student support. Thus, measures 146 whichtapbothsourcesandresultsofpeerarerelatedtottcse specifiosourceswtnicharedeterminedbytheoutconesofpeer. The results are not useable as evidence of causality. Another objective measure of potential peer that has been used is the hierarchical authority of an individual. This variable tas been seen as an index of potential power, as it provides the ability to influence others through formal authority (Hambrick, 1981: Blau & Alba, 1982) and through the control over resources (Fombrun, 1983). It has also been interpreted as the "fossilization" of previous peer, and henceanobjectiveindexofpowerthathasbeenusedtoinfluence decisions (Hickson, Hinings, Iee, Pennings & Schneck, 1971). Indices of authority have been consistently linked to power; the analyses have involved using authority to predict power in regression equationns, so most of the resulting interpretations have involved authority as potential peer. Its correlations with other indices of peer have been leer than other measures (Hambrick, 1981; Embrun, 1983; Blau & Alba, 1982) , and its patterns of relationships with other determinants of power have been inconsistent and not as supportive of hypotheses as subjective measures (Hambrick, 1981; Hinings, Hickson, Pennings 8 Schneck, 1974) . Provan, Beyerandlcruytbosch (1980) usedanindexof ”net dependence" as an objective measure of potential peer. This was a relative measure of the preportion of agency funds that came from the UnitedWay, comparedtotheproportionof UnitedWayfundsthatwentto the agency. First, this made the mistake of using the allocation of resourcestg, ratherthanfrom, asocial actorasanindexofthe dependence upon that actor. In addition, it epitomized the conceptual 147 problemlwith objective indices of potential power: they attempt to measure dependence or its causes, rather than the ability to influence. All of the objective measures of potential power lack.construct validity; they reflect conceptual confusion as to what power is. Power is the ability to intentionally influence other social units: almost all of these indices actually measure the extent to‘which the social units have access to various sources of power. It assumes that the control over various resources and formal positions automatically engenders the ability to determine others' behavior. Objective Measures of Enacted Power Objective measures of enacted power involve using non-subjective indices of power that has been used to influence others; they use the outcome of the enaction of power to determine how much power parties have. There are several conceptual criticisms of the use of these measures. Pfeffer (1981) holds that these measures may not actually reflect the use of power, but instead, the ability to predict the way decisions will be made, and align.oneself with the winning side (pp. 44- 47). Using enacted power as a measure of power also assumes that the ability to influence others is always used. This has been contradicted on both theoretical and empirical grounds (Provan, Beyer & Kruytbosch, 1980; Kipnis, 1976: Pfeffer, 1981). Some Objective measures of enacted power have shown consistent relationships with.other indices of power, such.as committee representation (Pfeffer & Salancik, 1974). Unfortunately, they often resulted in difficulty interpreting the relationships found among variables, as described above. 148 Other objective measures of enacted peer have not denonstrated significant correlationns with one another, even within the same decision area (Provan, Beyer 8 Kruytbosch, 1980) . It seened that the more assumptions implicit in the measure about how existing peer has beenused, themoreopportunity forthemeasuretotapotherconstmcts besides peer. For example, the measures of the percentage of budget requestswhichweregrantedtoanagency may have tappedthetactics used by the agency to influence the United Way (e.g., connsistently requesting a larger budget than they wanted, and consistently increasing their budget requests). In general, the intercorrelations between objective and perceived measures of peer have been lev. Provan and his colleagues (1980) found intercorrelations between .00 and .42 among perceived and enacted measures of power; of these, only two out of ten were statistically significant. Pfeffer and Salancik (1974) found much higher correlations among the two types of measures; all were significant, ranging in size from .42 to .66. Heever, convergent validity indices were far leer than many of the relationships between the objective measures and the distribution of resources, the ”outcome" of peer. The empirical evidence suggests that the two types of measures tap different constructs. The key limitation of enacted measures of peer is their lack of construct validity; they do not measure power itself, but either its outcome or its sources: often they tap variables, such as formal organizational roles, which act as both cause and effect. 149 General Problems in the Measurement of Peer Twoaspectsoftheconstructofpeerhavenotbeenaddressedin its measurenent. The relationship-specific nature of peer has not been well assessed by existing measures, whether perceptual or objective. Only one team of researchers out of all those exploring the topic alleed the specification of different amounnts of peer for different relationships (Provan, Beyer & Kruytbosch, 1980): these researchers assessed the power of each agency over the United Way, rather than their power over all other agencies. This was done with both perceptual and objective measures; the latter were severely flawed in concept. Those relationship-specific measures which indicated validity on connstruct and convergent grounds, lcwever, were single item indices; hence their reliability is unknown. They tapped the power of organizations over one another . There were no individual level measures of peer that were sensitive to peer differences across relationships . Thesecondaspectoftheconnstnctofpowerthathasnctbeen adequately addressed in its measurement is its issue-specific nature. Some researchers have Operationalized peer in such a way as to only allow the identification of one issue, either with multiple item indices of one decision area (Blau & Alba, 1982) or with single item indices of power (Pfeffer s. Salancik, 1974: Provan, Beyer 8 Kruytbosch, 1980) . Several researchers have used multiple-item indices of power, particularly in regards to influecing decision making in different areas (Hinings, Hickson, Pennings & Schneck, 1974; Hambrick, 1981: Fombrun, 1983): one measure even alleed for the identification of the 150 number of different areas wherein an actor had influence over the decisions made (Hinings, et al., 1974). Most of these data were aggregated across all decision areas for analyses of power (Hinings, et al., 1974: Hambrick, 1981). With the exception of one study that analyzed power over two different types of decisions separately (Fombrun, 1983), they failed to identify the differential relations among power over specific issues and other variables. Yet, when the power held over different issues was described, social units had different amounts of power over different issues (Hinings, et al., 1974). These measures had several other failings. Fbmbrun's (1983) two- issue distinction used two single item measures, making reliability uncertain. Hinings and associates (1974) multiple item measures were quite reliable, but they were specific to the work performed in those particular organizations. The issue-specific measures were developed from organizational inputs into those particular manufacturing organizations, and would not be applicable to organizations in other industries. In addition, the Hinings, et al. indices measured the power of subunits. No reliable, tOpic-specific measures of individual power have been reported. An additional limitation of the existing multiple-issue measures is that they all were limited to one specific type of behavior: decision making. .Multi-factor indices of power over other types of behaviors were not.evident. The conceptualization of power holds that it is intentionally used social influence that may vary across issues and across relationships. 151 None of the individual or subunnit measures included both of those characteristics of the construct. SEE This chapter reviewed the literature on the sources of peer in organizations, focusing on the definition and measurement of the construct of peer, and the identification of its sources. The aspects of the construct on which there was little consensus, either between or within studies, were resolved. The direct sources of peer were shown to be authority and dependence. Several sources of dependence were identified: the integration of the two major views resulted in the identification of the control over critical physical resources and control over critical informational resources as the two key variables. Recurrent limitations in the measurement of power were identified. The next chapter will formally integrate all the sources of peer specified in this chapter. These sources of power will be described, along with two other variables which have been found to relate to the sources of peer. The relationships among all variables will be specified: this will constitute the theoretical links of a nnomological nnetworkoftheantecedentstotheconstructof individual peerin organizations. Three individual difference variables which relate directly and differentially with power will then be disenssed. III. THEDRETICALLINKS INAWICALNEI‘OF POWER This chapter uses the information presented in the preceding review of the literature on the etiology of peer to deve10p theoretical links in a nomological nnet of individual peer in organizations. First it describes several constructs which have been posited as antecedent to peer in organizations. Next it introduces two demographic variables that rave been related to one of the antecedents of individual peer. Then it specifies the existece and strength of the links among all of these constructs: hypothesized direct and indirect predictive relationnships are differentiated into primary and secondary links. Iast, three individual differece characteristics which have previously been related with individual peer, but not its antecedents, are presented. A substantial amount of research suggested that the predominant causal mechanisms of individual power within organizations are authority and others' depedence upon the peer holder. The literature reviewed indicated that dependece stems from control over critical material and informational resources. 152 153 Constructs in the Nomological Net Peer Power is defined as the ability of an individual to intentionally influence the behavior or decisions of another. Because an individual can have a great.deal of power over one person, and little power over another, the individual's power is relationship specific: it is the amount of this ability held by a given person in reference to any other individual. Power is also issue-specific: a person can have power over one aspect of someone's behavior and not have power over another aspect. Power is thus further specified as the ability to influence different aspects of a person's behavior or decisions. Individual power in organizations focuses on the power of organizational incumbents to influence one another. This includes individuals at all organizational levels, and the amount of influence they hold over others at all I levels. Dependence Dependence is the extent to which an individual is needed in order for other organizational members to function effectively at work. Effective functioning includes both effective job performance and job maintenance. Effective job performance is defined as the successful completion of tasks that constitute an.incumbent's jObw Dependency on another may include that person's provision of sufficient information to complete all tasks, providing access to adequate material resources to complete all tasks, ensuring the delivery of all work inputs 154 necessary for task performance, providing information regarding the nature of all task requirements, or any other element of job performance. JOb maintenance is defined as continued job incumbency. A person can not function effectively at work if she has no job. Dependency upon another for job maintenance involves any basis of need for that person in order to maintain a position, including continued acceptance of the adequacy of the work level and necessary assistance in maintaining an adequate work level. The term "dependence" will be used to refer to the dependence of an organizational member on the power-holder. In every relationship, both parties have some degree of dependence upon another, even if that degree is zero. In every relationship one individual, person.A, can arbitrarily be specified as the power holder: his or her power over person B would therefore be the focus of that power analysis. This does not negate the power of the person B over person A, which may be stronger or weaker than A's power over B: it is simply an arbitrarily chosen focus. Dependence will be used to refer to the dependence upon person B on person A. Relative Control over Critical Material Resources This construct describes the relative control over critical material resources held by an individual. Material resources are considered to be any resources required for work that have conservational preperties, such that the more of a resource that is transferred from one location or person to another, the less is left in the original store. Material resources are distinguished from physical 155 resources, in that the former need not be manipulable physical entities. They include factors such as money and the work of individuals, along with manipulable physical entities snch as typewriters, desks and computers. Critical resources are those resources that are necessary for effective functioning at work: that is, any resource that is necessary for either effective job performance or job maintenance. This includes financial aspects such as monies required to purchase supplies or services needed for effective performance, physical resources such as computers, and supplies such as paper. The more essential the resource is for either effective job performance or job maintenance, the more critical it is. The actual control over those resources incltdes either of the two types of control that were identified in the previous chapter: acquisition of the resource , or determination of the flow of the resource. If person A can get access to superior equipment, which person B can use to perform work more effectively, this is control by acquisition. If person A can decide how long each enployee nas access to the equipment that person B requires to perform his job, this is control by resource flex determination. The relative central over those critical resources held by an individual is the focal variable: this takes into accounnt the availability of the resource elsewhere. Relative control means the amount of control over a critical resource held by an individual, compared to the control held by all other available sources of that resource. Despite complete control over the use of a key piece of equipment, the relative control will be low if others can acquire or 156 gain access to another such machine. Thus, person B will not be depedent upon person A, and the control over tl'e resource will not leadtopeer. Relative Oontrol over Critical Informational Resources This construct is the relative control held by an individual over information trat is critical in the effective performance of another's work, or to her job maintenance. It is parallel to tie previous construct, but refers to informational resources. Informational resources are any resources that are not conservational in nature, such tnat transferring some to another location or individual does not deplete tne original store. It includes any job-relevant information, such as tl'e knowledge and skills necessary to deal with non-routine events or problems that require solution in order for others to perform their work: an example of this is the ability of mainteance workers to fix disabled machines in a highly mechanized plant (Crozier, 1964) . It also includes information allowing one to owe by prediction, such as scanning the work environment in order to predict future demand for services . Information is critical if it is needed to cope with uncertainty that prevents the successful completion of job tasks. Just as with material resources, control over tne innformational resource may exist in two modes: acquisition or the determination of flow. Acquisition of critical information is exenplified by learning nowtofixtrecopymachinewnenitbreaksdown, orbyscanningtne environment to learn production plans, and therefore predict the workload for tie next monnth. Controlling the flow of innformation can 157 be accomplisted by being central in communication links, and thus able to disseminate specific information to everyone, or by limiting others' access to information. Examples of trese are altering tie information that is verbally communicated through a work group, and ”losing“ manuals that contain instructions for fixing tie copy machine. The relative control of this type of resource focuses on the control of person A relative to the control neld by all other available individuals. If an employee is the only person in a building with the knowledgeofhowtofixthecopy machine, andothers requirethecopy machine in order to do their work, this person will have a high degree of relative control over a critical informational resource. If one person on every floor knows how to fix the machine, this individual will have a lower degree of relative control. On the other hand, if every other person is a my machine mechanic, but they all are at lunch, on vacation or in a meeting, the individual will have a high degree of relative control: the others with control exist, but are not available and hence do not decrease the relative control of the resource. Autlorig Autlority is the belief held by an organnizational member tlat another member l'as the right to direct his/her belavior. This is a direct source of power, indepedent of the perception of dependence: if an individual believes that tel she must comply with tie wistes of another within a specific sphere of directives, the other will be able to intentionally influece him/her. 158 Autnority in organizations is determined primarily by hierarchical superiority over another. It is generally limited to a belaviors tnat are relevant to the performance of work or personal conduct at work. Other Antecedents of Peer T'ne preceding constructs have been described theoretically and Operatioally in the literature review. They have all been linked with peer itself, and evidece supports significant and fairly strong relationships among most of them: this will be described in detail later in this section. There are many constructs trat are tlought to relate to the sources of peer described above. Some of these are denographic characteristics, which are easily measurable. Two tlat have been empirically or theoretically supported are tenure and geder. £8223 Tenure is theoretically and logically linked with two of the antecedents of individual peer in organizations. Mechanic (1963) suggested that organizational tenure would lead to greater peer of leer-level enployees. According to Mechanic, peer is determined by access to and control over people, instrumentalities and information: organizational tenure allows one greater access to peOple and information. In this formulation, tenure is related to the relative control over critical information, in terms of control by acquisition, creation and determination of its f low. 159 Blackburn (1981) holds that ”one cannot gain peer at any level within an organization unless one is a meter of that organization. Hence, organizatioal membership of some duration” became tl'e starting point of his model of tie peer of lower-level enployees (p. 128) . The formal model rolds that organnizatioal tenure has direct causal links to tie relative effort and expertise that allow the control over critical contingecies tnat create peer. Thus , Blackburn' 8 formulation supports tne relationship between tenure and relative control over critical information proposed by Mecl'anic' 3 work. The existece of this relationship is thus firmly grounded in two of the theoretical works on the sources of power. T'ne size of the relationship, Iowever, is not Specified by tre tleory or by previous research. From a logical perSpective, the longer an enployee's tenure in an organization, tte more major problems trey are likely to rave experienced. They have an opportunity to observe how to solve such unplanned, unpredictable problems: this is critical information that allers them to cope with these contingencies. However, all enployees will not desire to or nave tie opportunity to acquire such innformation: both Blackburn and Medanic posit effort and motivation to acquire information as mediating variables in the development of power. Therefore, a weak relationship ought to exist between organizational tenure and control over critical information. Tenure is also logically linked with authority. Most organizations have some procedure for promoting successful enployees into higher level positions. The longer an enployee works in an organization, therefore, tie greater tie opportunity for promotion exists. Similarly, although some higher-level employees may be recruited from 160 outside the organization, a large pr0portion of them are likely to be promoted from within: therefore the average number of years of tenure in an organization is likely to be higher for employees at higher organizatioal levels than at leer levels. This means tnat a strong relationship ought to exist between organizational tenure and hierarchical position, and thus with autnority . eager Many researchers have examined tne relationship between the gender of an individual and his/ her formal authority: the relationship between being male and hierarchical superiority is well establisned. As of 1984, 33.6 % of all executives, administrators and managers in the united States were women (Blau & Berber, 1987) . Within this broad category, the preportion of women in upper level positions is even leer. "Accorng to one estinmate, 49,000 men and only 1,000 women held top policy-making jobs in major corporations in 1984" (Dipboye, 1987) . Some women who occupy jobs with titles indicating high levels of hierarchical auttority do not actually have it: a recent Gallup poll of women with the title of "vice president” found that, of the sample of 2000 who held that title, there were only 800 who actually held jobs at that 1eVEI (Dilboye, 1987) . When men are pronoted, it is more likely to be into positions of greater formal autl'ority: evidence indicates ttat this is not true for women. Stewart and Gudykunst (1982) found a significant correlation (r a .41) betweentlenunoerofpronotionsreceivedandtl'ehierarchical level attained by managers for men, but not for women (r = .03). ”After controlling for variance due to length of tenure, age, and years 161 of education, female enployees were found to receive more promotions than males, but to occupy leer positions in tne hierarchy tnan their male counterparts" (Dipboye, 1987). Within academia, women consistently occupy leer hierarchical and status levels than men. As of 1983, 10.6% of all full professors were women: 21.9% of all associate professors were female: 36.0% of all assistant professors were women: 52.1% of all instructors were women: and 37.1% of all lecturers and full time faculty of no academic ranks were women. The higner status academic positions are overwhelmingly neld by men. The tred is the same for academic administrators. As of 1983, there were 3,084 women in senior administration of accredited colleges and universities in America (including deans, directors, chief officers and vice presidents): as there were 2,824 institutions sampled, tnat averages 1.1 ferale executives per institution (Bognannno, 1987) . There is a clear, strong relationship between gender and hierarchical authority . The causes of this relationship are many and varied: tne relationship is consistent. Relationships Among the Antecedent Constructs and Peer A substantial amounnt of research suggests tl'at tre peer of individuals in organizations is engendered by tneir autnority and others' depedence upon them. The literature reviewed in tie previous chapter indicated that depedece stems from control over critical material and informational resources. Being male is associated with hierarchical autnority . Organizatioal tenure leads to increased 162 control over critical informational resources: it also is associated with hierarchical authority. The literature on the development of peer has examined all of trese constructs. Hoever, tie interrelations among all of them nave never been specified. Dependece and authority have been proposed to rave direct causal links with peer in mnuch of do existing research: these relationships have been examined the most frequently, and have the greatest amount of empirical and theoretical support. Tiese are considered to be primary links, because of the relative degree of certainty in their existece and size. Secondary direct linnks involve other antecedent constructs. A graphic representation of these constructs and flair direct links is presented in Figure 2. The lines between constructs depict linear relationships among them. Strong relationships are depicted with double lines and weak relationships with single lines. Table 2 indicates the size of the hypothesized relationships among all constructs, including both primary and secondary direct linnks, and indirect relationships: the latter will be described in the subsequent section. All existing relationships are hypothesized to be positive, with tie exception of geder (using an arbitrary designation of being fenale as "higher') . Direct Links Anong Antecedent Constructs As described in the literature review, peer has been theoretically and enpirically linked with both autl'ority and depedece: these are the primary links in the nomological net. Those who told authority over an individual will be able to direct tneir conpliance, and thus 163 Gender 1...; Authority Tenure 7// Control over Critical Materials \ mmence Control over Critical Information strong link weak link Figure 2 . Direct Linkages Among Antecedents in a Nonological Net of Power. 164 Table 2 Relations Among Antecedent Constructs in a Nonological Net of Peer Peer Autnor Deped CO! OCI Tenure Gender Power - Autl'or § - Depem s s - 004 (S) g _S_ - con (8) w s o - Tenure (S) § (8) (8) fl - Gender (S) g (S) (S) (W) 0 - Author = Authority Depend =- Dependence CCM = Relative Control over Critical Material Resources CCI = Relative Control over Critical Information strong relationship weak relationship no relationship ozon ll u ll _ = direct relationship ( ) 8 indirect relationship 165 intentionally influece them: peer is therefore hypottesized to have a strong positive relationship with authority. This relationship has been enpirically supported by several researchers. Hambrick (1981) found the hierarchical authority of executives to be the best predictors of their peer: tte correlations between do two variables were r s .71 for two industry samples, and .49 for a third. Blau and Alba (1982) found a correlation of .48 between these variables. Fonnbrun (1983) found tnat being in management, with the autlority implicit in such positions, was the strongest predictor of peer: within non-nuanagenent jobs , hierarchical position was the strongest predictor of individual peer . Dependence upon a peer holder is hypothesized to have a strong relationship with tlat person's peer. Following Emerson (1962) . Salancik and Pfeffer (1974) , Hickson, Hinings, Lee, Schneck and Pennings (1971) , and Jacobs (1974) , depedece is viewed as an immediate, direct determinant of peer. Individuals who are depedent upon anotner for desired outcomes which cannot be attained elsewhere will be responsive to the directives and desires of tnat person. This hypothesis is not supported by empirical evidece. Despite the many theoretical fornnnulations of this relationship, and do enpirical analyses of theories based on tte notion, tte measurement of tie construct of depedece itself nas only been attenpted once (Provan, et al., 1980) in an index of questionable construct validity. The consistecy of the strong theoretical links between peer and depedece, and He strength of the ennpirical links between all of the antecedents of depedece and power indices, are both quite substantial: trey form tl'e basis of this hypotlesis. 166 The integration of resource control and the critical contingenncies theories of intraorganizational peer results in the next set of hypotheses, which are secondary level, direct links. Dependence results when the individual has a high degree of control over either critical material resources or critical informational resources , relative to others' control over such resources. This produces two parallel hypotheses. First, dependence on the power holder is strongly related to that individual's relative control over critical material resources. Second, dependennce on the power holder is stronngly related to that individual's relative control over critical informational resources. These hypotheses are directly supported by the theoretical literature, and indirectly supported by the empirical literature. Virtually all of the research reviewed in the preceding chapter supported the underlying thesis of one or both of these hypotheses. As was nnoted above, the empirical research has not involved the formnal measurement of the construct of dependence. Heever, the ability to cope with uncertainty was empirically correlated with peer, the proposed direct result of dependence, by several authors. Hinings and associates (1974) found a relationships betweenn .44 and .81 between indices of thom variables. Hambrick (1981) found eight significant correlations out of twelve, ranging from .20 to .38, between a measure of acquirinng information to cope with uncertainty in the environment and peer. Blau and Alba (1982) found strong correlations between centrality in work-related communication networks and power: they also found that frequent communication with those outside one's work group significantly predicted peer. Pombrun 167 (1983) found that centrality in communication networks was a significant predictor of power over two different types of issues. Similarly, the relative control over critical material resources was also strongly related to power. Salancik and Pfeffer (1974) found correlations ranging fromn.36 to .77 between the acquisition of external funding and measures of power. They also found very strong correlations between allocations of internal funds and measures of power: the correlations ranged from .35 to .90 for the most critical of these resources. Despite the probable influence of common method variance in the measure that produced the strongest of these correlations, the other measures still indicated strong relationships. Thus, substantial empirical evidence supports the notions underlying the strong correlations in previous research: that relative control over critical informational resources is strongly related.with dependence, which is a source of power: and that relative control over critical material resources is strongly related with dependence, which is a source of power. Positions of formal hierarchical authority are characterized by several factors. The first is the right to direct behavior, or perceived authority, inherent in the role: this was argued at length.by weber, (1968), and to a lesser extent by Hambrick (1981), Astley and Sachdeva (1984) and Hinings, et a1. (1974). This results the hypothesis that the hierarchical superiority of power holders will be strongly related to perceptions of their authority, or the right to direct behavior. As hierarchical superiority is generally held to be the major source of this belief, and as the link is so strong that the former has repeatedly been used as a measure of the latter (Blau & 168 Alba, 1982: Fombrun, 1983: Hambrick, 1981: Hinings, Hickson, Pennings & Schneck, 1974) , this relationship is hypothesized to be extremely strong: although the two variables are conceptually distinct, it is hypothesized that their relationship will be sufficiently strong to indicate that they tap the same underlying construct of authority. The second hypothesis engendered by these relationnships holds that both the formnal and perceived authority of the peer holder will be strongly related to his/her peer, independent of the amount of dependennce on the peer holder. This hypothesis is based on the theoretical postulates of Astley and Sactdeva (1984) , Hambrick (1981) and Bacharach and Iawler (1980) , who posit strong independent causal effects of both dependence and authority on power. They are also supported by the empirical findings of Hambrick (1981) , Blau and Alba (1982) , and Fonrbrun (1983). Blau and Alba (1982) found indices of work-related communication flow cenntrality and formal position to each independently predict power. In addition, membership in decision-making groups had an independent effect on peer. Fombrun (1983) found both formal position and centrality in communication networks to independently predict peer. Within a subsamnple of only technical workers, he found expertise and hierarchical level to independently predict peer, as well. Hambrick (1981) found both resource controls and authority to have strong independennt relationships with peer, by examining partial correlationns among peer and both hierarchical level and two indices of controlling critical information. 169 A different characteristic of positions of formal hierarchical superiority, distinct from the perception of the right to direct behavior that they egeder, is the function of the position. These positions generally involve the responsibility, and thus the ability, todecidehowandwlentoadministerrewerdsandpunishmentsto subordinates. They often include the responsibility to direct the behavior of subordinates, providing them with necessary task direction and access to both material and informational resources to solve problems in tte work process. Thus, those incumbents of hierarchically dominant positions have a significant amount of control over both critical informational and material resources. This association between authority and control over critical resources implies tnat, all other things being equal, the relative control over critical resources will be positively related to formal authority. However, the greater amount of conntrol over resources held by an incumbent of an authority position does not have an impact on the control over resources reld by the subordinate's colleagues or subordinates. The size of the effect of the superior's control over resources on tie relative control of resources will deped upon the amount of control over those resources held by others, and the number and availability of otters who have such control. Inmost jobs, thereareagreatmanysourcesofuncertainty tnata hierarchical superior does not lave information to cape with (cf. Crozier, 1964: Hambrick, 1981): these vary from having the skills needed to fix broken machinery, to details of efficient task performance, to caping with unpatterned variability in work inputs. 170 Therefore, the link between formal authority and tne relative conntrol over critical innformation ought to be significant but weak. Trere are also a number of physical resources tnat are controlled by others beside the hierarchical superior: these include factors such as tne physical environment of work and the processing and transfer of materials that are used as inputs into the individual's work process. Heever, the acquisitionn and transfer of many material inputs are formally, rather tl'an informally decided upon and determined. This formal decision making and acquisition process is most often a part of the responsibility of the position of an hierarchical superior. The result of this is that the superior will control a significant proportion of material resources. In addition, most superiors are responsible for recommending promotions or financial rewards for subordinates. This indicates tnat a greater prOportion of materials tl'at are critical for either job performance or job mainteance are conntrolled by the supervisor's decisionns. Therefore, the linnk between formal authority and the relative control over critical material resources is hypothesized to be strong. Hierarchical superiors are also responsible for tie firing of tneir subordinates: thus, employees are largely dependent upon tneir superiors for job mainteance. This results in tne hypothesis of a direct relationship between authority and depedence , which does not require the control of material or informational resources to impact on dependece. Tie otter aspect of dependence is not so strongly connected to authority: effective job performance is only partly determined by tl'e hierarchical superior, outside of tie tne superior' s 171 control over critical resources tnat determine job performance. The relationnship between authority and depedece is hypotl'esized to be strong. The evidece supporting the relationships among trese variables and geder and tenure were presented above. Based on tnat evidence, it is hypothesized that there is a strong relationship between being male and hierarchical authority. Due to the internal selection and promotion process, it is also hypothesized tnat there is a strong relationship between tenure and hierarchical authority . As was theorized above, employees who have worked in an organization for many years tend to deve10p more communication links: Mechanic (1962) referred to this effect of tenure as increasing ”access to organizational actors.” Greater tenure allows one tne opportunity to form more communication links, and to become central in communication networks. However, tenure does not require this increase in communication centrality: individuals will vary in the extent to which they deve10p such links. Therefore, it is hypothesized that tnere is a weak relationship between tenure and the relative control over critical information. Indirect Links Among Antecedent Constructs Several of the constructs in the nomological net have not been theoretically or empirically related to one anotner previously. However, the network of interrelationships hypothesized would logically result in some degree of empirical relationships among all variables. In order to most fully compare the theoretical links among constructs with empirical linnks among variables, all linnkages should be enamined. 172 The size and existence of these hypothesized relationships have been arrived at through the use of several algorithmns, based on tne theoretical ly and empirical 1y grounded relationships described in the primary links. All of these hypotheses concern constrncts trat are indirectly, rather than directly, related with one anotner. Tl'ese hypotnesized relationships were based on the assumption of shared variance between those directly unrelated variables tl'at share a logical link through a third construct. Constructs tlat are related by a logical causal flow are held to have some measure of relationship. In specifying the size of these relationnships, a few heuristics were used: two directly unrelated constructs that each had strong relationnships with tie same intervening variable, and had a one-way logical causal flow binding then, were held to be strongly related with one another: two directly unrelated constructs that tad one strong and one weak relationship with the same intervening variable, and had a one- way logical causal flow binding then, were leld to have a weak relationship with one another. Any constructs that are not directly related were hypotl'esized to be unnrelated outside of the effect of intervening constructs: partialling out the effect of the intervening constructs would result in no relationship between any two so connected. Based on these rules, tie fol lowing relationships are hypothesized. The indirect sources of power are tne relative control over critical innformational and material resources: these provide a basis for depedence, which is a major source of peer. Therefore, there is logical rationale to assume that the indirect sources (resource controls) and power will be related. It is hypothesized tnat peer is 173 strongly related to its indirect sources, control over critical materials and innformation. Thegederandtenureofpeerholdersarebothviewedasbeing strongly linked to authority. Autnority is a major source of peer. Peopleinautnoritytedtohavepeer, andtneyalsotedtohave greater tenure and be male. Thus, it is hypotlesized that peer is strongly related with geder and tenure. Peeple in autnority ted to have greater tenure and be male: such incumbents of positions of authority also ted to have others be depedent upon them, because trey told autl'ority. It is therefore hypothesized that dependence is strongly related to both tne geder and tenure of tie peer holder. The relationship with tenure may be slightly larger, because those with longer tenure tend to have greater control over critical information: this also egenders others' dependece upon then. PeOple in autnority ted to have greater tenure and be male: such incumbents of authority positions also tend to rave control over critical material resources . It is therefore hypothesized tl'at gender and tenure are strongly related to control over critical material resources. In addition, those with autnority are somewhat more likely to have conntrol over critical informational resources. A weak direct relationnship was hypothesized earlier between tenure and information control: duetothereasoningpresentedhere, aweakindirect relationship is hypothesized between geder and information control . There is no logical basis for shared variance or causal linnks between geder and tenure: thus, it is hypothesized tnat there is no relationnship between them. Likewise, there is no logical basis for 174 shared variance or causal linnks between control over critical materials and control over critical information: it is also hypothesized that ttere is no relationship between tlese variables. Other Variables that are Related Directly With Power The nomological network of predictive relationnships described above includes two constructs, geder and tenure, which are hypotnesized to have varying relationships with causal constructs in the network. In order to validate the measures of peer, it would be beeficial to examine tre relationnships among peer and variables that have previously been srown to have differential relationships with peer itself. In this way, discrinninant as well as convergent validity may be used. In addition, both of the direct antecedents of power in tle nomological network, authority and depedece, are perceptions of relationnship characteristics. It would further tne validation process to examine relationnships among peer and individual differece characteristics. Several such variables are prOposed. First, two personality characteristics, tie need for dominance and the need for affiliation, have been extensively examined in previous research, and found to be differentially related to peer (Murray, 1938: MoClelland, Atkinson, Clark & Lowell, 1953: Atkinson, 1958: Stewart, 1982: Steers & Braunstein, 1976: Meday, 1979: Dalton & Mar, 1979). Individuals with a strong need for dominance are motivated by controlling others in tneir environment, while people ”with high affiliative needs are concerned with establishing and maintainning relationships with otters" (Stewart & Ctester, 1982, p. 192). 175 Steers and Braunstein (1976) found that individuals with high need fOr dominance preferred to be‘group leaders and to play a major role in determining group performance: they also tended to perform highly, be supervisors, and show strong leadership abilities. Mcwday (1979) found that those with high need for dominance were more likely to attempt to influence others at work. Therefore, it is is hypothesized that there will be a significant relationship between the power holder's need for dominance and his/her power. The need for affiliation, however, has not been related with power. Steers and Braunstein (1976) found that need for affiliation was negatively correlated (although not at a level of statistical significance) with.desiring to be a.group leader, to play a major role in determining group performance, and to control one's own workplace. It.was also negatively correlated, again at nonssignificant levels, with holding a supervisory position, and leadership abilities of control and persuasiveness. Several studies have supported the notion that those people with a strong need for affilation avoid behavior which offends or disturbs others (Stewart & Chester, 1982: Sorrentino & Sheppard, 1978). Individuals‘with a strong need for affiliation tend to be more anxious when being rated by others on how likeable they are (Byrne, 1961), to avoid making comments to group members that might create contention in a group (Exline, 1962), and to perform at higher levels only if high performance will please others or maintain important relationships (Stewart & Chester, 1982). 176 Tre definition of tne construct of peer in this study focuses on influence in general, rather tnan on coercion or the overcoming of resistance. Influecing otters may be acconplisned through positive, affiliative means, as well as through doninating means. Therefore, it may not by assumed ttat the need for affiliation will have a negative relationship with peer, as it is measured and defined here. However, based on previous research, the need for affiliation is not likely to have a strong relationship with peer. Therefore, it is hypotl'esized tnat tl'e relationship between an individual's need for affiliation and his/her peer will be weaker tran the relationship betweentheneedfordoninanceandpeer. me last individual characteristic is considered. The intention to turnover is an individual difference characteristic which is hypothesized to have a negative relationship with tie amount of peer mm by an individual. Although tnere have not been direct reports of such relationships , it is hypothesized that individuals with a great deal of peer will be less likely to desire to leave their position. Supportive evidece was provided by Bachman (1968) , who found a correlation of .42 between the announnt of overall influence mm by University groups and faculty unwillingness to leave tneir University. Summary This chapter described the constructs in tne nonological net around individual power in organizations: trese constructs were identified in tre literature review as having been consistently related with peer. Trese constructs include direct sources of peer, which are dependence 177 and autnority, and the indirect sources of peer, which are tne relative control over critical informational and material resources . Tne two new variables of geder and tenure were introduced, and research was presented trat linnked them with tie sources of peer. The hypotlesized relations among all of tiese constructs were specified, in terms of primary links which were most thoroughly grounded in theory andresearch, secondarydirect linkswhichweresomewhat less thoroughly grounded, and indirect links which were grounded in logical deduction. Three new individual difference characteristics were presented , and hypotheses were drawn regarding their direct relationships with peer. Many limitations of the existing measures of peer have been identified. A new measure will be presented ttat edeavors to avoid tnose limitations by tapping the construct of individual peer in organizations as closely as possible. The next chapter presents the methodology employed to measure the construct of peer, and to assess tne validity of this measure using tne relations described in the nomological net. The subject pepulation, measures of all constructs, pre-test results, and analytic strategy are all presented. Significant attention is paid to the measurement of tie constrncts, by careful grounding of all measures in tie definitions of the constructs and, where possible, using previously validated measures. IV.W This section presents tre metnodology used to Operationalize peer, and provide enpirical evidence of tie construct validity of this measure. It describes tine subject population, the measures tnat were used to tap the constructs in tre nomological net, pre-test results for these measures, and the design of the analyses to assess construct validity. Subjects The subject population used for data collection was tre non-famlty staff of a large, public Midwestern University. Pairs of employees were randomly sampled from tte population of enployees. T'nese included admninistrative, supervisory, clerical, technical, mainteance and all other support staff. (re trousand, three hundred pairs of employees were selected for participation. The sample was comprised of two sub-samples, one consisting of pairs of co-workers at tte same hierarchical level, and the second consisting of pairs of supervisors and their direct subordinates. For tne first sub-sample, eight hundred enployees were randomly selected from computerized listings of all non-faculty personnel. Each employee was randomly paired with anotter employee with the same job title, in the same department. If the department 178 179 was small enough tlat there was only one employee with trat job title, a person in anotl'er job at tie same hierarchical level was selected to be the "partner.“ The second sub-sample consisted of five hundred pairs of enployees at different hierarchical levels. Five hundred supervisory personnel were randomly selected, also from a computerized listing of all supervisors at all levels: this inclnded managers and executives. Each supervisor was randomly paired with anotner employee in the same department, one hierarchical level leer. Half of tiese pairs were randomly selected to explore the peer of the superior over the lower level employee, and tne other 250 pairs were targeted to examine the peer of the lower level employee over tre superior. Procedure The process of identifying subjects resulted in tie identification of 1300 pairs of employees in all positions, and at all hierarchical levels. Because the computerized personnel listings did not include information regarding work group assignments, it was not possible to identify employees who actually worked together. In order to ensure that each employee had a working relationship with his/ier partner, several procedures were used. For pairs of co-workers , three questions were included at tne beginning of tie questionnaire, asking respondents to describe how closely they knew or worked with tleir partner. All respondents wno indicated tnat trey were not at all close with tneir partner on all three items were excluded from data analysis. 180 To ensure that pairs of enployees at different hierarchical levels were actually supervisory/ subordinate pairs, a more time-consuming process was undertaken. First, small departments were examined by means of a listing of the numbers of employees in each job title in each department. If tnere was only one enployee in a supervisory capacity at a hierarchical level above the level of tie designated subordinate employee, and less than four enployees at tl'e hierarchical level of the subordinate, it was assumed tnat tne person selected as the supervisor was indeed in a supervisory relationship with the person selected as the subordinate, because there were no other supervisors above them in the department. In all other departments, there was no way to ensure tnat the particular person selected as the supervisor was indeed supervising the particular subordinate 8/ he was paired with. Because of the sensitive nature of these relationships, it was deemed important to minimize false pairings. In addition, the tOpic of peer and depedece in supervisory-subordinate relationships was considered to be more sensitive for most subordinates tnan for most supervisors: they had more potential repercussions for participation. Therefore, tle subordinates were contacted by telephone. The study was briefly describedtothem. Theywereaskedfortleirpermissiontosend questionnaires, and for the name of their supervisors. For all pairs not confined by tie process of exanmining tie organnizational crart, only tnose dyads in which tle subordinates were actnally contacted and agreed with these requests were sent questionnaires. Of tte 500 computer-generated subordinate/ supervisory pairs , 363 were confirmed by oe of trese two processes. 181 As an additional check on this relationship, a question was included at the beginning of each of the supervisory or subordinate qnestionnaires, asking if he partner was tl'e individnal's subordinate or supervisor. After pairs of enployees were identified for participation, one person in each pair was randomly chosen to be tte ”power holder." In half of the cross-level pairs, the supervisor was randomly selected to betnepeerrolder, andintheotlerhalftresubordinatewas identified as do peer holder. The peer holder was sent a questionnaire inquiring about his/ her peer over the person he/ste was paired with: the name of the "power recipient" was written in on the t0p of tnese questionnaires. The peer recipient was sent a parallel questionnaire inquiring about the peer nolder's peer over him/herself: the name of the peer folder was written in on tne top of these questionnaires . Thus , respondents were directed to describe tneir peer relationship with one specific co-worker. All questionnaires included measures of power, dependece, authority and control over critical informational and material resources. only the peer-holder questionnaires included measures of geder, tenure, need for dominance, need for affiliation and intent to turnover. Clo-worker questionnaires inclnded three questions regarding tte closeess of work: supervisory and subordinate questionnaires inclnded instead a question confirming tnat relationship. Thus, six formns of tte questionnaire were sent: peer rolder supervisor, peer recipient supervisor, peer holder subordinate, peer recipient subordinate, peer rolder co-worker, and peer recipient co-worker. A 182 cOpy of each of these questionnaires is presented in Appedices A, B, C, D, E and F, respectively. Questionnaires and cover letters were sent to each employee at work. Two weeks following tne dissemination of the surveys, it was discovered tnat sone of the subjects who received questionnaires were apprehensive about its purpose. In particular, subjects who were members of one particular union which had been involved in protracted, difficult negotiations with tie University management evideced sone suspicion regarding potential non-research uses of tie survey . Therefore, follow-up surveys were delayed for two months. A more detailed description of the study was included in the cover-letter sent with all follow-ups. Follow-up surveys for co—worker pairs were limited to tnose non-responding employees in pairs in which one partner had returned a completed survey from tne innitial mailing. In order to address potential apprehension regarding the survey, an additional step was taken in subordinate-supervisory pairs . All non- responding enployees in subordinate-supervisory pairs in which one partner rad returned a completed survey were contacted by telephone. The study was briefly described, and each person's permission was obtained before sending a follow-up survey. In addition, thirty-seven new supervisory-subordinate pairs, who had not previously been sent surveys, were contacted by telepl'one: both tte supervisor and tie subordinate in tress pairs were contacted. The study was described, tleir permission was obtained, and a survey was sent to trese additional 74 employees. 183 All together, 2400 enployees, in 1200 pairs, were sent surveys: 400 pairs were supervisory/ subordinate, and 800 were co-workers. Inconplete surveys, and those in which tne relationship checks indicated that tne employees were not actual co-workers or subordinate/ supervisors, were eliminated from data analysis: a survey was considered incomplete if less tnan 80% of any measure was not completed. Five hundred and ninety-nine individuals returned usable surveys, resulting in a response rate of 25.0%. 0f ttese 599 conpleted surveys, 331 were from power recipients, and 268 were from peer nolders. Twelve of the peer holders completed all items on tie questionnaire except the self-descriptive denographics of tenure and geder, resulting in 256 peer nolders wno responded to all items. In addition, 262 of the completed surveys were paired with one another: tnat is, there were 131 pairs in which both individuals responded to all variable measures. The response rate for members of tne union undergoing difficult contract negotiations was 23.0%, which is conparable to the overall response rate. W3 bunch of the existing research on peer displays problematic measurement of tne construct of peer, and of related variables. Therefore, effort was taken to directly and reliably tap tne constructs which form tie nomological net. Multiple measures of peer were used, and evidence of construct, convergent and discriminant validity was sought. 184 The preceding chapter described several variables tnat were proposed to have relationships of different strengths with peer and with one anotner. The existence and size of each of those relationships was specified. Each of these constructs has at least one measure. T'ne construct of greatest interest, peer, has several Operatioalizations. Asumnaryandconparisonofall measuresusedinthisstudy is presented in Table 3. This table describes tne number of items in each instrument, which individual responded to it, which individual it describes, and in which Appedix it can be found. Measuring Power A Topic-Specific Measure of Peer The research previously reviewed showed that there is a need for an index of individual peer that assesses all aspects of tie construct: innfluece of individuals over one another that is sensitive to variance across contexts and across relationships. Such a measure is presented in Appedix G. It is a thirty-four item Likert-type questionnaire, asking employees to describe tl'e amount of influence neld by anotler organizational member over different aspects of their work life. he first step in the development of this measure was tne geeration of items. Topic-specific items were generated from reviews of the literature. Tlese items inquired abont tie amounnt of influece neld by a co-worker over the employee: this could be a supervisor, subordinate or peer. Each item dealt with a different aspect of work life: nearly fifty items were created. They were designed to be specific enough to CWSTRUCT Autfority I Autfority Dependence Control of Crit Maters Control of Crit Info Tenure Geder NDominance 185 Table 3 Measurement of the Constructs MEASURE NITS Specific Specific Generic Geeric Symbol W1 Perceived Perceived Heirarch Perceived Perceived Perceived Perceived Perceived Perceived Manifest Needs or N Affiliation Manifest Intent to Turn over NeedsQr 34 34 3 3 WW mm 11 11 15 15 10 10 RESPQ‘IDANT peer recip peer folder power recip peer folder power recip power folder peer recip peer holder archival data peer recip peer folder peer recip peer folder peer recip peer folder peer folder peer folder power folder peer folder peer folder (NE DESCRIBED APPDIDIX power folder power holder power folder peer folder both both peer folder peer folder peer folder peer recip power recip peer holder peer folder peer folder peer folder peer folder peer folder peer folder peer folder peer folder G 186 assess different aspects of work, but geeral eongh to be applicable to a wide variety of jobs. T‘feterm'poer'wasnotusedinthequestionnaire, toavoid response bias caused by tfe negative connotation of tfe term (Jones, 1984: McClelland, 1969: Kipnis, 1976: Pfeffer, 1981). The “how much influece does this person fave” format was adopted because it was tfe most consistently used technique, and when conpared to otfer formats, dencnstrated the least response bias (Provan, Beyer & Kruytbosch, 1980) , the most agreement across raters (Hinings, Hickson, Pennings & Schneck, 1974) , and tie most consistent support for tfe linnks with the sources of peer tfat were identified. Tfe researcher then elimninated items tnat were vague, overlapping or overly specific. Two versions of these items were written, due to different terminologies used by the different types of occupations told by tfe pre-test pOpulation. The first of these was tfe staff and administrative personnel of academic institutions : these questions were phrased to ask about the influece held by the respondent's "supervisor.” The second of these was academic faculty: tfese items asked about the influece held by the respondent's “chairperson“, and replaced a question abont ”bonuses and overtime" with an item on ”summer support and other non-salary pay“. The resulting list of qnestions were Q-sorted by employees in a variety of jobs in an academic institution: faculty, staff and administrators were used. All seven respondents were naive to tfe subject and the research. Four categories emerged consistently from a majority of tfe Q-sorts: the social environment of work: tfe physical environment of work: work itself: and money and advancenent. 187 Category headings were created, and tfe Q-sort procedure was continued with another nine naive subjects, again from a variety of positions. The majority of the items consistently were sorted into thesefactors. Eightitemsdidnot, andtheyseemedtocovertwo topics: tfe timing and scheduling of work, and personal feelings and issues. These two categories were added to the first four, and tie Q-sort procedure was continued with yet anotfer set of naive subjects. This process resulted in a high degree of consistency in tfe categorization of items. All six categories appeared to be relevant to perceptions of the influece held by organizational superiors . In order to ensure that this questionnaire was not only applicable to one' s supervisor, another version of the questions was produced tfat innquired about tfe innfluence of any one co-worker (administrative or staff) or colleague (academic faculty). These items were also Q-sorted by a new set of naive employees, and the results were consistent with the preceding findings. All together, twenty Q-sorts were conducted with tfe six categories. The questionnaire that resulted consisted of 35 items, which formed six categories of work life tfat may be affected by another individual at work. Response categories used a five-point Likert-type scale, consistent with tfe perceptual measures in previous research (Hambrick, 1981: Provan, Beyer & Kruytbosch, 1980: Hinings, Hickson, Pennnings & Schneck, 1974: Pfeffer & Salancik, 1974): anchors were ”no influece,” ”a little influence,” "a moderate amount of influence," "a great deal of influence,” ”total influence”. It was designed. to be specific to tfe work environment, and be both relationship-specific and tapio— 188 specific, annd yet general enough to be used across a variety of job types. The measure was pre-tested with a sample of 42 respondents, to ensure reliability of the scales. Details of the pre-test procedure are presented in Appendix Q. Alpha coefficients for five out of the six scales were acceptable, ranging from .81 to .91; these scales were either unnchanged, or single itenns were elinminnated for parsimony. Thescaleofpoweroverthesocial environmenthadanalpha coefficient of .68. One item was removed, and two new items were added to this scale. The pre-testing procedure resulted in the 34 item questionnaire, tapping power over six different aspects of one's work life. It is presented in Appendix G. A Generic Measure of Power A second perceptual measure was developed to validate the topic- specific measure of power. It is simnilar to the items used in the existing literature in its wordinng, and like prior measures does not distinguish among different tOpics or aspects of work. It used multiple itene to allow assessment of internal consistency; however, it describedpoweroveranindividual intheabstract, ratherthanpower over specific topics. This measure of power is presented in Appendix B. This index consisted of three items, asking employees to describe theamountof influence, control orpoerthepowerholderhasoverthe power recipient at work. It used the sane response format as the topic specific measure, which is consistent with those used in previous measures of power (Hambrick, 1981; Provan, Beyer & Kruytbosch, 1980: 189 Hinings, Hickson, Pennings & Schneck, 1974; Pfeffer & Salancik, 1974). These generic power items were interspersed among the topic-specific ones; in order to avoid contaminating other responses with the potential negative bias produced by use of the word ”power,” this item was last in the section of the questionnaire. A Non-evaluative Symbol of Power Both measures of power described above asked for individual evaluations of the amounnt of influence held by an individual over another. Both are therefore Open to response bias caused by social desirability, halo error, leniency, severity and central tendency errors. A non-evaluative index would be helpful for the validation of the measure. There are no enacted measures of individual power that would logically be valid across different jobs. The symbols of power that have been recommended, such as office placement and size (Pfeffer, 1981) are not relationship-specific. An alternative symbol of relationship-specific power, based on logical grounds, was therefore created. In most relationships between enoloyees, individuals with less power are likely to contact individuals with more power. Workers wro are dependent on another for information they need to do their work are more likely to go to that person to get the information they need; workers who are depedent on another for physical resources they need togettheirworkdonearelikelytogotothatpersontogetaccessto the resources. 190 The noneevaluative symbol of power was the frequency of contacting others; those who are less powerful were presumed to initiate more contacts, while those who are more powerful were presumed to initiate fewer. This variable was assessed with three questionnaire items: these are presented in Appendix I. All three involved double questions. The first inquired which party calls, goes to, or contacts the other more often, followed by a second question that inquired how many times more often. This relied on the respondents to make the relative judgement themselves, but used two items for clarity and face validity. Measuring;0ther‘Variables in the NOmological Network Authori Authority was defined as the belief that a person has the right to direct one's behavior. It was presented as stemming primarily from formal hierarchical position in work organizations; it has frequently been operationalized by this variable (Hambrick, 1981: Hinings, Hickson, Pennings & Schneck, 1974; FOmbrun, 1983). Authority is viewed as one source of power: if others believe that they ought to comply with requests, they are likely tordo so. This study used two measures of authority. Formal hierarchical authority was tapped by the number of levels of hierarchical superiority or inferiority of the power holder over the power recipient in eadh pair. This information was gathered prior to questionnaire administration: it.was used to select pairs of respondents. In addition, one question.was included in the questionnaire as a check on the superior-subordinate relationship. Only respondents who answered 191 in the affirmative were included in data collection. This variable was scored from the initial sets of partners identified for participation in the study. This technique measures relative hierarchical superiority, rather than the absolute hierarchical level. It limits the variable to different levels within a chain of comnand. Because of this, an individual with one level of hierarchical superiority is the direct supervisor of the other individual in the dyad, and therefore has the formal right to direct behavior within certain role constraints . For trose respondent pairs in which the power holder is the co-worker, 8/ he has no formal right to direct the behavior of the other. For ttose pairs in which the power holder is the subordinate, s/he is obliged to follow the work-related directives of the power recipient. Unlike absolute measures of hierarchical authority, this Operational ization takes into account the level of the power recipient. In addition, it more closely taps the construct of the perceived right to direct behavior than do absolute measures; varianoe in that right is closely tied to position in a direct chain of comnand. A second, perceptual measure of authority was created for this study. Previous research on the power of individuals has operationalized the construct as formnal position, ignoring the fact that the construct is, by nature, perceptual. The new measure is presented in Appedix J. It consists of 5 items, inquiring about the other person's right to directbehavior. Threeoftheitems, numbers 1, ZandS, tappedthis directly by asking about this right; the first two of these were adapted from items wordings used to identify what proportion of power 192 had a "base" of auttority (Bachman, Smith & Slesinger, 1966: Student, 1968; Thamhain 8 Gemmill, 1974). The other two questions, items 3 and 4, attempted to assess autlority indirectly by inquiring about the subject's habitual response to the recommendations and decisions of the other. This was based on the notion that people generally respond to the direction and suggestionsofttosewtotheyseeashavingtherighttodirecttheir behavior. All five itenns used a f ive-point Likert-type response format with anchors of ”strongly agree," "agree," "neither agree nor disagree,“ "disagree," and "strongly disagree.” The index was pre-tested for reliability: details of this procedure and results are presented in Appedix Q. As the alpha for the five-item scale was .78, it was used without change. Dependence The other direct theoretical antecedent of power is depedence. Only one of the previous studies directly measured dependence: this was the depedence of one organization upon another for finnanncial support. As dependence at the individual level has not been assessed, a new measureoftheconstructwasdeveloped. Dependencewasdefinedasthe extent to which someone is needed to perform effectively at work. This was specified to include two factors of effective functioning: job performance and job maintenance. The measure that assessed this construct contained questions directed at both trose subsets of the variable. while all other variables referred to the characteristics of the 193 power holder, this variable referred to the dependence of the power recipient.gn the power holder. The focal subject was reversed: this is evident in the questionnaire, presented in Appendix K. Items 1 through 7 focused on dependence for effective work performance. The literature described dependence in terms of needing information to deal with.critical contingencies, or needing material resources in order to accomplish.work. Both those foci of dependence were included in the measure,the former with items 3, 4, S and 7, and the latter with items 1 and 6. The measure also included a general item, number 2, that focused on relying on the person in order to accomplish work. Items 8 through 11 focused on dependence for job maintenance. They all addressed the potential effect.Of the other person for job maintenance and the maintenance of a career in the organization. The dependence measure was also pre-tested, resulting in an alpha of .88. Relative Control over Critical Material Resources This variable is conceptualized as the amount of control held by an individual over material resources needed by another to perform his/her work, relative to the amount of control held by all other individuals over those resources. In previous research, this construct has been Operationalized by use of related variables, sudh as formal authority (Pembrun, 1983): the number of interorganizational linkages among agencies, which allow them to "mobilize power in an effort to obtain scarce resources" (Provan, Beyer & Kruytbosch, 1980, p. 205); and the departmental contribution to university resources such as number of 194 graduate and undergraduate students , amount of grants , and business and professional contacts (Salancik & Pfeffer, 1974) . Direct measures of this variable at the individual level have not been reported. Tre specific resources that have been subject to scrutiny are different for every study, making generalizations difficult. In addition, no framework exists to organize the types of physical materials that exist, and may be critical in the performance of a given job. This study offers such a framework. Reviews of prior research found that control over resources can be maintained in two ways: through acquisition of resources, or through determining their flow. The measure of this construct, presented in Appedix L, included both of these factors. It was administered to both tne power rolders and their recipients. This instrument contained five subsections , each focused on a different category of material resources required to get work accomplisred: materials and supplies: equipment, tools and machinery: services and kelp from others: physical environment; and monetary resources. These categories were identified by Peters, O'Connor and Rudolph (1980) , in a stn.dy of the factors that prevented employees from getting their work done. Trey used tie critical incidents technique, and Q-sorted tne resulting factors into eight dimensions. These dimensions, then, describe tlose factors that are critical for the performance of work. The eight factors identified in the study were: job-related information: tools and equipment: materials and supplies: budgetary support: required services and help from others: task preparation: time availabil ity: and work environment. Tre informational , task- 19S preparation and time dinmensions were deleted, as this measure is focused on control of material resources. Tie control over critical materials measure had fifteen specific items, three dealing with control over each of tie five types of material resources. Each item requested a relative assessment of the control heldbythepowertolderineachpair, asopposedtothe control neld by other people. Each item presented a slightly different way of tapping control: the ability to acquire and tre ability to control the flow of materials were both included. This measure was also included in the pre-test. Alplas for the five three-item scales ranged from .68 to .95, indicating acceptable reliability: the entire fifteen item scale had an alpta of .87. Thus, tne relative control over five different material resources may be reliably measured. To tre extent ttat these scales intercorrelate highly, there is evidence for a single underlying characteristic of control over important material resources. Relative Control over Critical Informational Resources Tte relative control over critical information was defined as being parallel to the relative control over material resources: it is tle control of tie power nolder, relative to all otners, over information or knowdedge tnat is vital in tfe effective functioning of the power recipient. This is information tl'at allows one to cope with critical contingecies, or unnforeseen problems trat arise and amp tne transformation process. Like tte otter control variable, tlere are two ways of creating control: through acquisition and through determination of tie flow of tre resource. 196 Hinings, Hickson, Pennings and Schneck (1974) specified three different ways of using information to cope with critical contingecies: by acquiring information about unknown areas, by preventing problem caused by unplanned variation, and by coping with problem as trey arose. While conceptually inclusive, this typology is quite geeral: for example, there is an entire lost of ways to cope with problem tnat nave already arisen. Hambrick (1981) was somewtat more detailed in approach: he described scanning behaviors as ways of acquiring information to predict or deal with critical contingecies . This technique certainly fits the first category of coping listed by tne earlier researchers, and probably involves tle latter two. The measure of relative control over critical information used in this study was based on these two works. It consisted of ten items trat asked both power tolders and power recipients to indicate the amount of control neld by tne power folder, relative to tre control held by others, over the flow of critical information. It included item specifically focused on predicting problem, communicating information to reduce their effect, and dealing with those trat have arisen. This measure is presented in Appedix M. Item 1, 3, 5, 6, 7 and 9 focused on tne acquisition of critical information: tnese innquired about wnat infornmation, including expertise, thepersonhasorhasaccessto. Theotheritem, 2, 4, 8 anle, weredirectedattl'econtroloverits flow: theseaskedabout tre person's willingness or frequecy of sharing the innfornnation with tie subject. The pre-test of this measure found an alpha coefficient of .90, indicating acceptable reliability. 197 Tenure and Gender Two other variables in the nomological net are tl'e denographic characteristics of tie organizational tenure and gender of tie power nolder. The single items tnat measure trese cl'aracteristics are presented in Appedix N. As trey are denographic variables, tney were directly assessed. Need for Dominance and Need for Affiliation The need for dominance and the need for affiliation have been examined fairly extensively in previous research (Murray, 1938: McClelland, Atkinson, Clark & Lowell, 1953: Atkinson, 1958: Stewart, 1982: Steers & Braunstein, 1976: Mowday, 1979: Dalton & Todor, 1979). The Manifest Need Questionnaire (Steers s. Braunstein, 1976) was deve10ped and validated to measure these two needs, among others. These scales each consist of five item, all of which innquire about betaviors at work which would reflect these needs: one of tie item in the need for Dominance scale, and two of the need for Affiliation item, are reverse scored. These scales are presented in Appedix 0. Previous research has supported the reliability and validity of trese measures. Steers and Braunnstein (1976) presented three validation studies, which validated the measures against self-ratings of trese needs using tl'e Personality Research Form, expert ratings of tre subjects' needs, laboratory experiments involving work group preferences , and reed-related work attitudes. This study found internal consistecy reliability indices (alpha coefficients) of .56 for nAffil and .83 for nDom, and test-retest reliability indices of .75 198 for nAffil and .86 for nDom. Mowday (1979) used the need for Dominance index in his analysis of the use of power in organizations, but did not report reliability indices. For use in this study, the response scale was slightly altered. Tre measures were deve10ped with a seven-point Likert-type scale: in this study, the scale was reduced to a five-point Likert-type scale. Response categories were: always, usually, sometimes, seldom and - never. Reliability analysis found alpha coefficients of .78 for the need for dominance scale, with the deletion of one item. Tne need for affiliation scale showed far less reliability, with an alpha of .02. Two items were deleted from scale formation, justified by an increase in the alpha to .42. Intent to Turnover A measure of the intention to turnover has also also been previously created and examined (Seashore, Iawler, Mirvis & Camnmann, 1982) . This index consists of three item, two using an agree-disagree Likert-type response format, and tie third using a likely-unlikely Likert-type response format. The measure was found to have a coefficient alpha of .83, and to have substantial negative correlations with job satisfaction and involvement. This measure was used with five-point Likert type response scales, and is presented in Appendix P. Reliability analysis supported the deletion of one item, resulting in a two-item scale with an alpl'a of .75. 199 Analyses This section describes the analyses used in the construct validation of the measure of individual power in organizations. This involved two stages: first, the measurement model was examined, to ensure adequate empirical representation of constructs in the nomological network: second, the theoretical model was examined, to assess the relationships among measures of the constructs. Both.of these stages consisted of several steps, the order of which was determined by their priority in the validation process. Examination of the Measurement Model The first step in the validation process was the establishment of the viability of the measures used to assess constructs in the nomological network around power. This entailed four types of assessments: the structure underlying the measures: the independence of measures which purported to assess different constructs: the congruity of measures which purported to assess different constructs: and the reliability of measures. In order to assess the structure of the measures, factor analytic procedures were used. Because the tOpic-specific power measure was deyeIOped through a repeated.Q-sort.procedure, which produced consistent identification.of the six categories, this measure was examined with a two-stage factor analysis: a confirmatory factor analysis was followed by an exploratory factor analysis. Other measures which were believed to tap multiple constructs included the measures of the control over critical materials, control over critical 200 informational resources and dependece: trese were examined with single stage factor analysis. Toprovideevidencetlattnemeasureshadfidelitywiththe constructs they assessed, the independence of measures of closely related constructs was examined. Factor analysis was used for this purpose. T'neindependenceoftnemeasuresofpower,depedeceand autlority was analyzed, as was tte independence of tre measures of control over informational and material resources. Three measures of power were used: He congruity among tie mnultiple measures of each construct was assessed with correlational analysis. Last, tte internal consistecy reliability of all multiple-item measures was assessed by means of inter-item and item-total correlations, and Cronbach's alpha coefficient: only tl'e latter indices are reported. This process was used for all measures except the indices of tenure, gender and hierarchical autrority , which consisted of single item. Examination of tie Theoretical Model The second stage of analysis involved examining tte enpirical relations among the Operationalizations of power and other variables in the neurological network. The hypotl'esized relationships among tl'e constructs in tie network were presented in Table 2: the relationships among tte measures of the constructs slould reflect ttese relationships. Some modifications of the enpirical analyses were made, based upon the results of tre analysis of tie measurenent model. As most of tie variables were measured by perceptions, the estimates may be biased by social desirability or other response 201 errors. In order to minimize the effects of these types of errors, the perceptions of both members of the power relationshipnwere examined. The different raters were treated as different methods of assessment, and aggregated to produce estimates of constructs. The empirical examination of relationships among measures was ordered by the primacy of the linkages among the constructs in the nomological network. This ordering was described in the preceding chapter. The primary level relationships were those among power and its immediate antecedents, authority and dependence. Secondary level relationships were those between power and the individual difference characteristics of the need for affiliation, need for dominance, and intent to turnover: these were hypothesized to have differential direct relationships with power. Tertiary level relationships were those among the immediate sources of power and their direct antecedents. Last, the indirect relationships among the control over critical resources and power were examined. Oorrelational analyses were used for the first three analytic steps, and a two-stage, two-step regression analysis was used for the last one. V. RESULTS This chapter describes the results of the analyses conducted to empirically validate the measures of individual power in organizations . First, the measurenent model is examined, to identify the extent to which the newly created questionnaires provide independent, reliable measures of the constructs they were designed to tap. Second, some modifications are made in the relationships to be tested, based on these results. Third, the relationships among the variables are examined, and their fidelity with the theoretical relationships are described. Part 1: Establishing the Measurement Model In this section, the structure, distinctiveness and reliability of measures are described. There were slight fluctuations in the number of respondents included in the factor analyses, as each factor analysis excluded respondents who failed to complete a single item of the measures being analyzed. The number of respondents was also reduced when examining the "symbol of power” measure, as it was occasionally not completed. All analyses of the measurement model included between 544 and 594 respondents. 202 203 Measures of Power Three measures of power were created for this study. First, tie structure of the topic-specific measure of power (see Appedix G) is examined. This instrument, as redesigned following tle pre-test, consisted of 34 item which tap power over six different aspects of work: the timing and scheduling of work: the work itself: money and advancement: the physical environment of work: tne social environment of work: and personal issues at work. The factor structure of the scales was examined using confirmatory factor analysis. A second-order exploratory factor analysis was condncted on the factors created in the first first step. Finally, internal consistency reliability was examined. The Multiple Group's program of Hunter's (1987) PACKAGE system was used to perform confirmatory factor analysis and produce inter-factor correlations. Alpl'a coefficients of these six factors ranged from .88 to .94. The factor loadings produced by the confirmatory factor analysis are presented in Table 4. Item are listed in order of the centrality of their content to tie construct trey were inteded to tap. Oommnalities were used in tne diagonal of tie inter-item correlation matrix. The results snow ttat most item loaded quite strongly on tne factors to which tley were assigned. However, most item rad factor loadings tnat were nearly as high on at least one other factor, and were quite high (over .60) on most of the factors. This indicates tlat tne respondents teded to respond similarly to item in all six areas of influece. Supporting this are the high correlations among tle factors, shownn in Table 5. 204 Table 4 Factor Ioadings from Confirmatory Factor Analysis of the Topic-Specific Power Measure Item 5 Work M Personal Social Physical Money/ Ennvt advancemnt 1 .84 .81 .77 .68 775' T76"""" 13 2‘9 .87 .83 .74 .88 .89 5 ‘;_3 .79 .77 .68 .72 .69 16 . 0 .90 .81 .76 .83 .81 34 :22 .80 .83 .74 .74 .68 22 ;_g .79 .83 .71 .75 .73 14 .86 3&2 .75 .77 .75 .66 8 .73 ;_g .71 .75 .72 .68 15 .89 ;§_ .77 .75 .76 .73 29 .78 ;_g .78 .74 .81 .79 7 .81 . 3 .76 .74 .79 .84 30 .85 788 .85 .76 .71 .70 31 .80 E .78 .72 .65 .61 20 .66 .63 :11 .68 .57 .57 33 .82 .82 .85 .76 .72 .74 35 .78 .72 Egg .68 .71 .74 19 .80 .72 . 1 .68 .74 .71 2 .63 .67 :52 .59 .65 .66 26 .69 .75 .72 3§Z .68 .63 27 .41 .43 .45 :2; .37 .29 21 .68 .73 .72 :éi .38 .57 25 .30 .32 .34 :59 .27 .21 32 .70 .74 .70 ;§§_ .67 .58 17 .71 .70 .76 192 .64 .61 9 .80 .84 .73 :1; .80 .78 11 .74 .71 .75 .68 3§§ .76 12 .83 .81 .75 .73 .85 .77 18 .74 .70 .72 .64 T84 .81 10 .80 .79 .75 .71 E .78 4 .58 .54 .62 .43 .66 .78 3 .83 .82 .80 .68 .82 79$ 6 .56 .57 .56 .53 .65 :72 23 .84 .80 .81 .68 .82 ;_4 28 .80 .76 .81 .64 .76 .83 N3565 205 Table 5 Inter-Factor Correlations of Topic-Specific Power Measure work Timing Personal Social Phys Envt Money/ advanct work 1.00 EB .98 1.00 Personal .95 .92 1.00 SocIST .85 .89 .87 1.00 Phys Envt .91 .88 .87 .81 1.00 Money .89 .85 .88 .73 .91 1.00 All correlations are corrected for attenuation.due to unreliability of measurement N a 565 Table 6 SecondsOrder Factor Loadings for TOpic-Specific Power Measure Factor General Factor Eigenvalue Pot. of var. work .987 5.401 90.0 Timing .970 .301 5.0 Personal .962 .134 2.2 Social .874 .100 1.7 Phys Envt .932 .058 1.0 Money .902 .007 .1 N=565 206 A second-order exploratory factor analysis was conducted, to identify wwhether tfe original six factors clustered together. The S;PS_S_ Factor program was used, with principal axis extraction. This procedure identified one overall factor: tie second order factor loadings are presented in Table 6. This indicates trat individuals teded to respond to all item similarly. Despite the fact that repeated Q-sorts found tnat individuals distinguish among six different aspects of influece at work, these differences did not appear in peOple's responses to tne questionnaire measure of tie amounnt of power treyteld, oraco-workerheldoverthem. Itmaybettattnosewto hold power over one aspect of work life also hold poer over other aspects: alternatively, this instrumnt may be insufficiently sensitive to differeces, allowing response patterns to overwtelm existing differeces. As tte six factors were not empirically distinguisted, tre item were collapsed into a single scale, consisting of responses to all 34 item. Inter-item correlations and item-total correlations were examined, and strog inter-correlations were found. Cronbach's alpha, anotner index of internal consistecy reliability, was quite high at .98. Twootler indicesofpowerwereused: anissue-geeric perceptual measure, and a symbol of poor measure. The reliability and convergent validity of trese measures was examined next. The generic poer measure consisted of three item inquiring about geeral amount of influece, control and poor teld over anotter at work. The alpta for this three-item scale was .93. A symbol of poer measure was created, consisting of three 207 questionns regarding the frequency of going to see, contacting or calling the other person; it was hypothesized that the more power one holds over another, the less likely one is to take action to contact the other. This three-item measure had an alpha coefficient of .69. The correlationns among all six sub-scales of the tOpic-specific measure, the entire 34 item topic-specific questionnaire, the generic power measure and the frequency-of-contact symbol of power are presented in Table 7. Only 544 respondents were included in these results, because the symbol of power section of the questionnaire was occasionally not completed. Strong correlations among the topic Specific and topic generic power measures provided evidence of convergent validity (r ranging from .80 to .93); however, the frequency of contact index did not appear be a valid measure of individual power (r fronm .01 to .04, none significantly different from 0.0) . This measure was therefore not inncluded in other analyses. Measures of Other Variables Other variables measured in this study were dependennce, authority, control over critical information, control over critical materials, gender, tenure, need for dominance, need for affiliation and intent to turnover. The independence, internal structure and internal consistency reliability of the newly developed measures were examined. Dependence Dependence was measured with two questionnaires. (he was directed at dependence on another for maintaining one's job, and the other was directed at dependence on another for solving problem that stop the accomplishment of one's work. The first consisted of four item, the Scale work Persl Mony Tim Socl Phys Total Generic Symbol* 208 Table 7 Correlations Among Power Measures work Persl Mony Tim Socl Phys Total Gener Symbol 1.00 .87 1.00 .84 .80 1.00 .92 .84 .81 1.00 .81 .79 .70 .85 1.00 .85 .79 .84 .83 .77 1.00 .96 .92 .90 .96 .89 .91 1.00 .88 .84 .90 .87 .80 .85 .93 1.00 .03* .04* .03* .01* .01* .05* .03* .04* 1.00 * = Correlations with Symbol are not significant at p < .05 All other correlations are significant at p < .001 N = 544 work Persl Mony Tim Socl Phys measure Tetal Symbol work itself sub-scale of topic-specific power measure personal issues sub-scale of topic-specific power measure money and advancement sub-scale of t0pic-specific power measure timing and scheduling sub-scale of topic-specific power measure a social issues sub-scale of tapic-specific power measure physical environment sub-scale of topic-specific power summed tapic-specific power measure topic-generic power measure relative frequency of contact power symbol 209 latter of seven items. In order toldetermine whether these two areas of dependence exist as separate constructs, the eleven items regarding dependence were factor analyzed. The principle axis factor identification process of SPSS was used, and the factors were rotated orthogonally. This process resulted in two distinct factors; the rotated factor loadings are presented in Table 8. The first factor contained items focused on dependence for work performance, and the second contained items on job maintenance dependence. The rotated factors had an estimated covariance of .06. Both dependence measures were found to have a high degree of internal consistency. The four-item job-maintenance dependence scale had an alpha coefficient of .91, while the seven-item job performance dependence scale had an alpha of .93. Authority Authority over another was measured two ways. First, a five-item scale was used to assess the perception of authority to direct behavior. The second measure of authority was formal hierarchical superiority. This was controlled by the selection of subjects to have one of three values: one level of hierarchical superiority for the power holder (e.g., direct supervisory position): no hierarchical difference between the power holder and power recipient (e.g., co- workers); or one level of hierarchical inferiority for the power holder (e.g., direct subordinate position). These were relative measures, and were coded as +1, 0 and -1, respectively. The independence of the measures of authority, dependence and power was explored by means of a two-stage factor analysis. First, a 210 Table 8 Rotated Factor Loadings for Dependence Measures Item # Factor 1 Factor 2 2 .81 .18 n 75 .17 3 I77 .28 4 22 .27 6 .75 .29 7 :7: .29 s :5; .47 11 .24 .87 9 .20 785 a .29 75 10 .32 _._"7_§ Eigenvalues 6.07 1.40 Items are presented in order of factor loading. N = 594 211 confirmatory factor analysis was conducted on all power, depedence and authority item; this resulted in a correlation matrix of factor interrelationships. This matrix is presented in Table 9. An exploratory factor analysis was then conducted, to confirm the independence of the power , depedence and authority measures . Principal axis extraction and varimax rotation were used for this analysis; the rotated factor loadings are presented in Table 10. This analysis indicates that only two factors underlay the responses to these measures. The first factor contained power measures and hierarchical authority; the second factor contained dependence measures and perceived authority. Thus, the empirical responses to the measure of hierarchical authority were redundant with power , and responses to measures of perceived authority were redundant with dependence. This finding makes the examination of empirical relations among power, depedence and authority impossible with this data. Either individuals are not able to distinguish power from hierarchical authority , and depedence from perceived authority , or the measures used were insufficiently sensitive to existing differences. This required the revision of relationship testing; this will be connsidered in a subsequent section. Relative Control over Critical Resources. Relative control over critical resources were measured by two questionnaires. Control over material resources was tapped by three item that were repeated for five different types of material resources. Control over infornmational resources was tapped by ten item tl'at inquired about control over access to information about how 212 Table 9 Correlations Among Power, Dependence and Authority Factors Scale SPower GenPower Pcd Au Hier Au wrk Dep Job Dep SPower 1.00 GenPower .92 1.00 Pod Au .68 .69 1.00 Hier Au .55 .62 .35 1.00 er Dep .50 .47 .71 .02 1.00 Job Dep .67 .68 .59 .42 .57 1.00 All correlations are significant at p < .001 N = 565 SPower = Issue-specific power measure GenPower = Issue-generic power measure Pcd Au 3 Perceived authority Hier Au = Hierarchical authority wrk Dep = Dependence for work performance Job Dep = Dependence for job maintenance 213 Table 10 Rotated Second-Order Factor Ioadings for Power , Dependence and Authority Factors Factor Factor 1 Factor 2 Gen Power .87 .43 SPower 79' .47 Hier Au E .00 er Dep .05 .98 Pod Au .47 :7: Job Dep .54 ._56 Eigenvalues 3 . 72 . 87 Item are presented in order of factor loading. N = 565 Gen Power = Issue-generic power measure SPower = Issue-specific power measure Hier Au = Hierarchical authority Pcd Au 3 Perceived authority Job Dep = Dependence for job maintenannce er Dep =- Depedence for work performance 214 to solve problem that stop work, and information tnat allowed one to predict problem that stop work. First, the independence of these two types of resource control was examined, by means of factor analysis. All resource control item were factor analyzed, using principal axis extraction and varimax rotation. The resulting rotated factor loadings are presented in Table 11: they indicate that there are three factors that underlay responses to these item. Information control item loaded onto a single factor: the two different types of information were not found to be distinnct. An unexpected finding was that material control item loaded onto two separate factors: control over money, and control over all other types of material resources. Control over the physical environment item loaded fairly strongly onto both factors. Thus, empirical results indicated trat three types of resources were perceived to be differentially controlled: information; money: and equipment, supplies and services. Three scales of control over critical resources were created. The environment items were not inncluded in scale construction, due to high loadings on two factors. The alpha coefficients for both the nine-itenn materials control scale, and the three-item monetary control scale, were .97. The ten-item scale of information control had an alpha of .96. A summary table of the alpha coefficients and number of item in all multiple-item scales is presented in Table 12. 215 Table 11 Rotated Factor Loadings for Resource Control Items Item Factor 1 Factor 2 Factor 3 EqptCl .81 .29 .24 SuplCl In .26 .25 Sup1C2 . 1 .26 .27 EqptCZ .‘8'0' .29 .29 Seerl 7'76 .36 .26 Seer2 E .34 .28 Sup1C3 i_§. .33 .41 Seer3 ._§ .38 .42 EqptC3 fl .33 .46 EnvtCZ .64 .22 .56 EnvtC1 .62 .22 .52 InfoClO .18 .84 .07 InfoC9 .19 E .12 InfoC4 .24 . o .18 InfoC8 .20 780 .09 InfoC2 .26 T78 .24 InfoC7 .28 777 .22 InfoC6 .26 :5 .16 InfoC3 .33 .71 .26 InfoCl .34 .78 .30 InfoCS .37 17 .23 MonyC3 .31 .24 .87 MonyCZ .32 .23 E MonyCl .32 .22 ;_g EnnvtC3 .57 .24 .63 Eigenvalues 14.80 2.77 1.29 N = 562 EqptC = Relative control over critical equipment SuplC = Relative control over critical supplies Seer a Relative control over critical services EnvtC a Relative control over critical aspects of the environment InfoC a Relative control over critical information MonyC a Relative control over critical monies 216 Table 12 Cronbach's Alpha.Coefficients of all Multiple-item Measures Construct Measure # Items Cronbach's “.82 Power Topic-generic 3 .93 Power Topic-specific 34 .98 Dependence Job Maintenance 4 .91 Dependence work Performance 7 .93 Control over Supplies, equipment, 9 .97 critical and services materials Control over 3 .97 critical monies Control over 10 .96 critical information Intent to T0 2 .75 N Dominance 4 .78 N Affiliation 3 .42 217 Part 2: Revisions of the Model and Hypotheses Due to the examination of the measurement model, limited revisions were made in the relationships examined. Because of the empirical overlap between measures of authority and measures of power and dependence, relationships involving authority may not be examined with this data. This resulted in the simplification.of the model and hypothesis testing. The model depicting the nomological network.was revised to take into account the measurable variables. This revision is presented in Figure 3. As authority could not be distinctly measured, its relations with other variables were removed from the model. In addition, three discrete types of resource controls were identified, rather than two: the relationships among the third type of resource control and other constructs are specified in the revised model. Prior research on organizational power held that control over money and budget decisions was control over a type of material resource: it was posed as a cause of dependence, and therefore power. (Pfeffer & Salancik, 1974: Salancik & Pfeffer, 1974: Provan, Beyer & Kruytbosdh, 1980: Fombrun, 1983) Based on this research, relative control over critical monies is hypothesized tonhaye the same relations with dependence and power as the relative control over other critical materials. The specific hypotheses about relationships among all the revised antecedent constructs and power are presented in Table 13: this table includes power, dependence and its antecedent resource controls. The relationships between monetary resource control and all other 218 Control over Critical Materials Control over CriticalmDepedence Monies Control over Critical Information = = strong link Figure 3. Revised Linkages Among Antecedents in a Nonological Net of Peer 219 Table 13 Revised Hypothesized Relations Among Antecedent Constructs in a NOmological Net of Power Power Depend CCMater CCMony CCInfo Power - Depend S; - CCMater (S) 1§ - CCMony (S) ‘§ (8) - CCInfo (S) §' 0 0 - Depend 2 Dependence CCMater a Relative control over critical Material Resources CCInfo a Relative Control over critical Information CCMony a Relative Control over Critical Monies S = strong relationship 0 = no relationship direct relationship indirect relationship 220 constructs parallel the relationships between material resource control and other constructs. As control over both monetary and other material resources are granted by hierarchical superiority, it is hypothesized that the two constructs are strongly related. These hypothesized relationships among constructs were reflected in the empirical examination of relations among measures. The list of the relations that were empirically examined is presented in Table 14. These proposed relations include the individual difference characteristics of need for dominance, need for affiliation, and intent to turnover, which were not part of the formal model of antecedent variables in the nomological network. They are numbered in the order of their primacy in supporting construct validity. The next step was the empirical evaluation of the relations among measures. These were examined to identify the extent to which they match the proposed relationships, listed in Table 14. Part 3: Empirical Evaluation of the Revised Model This section presents the empirical relationships among variables used to measure the different constructs in the nomological network: correlation coefficients and regression analyses were used for this purpose. Due to the subjective nature of the measures of most of the variables, and the fact that all measures were drawn from a single questionnaire, response bias was likely. In particular, social desirability was a likely contaminant of responses: power holders may report higher levels of power, authority and dependence than actually Relation 2a 2b 2c 3a 3b 3c 3d 3e 3f 39 3h 4a 4b 221 Table 14 List of Relations Among Measures Independent Construct Measures Work-perf Deped Job-maint Depend n Dominance n Dominance n Affiliation Intent to T0 Control of Info Control of Mater Control of Money Control of Info Control of Mater Control of Money Control of Info Control of Mater Control of Info Control of Mater Control of Money Dependent Construct Measures Specific Power specific Power Specific Power Specific Peer Specific Power Work-perf Deped Job-maint Depend Work-perf Deped Job-maint Deped Work-perf Deped Job-maint Deped Specific Peer Specific Peer Specific Peer Control of Mater Control of Money Specific Peer Specific Peer Specific Power Proposed mdiating Relationship Construct Measures strong - positive - - this relat' p -- stronger than - this relat' p -- negative - strong -- strong - strong - (strong) -- (stronq) -- (strong) -- zero -- strong - complete Perceived mediationn Depedece (both scales) complete Perceived mediation De (both scales) complete Perceived mediation Dependence (both scales) Relationnships placed in parentheses ( ) are indirect relationnships. 222 exists to enhance their self image, while peer recipients may report lower levels for the same reason. In order to rednce these problem, data were collected from both members of each peer relationnship: two individuals' perceptions of each variable were assessed. Comparing different perceivers' views of different variables would rednce the inflation of correlations resulting from social desirability effects. However, both respondents have access to different information regarding the peer, depedece and control over resources relevant to their relationship. For example, the peer recipient may nave another source of critical information of which the peer holder isnotaware, orthepeerholdermaybeabletoaffectthejob security of the recipient through communication with a third party, without the knowledge of the recipient. Because such information may be held by either member of the peer relationship, there is no basis to assume that either perceiver's view of any variable is more accurate than the other perceiver's. To the extent that the two individuals have different information and referents, combining their views would improve the estimate of the construct being measured. In addition, to the extent that social desirability leads the peer holders to inflate responses and the power recipients to deflate them, the combination of responses would cancel out its effects. Therefore, estimates of each variable were obtained by averaging the responnses of both mennbers of each dyad. Because only some of the respondents were paired, the number of respondents was reduced to 262, aggregated into 131 estimates. Five individuals completed all measures encepttheindicesofneedsandintenttoturnover: thus, the 223 regression analyses used to examine the last set of relations involved a sample of 136. Aggregation of responses requires that both respondents were indeed describing the same constructs, although from different perspectives. Evidence of this is provided by the correlations across perceivers for all measures, which are presented in Table 15. This table shows that the correlationns across perceivers vary between .36 and .78: the strongest inter-perceiver correlations exist for the measures of peer. These relationships are far stronger than those typically found between supervisory and subordinate descriptions of behavioral variables. For example, agreement between subordinates and supervisors on perceptual measures of leader benaviors have been found to range between .00 (Mitchell, 1970, using the Leadership Behavior Description Questionnaire) and .04 (Besco & Lawshe, 1959) . The correlations among perceivers found in this stndy fall within the range of those typically found between perceivers within one hierarchical level , when describing characteristics of leader behavior or their subunnit. Day (1968) examined perceptual agreement among pairs of co-workers on descriptions of the same supervisor's behavior, using the LBDQ: correlations ranged from -.02 to .73, with most falling below .50. Jones and James's (1979) review of the organizational climate literature found that typical inter-rater reliabilities for perceptions of work group climate range from .06 to .35 (p. 207) . Thus, the perceptual connsistency of these measures is quite strong when compared to other well-validated measures of inter-individual relations and subunit phenomena. There is sufficient agreement to support the aggregationn of the dyadic perceptionns . 224 Table 15 Agreement Across Perceivers on All Measures Issue specific power measure: Issue generic peer measure: Work performance depedece: Job maintenance dependence: Perceived authority: Control over critical information: Control over critical materials: Control over critical monies: N= 131 Correlation Across Perceivers .70 .78 .46 .51 .46 .48 .42 .36 225 Testinlthe Proposed Relations Having aggregated across perceivers , the relationships among the sourcesofpeerandmeasuresofpeerwereexamined. Tab1e16 contains the zero-order correlation matrix used for this analysis. Alpha coefficients for the aggregated scales are presented in the diagonals. Primary Ievel Relations: Among Peer and Depedence Relation 1. It was proposed that depedece and peer are strongly related: there are two measures of the fornmer connstruct. The two correlationns that are used to examine this relationnship are .53 and .74. The stronger correlation exists between the power measures and depedence for job mainteance. Both correlations are above .50, the level Cohen and Cohen (1975) hold to indicate "strong" relationships (p. 56) . This proposed relation was supported by the data. Secondary level Relations: Among Individual Differences and Peer Relation 2a. It was proposed that the need for dominance is positively related with power. The correlation between the peer measure and the need for dominance index was .28, and significantly different from zero at the p < .05 level, supporting this relation. Relation 2b. It was proposed that the need for dominance is more strongly related with peer than is the need for affiliation. The former correlation is .28, while the latter is .19. Connfidece intervals were connstructed around the correlation of .28, by transformation to z-statistics (Blalock, 1960, pp. 305-307): they ranged from .12 to .42. The differece between these correlationns is 226 Table 16 Correlationns Among Power and Related Variablesa Sprr Jopr wrkDp CCInfo C(‘Mat (bony Sp Pwr (.99)b JOb Op .74** (.90) wrk Op .53** .54** (.92) CCInfo .52** .44** .70** (.94) CIZMat .68** .54“ .53** .68** (.95) copay .60** .52** .34** .56** .71** (.89) NDom .28* .25* .28** .39** .30M .13 NAfil .19* .06 .04 .13 .18* .12 Int T0 -.12 -.18* -.10 -.12 -.16* -.16* Sp Pwr = Topic-specific peer measure * p < .05 Job Up = Job mainteance dependece ** p < .001 er Dp = Work performance dependence (IZInnfo = Control over critical information CCMat = Control over critical materials CCMony = Control over critical monies NDom = Need for dominance NAfil = Need for affiliation Int To = Intent to turnover 3 all scales are combined across perceivers: N = 131 alpnas for combined variables are in the diagonal 227 not statistically significant. Thus, the evidece is in the hypothesized direction, but the differences are not strong enough to support relation 2b. Heever, the low internal connsistecy reliability of the need for affiliation measure (alpta of .42) makes any conclusions regarding this variable suspect. Relation 2c. This relation proposed that the intention to turnover was negatively related with peer. The correlation between the measure of power and turnover intentions was -.12, significantly different from zero at the level of p < .10. Thus, this relation received only weak support fronm the data. Tertiary Level Relations: Among_Resource Controls, Dependence and Peer Relation 3a. It was proposed that the control over critical information was strongly related with dependence for both work performance and job maintenance. The correlations among the measures of these variables were .70 and .44, respectively. One is substantially larger than .50, indicative of a strong relatienship (Cohen & Cohen, 1975) , while the other is somewhat smaller. This relationn was partially supported by the data. Relation 3b. This relation held flat the control over critical materials was strongly related with both types of depedence: the correlationns between the measures were .53 and .54, supporting this prOposed relation. Relation 3c. This relation held tnat control over critical monies was strongly related with both types of dependece. The correlation with job mainteance depedece was .52, but the correlation with work 228 performance dependece was only .34: only the former of these may be considered a strong correlation. Thus, this relation received mixed support. Relation 3d. This relation prOposed ttat the control over critical information was strongly related with the measure of peer. The correlation between the measures was .52, supporting this relationn. Relationn Be. It was proposed tnat the control over critical materials was strongly related with power: the correlation between the measures was .68, strongly supporting this relation. Relation 3f. This relation held that the control over critical monies was strongly related with power. The correlation between control over this resource and the measure of power was .60, strongly supporting this conntention. Relation 39. It was proposed tnat the control over critical information was unrelated with the control over critical materials. The correlation between measures of these two variables was .68: this relationn was not at all supported by the data. Relation 3b. This relation proposed that the control over critical materials was strongly related with the control over critical monies. This relation was strongly supported by the data, with a correlationn of .71. sen-enr- The zero-order correlations provided substantial support for relationns 1, 2a, 3b, 3d, 3e, 3f and 3h. Partial support was found for relationns 3a and 3c, and weak support was found for relations 2b and 2c. Relation 39 was not supported. 229 Fourth Level Relations: Depedece as a Mediator. The last set of relations proposed tnat the relationships between peer and resource conntrols are mediated by depedence. Because all measures are likely to share variance due to the measurement technique, a test of mediation was used that can account for high co-variation among measures. A two step hierarchical regression analysis was used (Cohen & Cohen, 1983: Kozlowski & Farr, 1988) , to examine both the total effects and the unique effects of the indepedent variables (resource controls) and the proposed mediating variable (depedence) on variance in peer. In the first equation, the independent variables are entered in a first block, and the mediating variables in a second block: a significant increase in variance in peer accounted for by the second block indicates that the depedece measures nave a significant unique effect on peer. A second equation is then constructed, in which the mediating variables are entered in the first block, to identify their indepedent effect on power: the independent variables are entered in the second block. If there is complete mediation, there should be no significant increase in variance accounted for by the second block: tnat is, resource conntrol should not nave a significant unique effect on power. These regressions are presented in Table 17. They indicate incomplete mediation, as the resource controls had a significant unnique effect on the measure of peer. These findings did not distinguish among the three types of resource controls: they sheed tnat all three resource controls together had a significant unique effect. Separate analyses were also Table 17 230 Hierarchical Regression of Resource controls and Dependence on Power R2 Total R2 Unique Combined RC RC Dep RC Dep and Dep N = 136 * p < .05 ** p < .001 RC = Resource Controls Dep = Dependence Table 18 Hierarchical Regression of Individual Resource Controls and Dependence on Power R2 Total R2 Unique combined Mater Mony Info Dep Mater Mony Info Dep RC & Dep Ctrl Ctrl Ctrl Ctrl Ctrl Ctrl Power .65** .46** .56** .09** .19** Power .63** .36** .56** .06** .26** Power .59** .28** .56** .03* .31** N a 136 * p < .05 ** p < .001 RC = Resource Controls Dep a Dependence 231 condnctedtoexaminethemediatingeffectofdepedenceoneachofthe three types of resource controls. These are presented in Table 18. Three separate sets of equationns were used for this analysis: each set identified the total and unnique effects of depedence and conntrol over one type of resource on peer. Relation 4a. This relation held flat the relationship between the control over critical information and peer was completely mediated by dependece. The unique effect of information conntrol in accounnting for variance in power was an R2 of .03: this was quite small, but significantly different from zero at p< .05. Therefore, this relation was not supported by the data. Relation 4b. It was proposed tnat the relationship between the control over critical material resources and peer was completely mediated by depedence. The unique effect of material resource control was .09, which was significantly different from zero at the p< .001 level. This relation was not supported by the data. Relation 4c. This relation held that depedence fully mediated the relationship between control over critical monies and peer. The unique effect of monetary control was .06, significantly different from zero at the p < .001 level: this relation was not supported by the data. None of the fourth level relationns were supported by the data. In all cases, the unique effect of depedece was far stronger tlan the unique effect of control over resources: however, all three types of resource conntrol lad a significant unique effect on peer. Thus, there was evidence that depedece partially mediated the relationships 232 between resource controls and power, but there was not evidence of complete mediation. m This section presented the empirical analysis of the data gathered to assess the construct validity of the newly deve10ped measure of individual power in organizations. Examination of the measurement model indicated that a single factor underlay responses to all six issue-specific power scales: the measure was therefore collapsed into one scale. It.was also found that measures of authority, dependence and power were not distinct: one authority measure overlapped with power measures, and the other authority measure overlapped with dependence measures. The examination of relationships among variables was simplified in response to these findings. The hypothesized relations among constructs were empirically tested by examining the relationships among measures. Four levels of relations were examined: the first three levels examined direct relationships, tested by correlations among measures. Results indicated either moderate or strong support for nine of these twelve specific relationships: two of the second level relations were only weakly supported, and one of the third level relations was entirely unsupported. Fourth level relationships were examined by means of a recursive hierarchical analysis: none of these relations were supported. The next section will discuss these findings, and draw'conclusions regarding the construct validity of the measure of power. Implications 233 of these findings for the measurement of other constructs and for the nomological network of antecedent constructs will be presented. ‘VI. DISCUSSION This chapter discusses the implications of the results presented in the preceding section. Empirical results generally supported three of the four hypothesized relations among measures. Methodological limitations of the study are presented: in particular, due to the inability to empirically distinguish authority from power and dependence, the testing of relationships was modified. Other potential methodological problems are discussed, and the techniques used to minimize their effects are described. Last, the implications of these findings for the measurement, theory and models of the causation of power in organizations are discussed, and directions for future research are identified. Testing of Empirical Relationships This study sought to identify the construct of power in organizations, and validate a measure of that construct through examination.of observable elements of a nomological network. The relationships among operationalizations of constructs were ordered by their centrality in this network. The first two levels involved variables that were directly related with power: the primary level examined the direct antecedent, dependence, and the secondary level examined relations between power and individual difference 234 235 characteristics. Tertiary level relationships involved the indirect sources of power, the control over critical resources: their relations withdependence, powerandoreamtherwereexamdred. Thefourthand last level examined the role of dependence as a mediator in the resource control-power relationships . The primary relationship was strongly supported by the empirical results: measures of power and dependence were strongly correlated. Two of the three secondary level relations were supported by the empirical results. The relationship between the measure of power and the need for dominance was significant, strongly supporting ore relation. The measure of power was negatively related with the intention to turnover: the relationship was significant at the p < .10 level , providing weak support for this hypothesis. Iast, the power measure had a slightly stronger relationship with the need for dominance than with the need for affiliation: the difference was not significant, but the trend of this relationship was in the hypothesized direction. The trends for all of these relations were in the hypothesized direction: the magnitudes of some findings were not sufficiently large to indicate statistical significance. The third of these hypotheses involved a previously validated measure of the need for affiliation, which had a fairly low internal consistency reliability in this study. Thus the third relation must be viewed as inconclusive. A larger number of respondents, and more reliable measures of the need for affiliation, would identify whether the relationships found are due to chance. 236 The third level relations were either strongly or moderately supported, with one exception. Measures of the control over critical material resources, informational resources and monetary resources were all strongly related with measures of dependence and power. However, it was predicted that material resource control would be unrelated to informational resource control: instead, measures of the two were strongly related. One explanation for this finding is the possibility that the control over critical information and the control over critical materials stem from the same sources. For example, holding a supervisory position might grant both control over necessary materials and access to information needed to c0pe with problems, such as fore- knowledge of upcoming changes, or access to others who have expertise in problem areas. Another possible reason for this finding is shared method variance among the measures of the two types of resource control. However, different question formats and different perceivers' estimates were used in an attempt to minimize this possibility. The fourth level hypotheses held that dependence mediated the relationship between resource controls and power. All of the relations among measures failed to support this: there was evidence that dependence partially, but not completely, mediated the relationship between resource controls and power. There are several potential reasons for this finding. First, it is possible that the measures of dependence did not capture all of the relevant aspects of the construct. Dependence upon another for two desired outcomes, successful work performance and job maintenance, were 237 measured. It may be that full mediation would be found if the measure included dependence for other outcomes. For example, one may depend upon another for esteem, for a stress-free work situation, or for access to higher level organizational members. Any of these might be relevant in the mediation of the causation of power. Another possible reason for the finding of partial mediation is thatthemodel ofthecausesofpoweriseitherincorrector insufficient. Other research has not directly measured dependence: perhaps the control over critical resources has a direct causal effect on individual power, as well as an indirect effect through dependence. The mechanism through which this would occur has no theoretical grounding, however, nor is one readily apparent. Another possibility is that the model of the etiology of power may be incomplete. The finding of partial mediation could be explained by co-variance between resource control and power that is caused by unmeasured variables that have a causal effect on both constructs. In particular, there may be factors that directly affect resource control, dependence and power. This section presented the major results of this study. Moderate or strong support was found for most relations at three of the four levels of primacy in the nomological network. Possible reasons for the non-supported relations were suggested: several of these will be discussed in more detail in a later section. 238 Potential Limitations of the Study The findings of this study may be limited by several methodological issues. These include the inability to empirically distinguish authority from power and dependence, the possibility of common method variance, the inability to empirically separate power over different content areas, and the possibility of incomplete measurement.of dependence. Measures of authority were not found to be empirically distinct from measures of power and dependence. Factor analysis of all measures of the three constructs identified two underlying factors, rather than three. Measures of authority were split, such that the hierarchical measure loaded on the power factor and the perceptual measure loaded on the factor of dependence items. This divergence between the two measures of authority, combined with the extremely high correlations between the generic and specific power indices, support the notion that the measure of hierarchical authority loaded on the construct of individual power. There are several possible reasons for this finding. First, the high loading of perceived authority on the dependence factor may have been due to covariance in responses caused by the shared questionnaire measurement format. However this is not likely, as any shared method variance would also be shared by the perceptual measures of power: these measures appeared in a different factor. A second possible explanation is that hierarchical authority loaded on the power factor because it is the dominant source of power between individuals in organizations. There may be an extremely strong 239 relationship between the two, which has not previously been identified due to previous limitations in the measurement of the construct. The sample for this study was selected in such as way as to limit hierarchical superiority to three discrete categories : direct hierarchical subordination, direct superiority and similarity. This measured authority as a relative, rather than an absolute variable: in addition, relationships with hierarchical differences were limited to relationships directly linked in the chain of comnand. This is unique in the research on organizational power in two ways: it takes into account the level of the power recipient as well as that of the power holder, and it limits the categories to those which have clear differences in the right to direct another's behavior. Typical measures of authority are absolute rather than relative. They examine the power of a particular person, at a fixed hierarchical level, over others who exist at a variety of levels. The differences between the level of the power holder and the power recipient are therefore unmeasured and uncontrolled. The measure used in this study controls this difference. This measure has greater fidelity to the construct of authority, generally defired as the right to direct behavior, than a simple counting of the hierarchical level of the power holder. Only direct superiority in a chain of command logically grants the right to direct behavior: someone else's supervisor has far less right to direct one's behavior than one's own supervisor. Authority may be so closely related to the power of an individual as to be indistinguisl'able from it. As the relative technique used to measure the construct of hierarchical authority has not previously been 240 used, it may be that the strength of this relationship could not previously have been identified. Regardlessofthecauses, thedatagatheredinthisstudydidnot allow a distinction to be made between authority and either dependence or power: the measures were not independent. his problem was identifiedbyatwostageexaminationofthedata: themeasurement model was examined first, followed by the assessment of the theoretical model. Because of this process, potential ambiguity of results was avoided by the revision of the relations tested: only those relations were tested in which all measures unambiguously tapped specific constructs. A second potential methodological limitation of this study was the possibility of response bias. Because most of the measures involve subjective descriptions of a sensitive aspect of an individual ' 3 relationship with another, biased responses may result. The possibility of this type of measurement error was anticipated, and steps were taken to minimize this problem. The sample of respondents was chosen in order to identify and reduce this problem. Pairs of respondents were targeted for data collection, and both perceivers' views of all measures (with the exception of personality characteristics and individual difference variables) were combined. Thistechniquewasusedtoobtainimprovedestimatesofthese variables. It was assumed that social desirability would bias estimates in different directions for the two members of the power relationship, and aggregating across the perceivers thus would cancel out some of this error. In addition, both perceivers have access to different 241 information regarding their power relationship, and aggregation was trought to improve construct estimates by including greater amounts of information. This technique reduced the statistical power of the hypothesis test, by reducing the number of respondents to 131. It was rot able to eliminate the effects of common metrod variance, or of individual measurement errors such as halo and central tendancy. An alternative technique for dealing with response bias would be to compare reSponses across perceivers: that is, compare the power holders' perceptions of one variable with the power recipients' perceptions of arother variable. This would reduce the inflation of correlations due to individual measurement errors, as there is no reason to believe that both respondants would commit the same errors: therefore a greater amount of error variance would be uncorrelated. However, this technique was deemed inferior to the aggregation strategy, as it would result in the reduction of information in each construct estimate: the knowledge of only one perceiver would be included in each index. It would also fail to reduce the effects of social desirability. In addition, it would result in the same loss of statistical power, due to the reduction in the nutber of respondents, that the aggregation technique produced. And neither comparing across perceivers tor combining perceivers ' responses would reduce common mettod variance resulting from the common questionnaire format. The aggregation procedure was supported on empirical grounds , as the perceivers' views correlated at levels far stronger than tl'ose commonly found when crossing hierarchical levels. The amount of agreement on power measures was quite high: correlations of .70 and .78 242 are extremely strong for perceptual agreement across levels, and even for agreement within levels. This supports the construct validity of measure: theredoesseemtobeacertainamotmtofabilitytoinflueoe that is perceived by both members of a work relationship. A third potential measurement limitation in this study is the lack of erpirical independence of power over the six different content areas tappedbythemeasure. Powerovertl'esixareaswasstrongly intercorrelated, and the two-step factor analytic procedure identified one underlying factor. There are two potential reasons for this finding. First, it may be thatpeoplewhohavepoweroveroreareaalsoterotohavepowerover other areas. That is, altrough the distinctions over these areas exist in peOple's conceptualization of their work relationships, they may tend to strongly co-vary in people's actual experience. A second possibility is that these differeoes do indeed exist in individual level power relations, but the instrument used in this study was insufficiently sensitive to pick up such distinctions. This might be due to the parallel construction of the question formats. However, such distinctions were found with this measure in the repeated Q-sorts used to develop it. At a conscious level, the distinctions among power overtlesesixareaswereclearlymadewmenusingthisinstmmt. Due to the process of measure development, the topic-specific power measure taps the construct of power more precisely than previous measures. However, at least with these respondents, no erpirical distinctions were found among content areas. The breadth of types of jobs held by respondents reduces the likelirood that this finding is 243 sample-specific: respondents existed in eight.different unions, and the sample also included some noneunion.employees. Because no empirical differences were found in responses, the six power scales were collapsed into a single scale. The resulting measure had an extremely strong index.of internal consistency reliability (Cronbach's alpha of .98). Thus, the measure of power is extremely reliable. ‘When aggregated across perceivers, as supported by their strong agreement, the reliability index increased to a substantial .99. ,A fourth potential measurement limitation was the Operationalization of the construct of dependence. As this construct had not been.directly measured before, the measure was deve10ped from the theoretical definitions of the construct: these are quite limited. They include Emerson's (1964) conceptualization, which held that dependence is "1) directly proportional to [a person's] motivational investment in goals mediated by [another], and 2) inversely proportional to the availability of those goals outside the... relationship" (p. 289). Pfeffer (1981) integrated a substantial amount of literature in his description of dependence as "having something that someone else wants or needs, and being in control of the performance or resource so that there are few'alternative sources, or no alternative sources, for obtaining what is desired” (p. 99). [Although.different types of critical resources have been examined theoretically and empirically, dependence on another for these resources has not been.directly assessed. It has been assumed to exist and result from control of whatever outcomes are considered ”the most critical and difficult to obtain" (Pfeffer, 1981, p. 101): this refers either to material resources, or to whatever ”characteristic or 244 capability“ (p. 112) that is needed to reduce uncertainty that stOps the flow of work. Emerson's laboratory work (1964) was an exception to this view. He operationally defined dependence as needing arother for the amount of weighting factor that the person controlled, in order to collaborate arnd jointly win in a laboratory game: it was implied through a laboratory manipulation, rather than being directly assessed. Virtually all other organizational power research views dependence as needing arother for materials, skills or information that are needed to continue work processes. The Operational definition used in this study expanded the theoretical notion of dependence to include needing arother for maintaining one's job as well as for successful work performance. This outcomewascensideredtobequiteimportantinregardtothe dependence of individuals upon one another. Although this conceptualization develOped the construct beyond its previous limitations, it is possible that this deve10pment was not sufficient to tap all relevant aspects of the construct. This section presented four methodological issues which potential ly limit the results of this study. Explanations of these issues and techniquestodealwiththemhavebeenpresented. Thenextsection will present the implications of all of the findings discussed above: measurement issues, theoretical issues arnd model develOpment and testing issues will be addressed. 245 Implications for Theoretical and Measurement Development Implications for Measurement of Constructs Related with Power Duetothenewapproachestakentomeasurementinthissundy, the findings provide interesting perspectives on several problenatic measurenent issues. This study attempted the measurenent of several conceptually related constructs: power, autlority, dependence, and control over critical material, monetary and informational resources. Prior research had assumed that dependernce was the mechanism through which the control of critical resources created power: therefore , it rad not been measured. The attempt to directly measure power and its two immediate causes resulted in a substantial problem in erpirical ly distinguishing authority measures from power arnd dependeoe measures. In order to understand the causation of power, dependence and autl'ority must be distinctly and unambiguously measured. This study found that perceptions of a person' s dependence upon another covaried quite strongly with perceptions of that person's right to direct one's behavior . The constructs of dependence and authority are distinct in theory: techniques should be found to empirically separate them, or else determine that that to differences exist in the reality of employees' experiences. Improvements in the measurement of dependence rest upon the improved specification of the construct itself: this will be considered inthenextsection. However,wmatevertheenhancementsinthe construct, new techniques of measuring dependence that do rot rely upon perceptions, and therefore are not open to method variance shared with perceptions of authority, should be developed. 246 Laboratory research might be better able to differentially assess dependenceandauthority. Insuchasetting, theextenttowhich individuals need others for outcomes they value can be directly manipulated. For example, a laboratory experiment could be designed where subjects perform an intrinsically unrevarding and complicated task, and have the opportunity to obtain the assistance of a cower with either low, moderate or high ability to perform the task. Differing reward structures would manipulate the motivational investment in successfully completing the task: individuals could be rewarded on the basis of task completion, or rewarded independently of task completion. High motivational investment and high co-worker ability would create high levels of dependence. This manipulation could be compared with subjective assessments of dependence: perceptual measures could be validated through this technique. The distinctiveness of dependence from autlority could also be examined, by measuring perceived authority and examining the strength of the relationships. The status and formal autiority of the co-worker could also be manipulated to provide more information. The measurement of autl'ority clearly requires more attention. The lack of empirical congruence between the measures of hierarchical and perceived authority found in this study was unexpected: rowever, it provides new information that may be quite useful for future work. The relative measure of hierarchical auttority more directly taps the construct of authority than preceding non-relative measures of hierarchical level. However, even this measure was not closely related to the perceived right to direct behavior. This implies that the traditional measure of absolute hierarchical level may be extremely 247 misleading: it may be only marginally tapping the construct of authority, and predominantly assessing other factors such as the control over material and monetary resources afforded by organizational stature. Other potential ”contaminants" of the hierarchical superiority measure include control over the flow of communication to other levels (particularly upper levels) in the organization, and prestige. Conceptually, autlority is the belief that someone has the right to direct arother's behavior. It is clear that this right is limited to behaviors that fall within certain role boundaries. It is likely that the right to direct some of dose behaviors is held predominantly by hierarchical superiors, while the right to direct other behaviors is also held by co-workers and possibly subordinates. For example, a co- worker with greater expertise may be vieed as having the right to tell arother employee how to accomplish a task. The perceptual measure of authority used in this study did not specify the different types of behaviors that a power tolder might have the right to direct. Theoretically, the right to direct what tasks are done, when tasks are done and how tasks are done may be differentially held by individuals at different hierarchical levels. Greater precision in specifying the types of beraviors that one ras autlority over would improve the measurement of autrority. Any future examination of the causes of power will require that the two constructs of autlority arnd dependence be enpirical 1y distinguished. However, improved measurenent of both constructs must reflect improved understanding of their content. 248 Implications for Theoretical Development This study identified several phenomena that have interesting implications for the development of the theory of the causation of power. First, it found that dependence did not mediate the relationship between resource controls and peer. This led to the examination of the construct of dependence, and the implication that, altrough this variable is intrinsic to a vast quantity of research on organizational peer, it is not clearly conceptualized. Second, this study identified a distinnction between concepts that had not previously been recognized: control over material resources as opposed to control over monetary resources. The attenpt to directly measure the concept of depedence identified an inadequate understanding of it. As has been discussed, depedencyhasbeenviewedasaunitaryneedforsomeoutcomethat another mediates: the dominant outcome of prior concern was the continuation of the flow of work. The operational ization of the construct was expanded for this study: it included dependence for resources needed to accomplish work, as well as depedence for maintaining one's job. The measure tapped dependeoe for two important outcomes: heever, there may be other, equally important outcomes that individuals are depedent upon others for, which were not included. The limits of this measure reflect the limits of the conceptualization of the issue. The results of this study indicated that depedence for successful work performannce and job mainteannce were enpirically distinnct, and had indepedent effects in mediating the relationship between resource control and peer. Future research snould examine whether other 249 outcomes are equally relevant for individuals at work. A likely third outcome is esteem, which results from the power tolder's mediation of erotional support or high regard. French and Raven's (1959) ”referent peer“ tolds that peer stems from a personn's being held in high regard. People do what such people espouseandaccomodatethembecausetheywishtobeseenas (andsee themselves as) similar to the admired ones. This implies that the- peer recipients are depedent on the admired individuals for high self regard. Emerson's (1964) post-hoc explanation of boy scout behavior was based on dependennce upon a group for high regard. In his formulation, respect was the outcome for which individuals depended on groups. This gave the groups peer over the individual. DeSpite the recognition of the importannce of this outcome in these early works, its importance at the work place ras not been assessed. The notion of depedence upon another for esteem, or any form of emotional well-being, has been ignored in the organizational literature. To the extent that peOple form attachments at work, and develop friends and social networks among their co-workers, the importance of this outcome is heighteed. Unlike either work performanoe or job maintenance, it may not be related to hierarchical superiority. However, itmaywell coinncidewithtlosewhohavetheexpertiseto solve problems at work: that is, it may be correlated with one type of control over critical information. Future theoretical and empirical investigation of the development of peer should include this aspect in the construct of depedence. 250 A second unsuspected finding of this study was the independence of different types of material resources. Previous conceptualizations of resourcecontrol cmsideredmonetaryresourcestobeonespecifickind of material resource, and held that control over it is essentially the same as controlling other materials (Pfeffer & Salancik, 1974; Salancik & Pfeffer, 1974: Pfeffer & Leong, 1977; Kotter, 1977; Kanter, 1979). This study implies that this may not be true. Empirical analysis found that control over money was quite distinct from control over other material resources, including services, supplies and equipment. Because previous research failed to compare control over money with other material resources, this distinction is new. The two types of resource controls were differentially related with dependence for performing one' 8 work. Control over monies may thus have a different causal effect than control over other material resources. Traditional assessments, following Salancik & Pfeffer (1974) and Hinings, Hickson, Lee, Schneck and Pennings (1971) would hold that the relative effect of each in creating power will depend upon the types of resources which are scarce in any given organization. However , it may be that monetary resources do not fit this pattern. Even when money is not scarce, more of it is continually sought by most social units. Thus, it is always treated as a critical resource. The relative effects of controlling each type of resource in different organizations and environments require further exploration. 251 Implications for Model Development This section presents the implications of this study for tie model of the causation of power, develOped from the review of the previous literature. This model guided the relations examined in construct validation. It includes the essential components of the theoretically and empirically supported causes of power. The results of this study indicate a clear need for further development of the model. A major element in this model was the notion that the control of critical resources creates dependence, which in turn grants power. However, empirical results found that dependence did not completely mediate the resource control-power relationships. If the construct of dependence has been adequately measured, it is possible that this model is wrong. Although a large prOportion of the research on organizational power holds that dependence is tne mechanism through which resource control creates power, perhaps it does not fully mediate the effect of control over resources on power. However, the theory provides no rationale for a direct effect. Logically, the ability to control access to critical information, monies or materials has no means of engendering power unless the recipient desires those resources, and would have difficulty acquiring them from elsewhere. The alternative, more plausible hypothesis is that the model is incomplete. It is quite likely that unmeasured variables tlat grant control over resources are also direct causes of authority, dependence and/or power. This would account for the lack of complete mediation. There are several variables that are likely to have multiple 252 associations with various constructs in tne nomological network relating the antecedents of power. The characteristics of the work performed by employees is likely to have a causal effect on tne relative control over critical resources, and may directly influence authority as well. The work may require certain types of abilities, skills and information, which may be needed to solve problems that stop others’ work. In this way, tie job would affect the KSAs (Knowledge, Skills and Abilities) of tie incumbent: these KSAs would constitute critical information. The work may involve tie creation of material resources or tl'e acquisition of monies, which are needed by others; alternatively, it may grant the ability to determine the use or flow of such material or monies. In either case, the work would directly grant the ability to control critical resources. Tl'e responsibilities of tre job will also directly affect the authority of the employee to make decisions over certain areas, and to direct tre behavior of others. Thus, the type of work done would affect the control over all three types of critical resources, as well as authority. Relevant characteristics of a job which might have such a causal effect on power include tte responsibility for equipment, decisions and individuals, and the levels of knowledge, skill and abilities required. Anothertypeofvariabletnatisnotinclndedintlemodel, butis likely to affect resource controls, dependence and power is individual personality. Specifically, Machiavellianism (House, 1989) may be an unmeasured variable that ras substantial effects on power. Machiavellianism is likely to cause individuals to differentially seek control over both material and informational resources; high Mach 253 pepple are far more likely to seek out control over the most critical resources in any environment. High Mach individuals are likely to seek tasks and work roles that inwolve the direction of others. They may also expand their authority through the manipulation of role definitions (e.g., changing their tasks, or the flow'of work, in such a way as to require that many others' actions and decisions be subject to the power-holder's review). In addition, high Mach individuals are far more likely to engage in information manipulation, in order to change others' perceptions of the options available and the ramifications of various actions. They are more likely to lie to achieve their goals, and use whatever tactic is necessary to win. By manipulating a person's perception of alternatives, they may create the perception of dependence with out affecting resource controls. In addition, Machiavellianism may have a direct causal effect on power through tactics such as lying and manipulation, without the mediation of dependence or authority. Thus, both the characteristics of the work performed and Machiavellianism of the individual employee may have direct causal effects on resource controls, dependence, authority and power itself. The hypothesized relationships between these two characteristics and the other elements of the model have been outlined above. Empirical analysis of these relations should be conducted, and the results used to expand the model. Iterative additions such as these may be necessary before the causation of individual power is fully understood. APPENDIX A SUPERVISOR POVER HOLDER QUESTICNNAIRE 254 APPENDIX A: SUPERVISOR m RIDER QUESTICNNAIRE mmunmmmmmmammumm. ”alumna: yun'rolatiauhipwith ,udmallmnfcrrimtothhm mummmnlatiafldpwidnthhmany. mmmomplmlyoonfldmtm: plouomailtlaemindaceloudmlop-umuitism. “yum-Imam than! PM E. andqroud intonation. l. lbwlanglavoymbma-hrofthisorglnlntim? you‘n/lunths (plot-carols”) 2. Watt-yearns? ale/m (pl-u. circle on.) 3. hddsprmmdirsctnhordimto? yon/no (plausireloau) PARTII. mummmnmmmonmnmywmwmaemmon FRI;- 'Iworklih. PluucuelodamrthatbutwtbmmtofinnmmhIwm a stunts-poctofworkflorthisp-rson. l-nono 4-agrsntd-l 2-alittlo s-oulplcu 3-amodcnummt 4. bunnnmflucesdoyoulavoowrmttamthhmbumdnyin his/harm S. MMMWdopummnowthisma-“atw 6. ”whammdoymemthism'sptmndngamluatim? Ho-o-ow-o . 7. hadnhnflmbywlawmthism'snhm N N N N N G u u u u u I. a a «b wk .- s. hmmflmdoymlawmlowthismmachimsstwuk? l: 9. hadninflmbywmmmmesp-rmmimbr ”CW? 1: 10. bufihflmdoywhmmdmthismmm? 1 ll. ”Whambymhwmmmhmmmmun l: 12. ”winnumcodoyouMwovcrummhmwcrhdth? l 1 l3. hummdoyNMwmfidsp-rmstm :9 N N N :0 u u u u u b b b a Q Q 14. mammammmmemm macaque-this mmdflfi 1:2:3:4: 15. mummaomhmnmmmmmmm mm? l:2:3:4: 16. ummWammemmm-Wuum l:2:3:4: l7. budhflmdoywtmomuatthum'sfimm 1:2:3s4: 18. Mnehlnflueesdoywlavoovsruutthispcamhsstdiffmum WWdIYI 1:2:3:4: l9. humane-60mm“nuptiaityofthispurun'sdiffmmh? l: 20. ”MMu-e-doymhswmtbmofdhdifmmm puma-«wax? N to U as on 21. manhunmdoymMomhtthisp-nmaystommatwork? 22. ”Immanubymmmthmofmthhmhnmmin? H :- :- w- . to N N N N on so u u u 9 fi fi . O a 23. bMMmdoyunlmnommmmmm'smu? “GUM“ 255 24. hadinfluncudoywhwmlmfldsmmwoh/duum 1:2:3:4:5 25. minuchinflueeodoymnavoov-rlowoftmthhmmmdatwith friable“ 1:2:3:4:5 26. Ebwnrsninflueendoymlavoovcrlowdifficultthism'swukifl 1:2:3:4:5 27. mummnmdoywmmflummdtomtmsmwm g.- I. N so In! a fi 00 u: 28. bufihflmbywlavoovcfiothismuu launch):- weaklwith? 29. budinflmbywhwmmmmnmmpcmdouatwuk? 30. ”womanldommcmthismatm 31. Ebwuehinflmdoymlawmmidnpcplothhmlochlimwithatwum HHFHH I 32. hmmflmdoymhwmthism'smw 33. mmmnwdoywlawmdattinfldsmmmorlmm g.- u 34. hadhflmdoyouhavocvormmmismgcuhnmrydsw 35. mummnueecdoyunnavoovsrnwtmspcmaganimhis/htwuk? 1: g..- u u N N N N N N N N a U U U U U U U U u b :5 «b b «b fi 5 «D co 0: M u: M u: U: u: M 36. ”winnmdoymhawmrmmmmdaudflnm poplowhiloM/duisacttallywuking? l: 37. mammnueeodoymhawmrlowlmolwdthiopsrmiswlthhisfimwork? 1:2:3:4:S N .0 U to . on m 38. mammamdoywmwmflatmthhmmtogct his/YUM”? 1:2:3:4:S 39. munchinflu-eodoymhavsovcrlovatisflodthismhwithmher MM? 1:2:3:4:5 40. acwuadnpowurdoymhvoovsrthispsrmatwuk? 1:2:3:4:5 mung. mmmmmmwmdumjcbaMymrrshtiauhipwimthhpm. PIeaumthnscalobolcwtoanlw-radnqwmtian,udcircloflemuatcorrupudswiflnMMyou agrsowithadnltat-mt. 1-stru'91yagroc 4-diagrso 2-agn- S-mmslydinsn- 3-noitleragrecnordinagros 41. answering w position, I have a legitimate right to upset that my mggutions willbocarriodant. l:2:3:4:5 42. Itavothorighttotsllthispsrsonwluttamtocbatwlattim. 1: N so u so ‘5 so u: 43. Thispsrsmmuaflyacc-ptstladocisiaulmabwtlmthingswghttobs donsatwork. 44. Thispcrsmgmllyfollcauyidouabwtlnvtosolvopnoblmatm. 45. Mommit'smbrumuuhmuatb/mmmdoatm. 46. mmmmmuumwummmmgumAumm. 47. flitrnwtu,tlnispsrscnwcmdlavoanx$:hardsrtimbinghisflur job. 48. IsolwprohlmflatstcpthhmfrulqottinqhiI/hcwukdom. 49. Map-rmlcdutomtoflndaxtdamctcffisctiwuytobthmgsatwork. so. Ilotthinpsrsmhowwtat'sr-llymtinuruofdolngagoodjob. 51. Ioftmgotloldotmcuflorflunpumflath/shummb autumn-r0. 1:2:3:4:5 HHHHHHH -8. NNNNNNNN so uuuuuuuu O. Obfibfibbh ” uwuwuwuuwuwuwuw pa 0. 256 52. “mmmmqudmmIMphh/hcflndauyto handlcthea. 1:2:3:4 53. HIWthnmmmtdolnyagacd-ufinjob, itwouldhnrt his/lnrdweuofmyingm. 54. IfIwInt-dto,1ou:1dm1yboabhtoqctfllhpsrmfird. 55. It‘smfu'flnilmtokqnmhmmmmm. 56. Myopinimofthismmmofhctmlowmhh/hu‘jdah. 57. Iwillproblblylmkfa'amjobintlamty-r. 58. Tommmwtthq. H :- H w-o o-o H N N N N N N u u u u u u u . I § Or fi b O b may. mmmmmommmmmmtmm. numbed: each s a E E 3 9 i 5 b. humor-often? (”anytime-mount“? 0.9.,twicouoftn.) dreamer-often. 60. a. dmmispsrsm,whidnauofyouallsthloflarmorsoftm? yw/thispurm/tlamm (mama) b. thalamus“? (mumnytimmporwodt? caucuses-often.) tin-ammo“. 61. a. mdmummmamammmmwmmmunnwm: you/thispersm/tl'osmmm (manage) b. biennium-soften? (roamnytimmoropcrwodc? aquatic-aorta.) timmoftm. 1-nadnmtluncvoryuua1u 4-m1ustlmcvaryomslu 2-thflanmymolu 5-m1mthanmryonsolu 3-aboutthomascvuyomolu 62. Planidmtitytlemurialsgg inflatflnisp-rmmdsflumostinordartoqcthis/lnrwork due. Mama's-you mutate-lemm— a.gotaccustothen? 1:2:3:4 b.qotthn£orthhpcrom7 1:2:3:4 c.1nflmdccisiauabentwnogouw 1:2:3:4 63. Plenum-nifytlugnipnmt,tcolsor tlatdnispsrmnosdstlomostinordcrtogot his/hormone. murdtomryomdioyu: wimtouatcxtmtcan—ym a.gotaccoutotlun? 1:2:3a4: b.9ltfilltorthilpntm7 1:2:3:4: c.1nflueendscisiauabentwlogmm 1:2:3:4: 64. Plauidcntifythsmvicughlgfmloflar otl'atthisp-rnmmdstlamostinordnrtogot his/Mrworkdmo. Wow” towtatutcntcanyou_ ~1.g¢tmtoth¢n? 1:2:3:4: hgotflubrthhm? 1:2:3:4: c.1nfluesdcciniauabmtwtogotsfld 1.2.3.4. 65. Plano identifier-Mott}:- wim(o.g.,rcaullghting, spec,otc.)tlntthis mmgmmmmgu Wei—thathavimflmbudiffumtmightstopthis magnum/hormone. Wmmcluyouworkwith,towmtoxtmtoanyou a.g¢taccastothea? : till : 1: b.gct bitmap-roan? 1, c.1nflmdchimsabmtuogcuw 1. ””N as... “U“ so... ... so UMMMUM offluflollowingqmtiaunmdnflntputbycimlhgmm,udmmprt now: “U“ “M“ MD“ '- 85. Ifindmolfmkingtotlouamduabcutnarbminaurmtcdnttm. 257 1:2:3:4:S b.90titfiorthispursan? 1:2:3:4:5 c.1nflmdochia'mwknqcuit? 1:2:3:4:5 “probluoceratwcrk, mmhmnymmulwflmormmdcumm Immaculwithfl-n. mmmmmnwmwmmmoms 67. doyouhavothouportiutouolvommtoanstcpthispursonfrm hisflnrgcttingwukdea? 68. doyoulalpthilpcrsmtogctuputiutololvo-jorgobluatwork? 69. doyoulawaccsutoinformtimtlathnudodto-olvonjaprobluatwum H H H H l. O. N N N N u u u u u 0. .0 I) ch ch b so u: u: U: u: 70. dompauminficnntimthatumddtooqswdthnjorpobluatwork? 71. canyougcttalpfrmothcrp-oplouohvothcapsttiutosolvonjnr “network? 1:2:3:4:5 72. bywflndantbowmptohl.thltnymupatm7 1:2:3:4:5 73. anympdifidanguflatmpmgtonhpoblufiorflusp-rmatwork? l:2:3:4:5 74. mymwillingtolctthbporlmmbcmdofdangcdatmgoim tommlubthinfinrattotk? 1:2:3:4:5 7S. mywablotoanticiptsproblubcfonthnylappm? 1:2:3:4:5 76. doyouunailyfioro—umthispnrsonofpoblubcfionfleyoeuqfl 1:2:3:4:5 9mg. miss-etimuksabentlowywukotom. Plauuutlaulobalwtoducribsmalfat WE. l-alwayu 4-ulnhm 2-usun11y S-nlvsr 3-mtim 77.nm1hawadoico,1trytowcrkinagro:pinltadofbymyulf. 1:2: : : 78. Isukunactivorolointnaladorshipofagrunp. 1:2: : : 79. Ipsyagaoddalofattmtimtotlefuling-ofotlnrsatwork. 1:2: : : so. IavoidWirqtoinflumduoamxdutou-Mmquy. 1: 81. Ipofcrtodomyomworkandlotodmdothirs. 82. Ifinduyulfcxganizinquddirocdmdoactiviti-ofothrs. 83. Immdiagromflwithoflmmly. 84. IntrivotogainnoroomtrolovortlomtIaruIduatwu‘k. HHH O... NNNNNNN O.“ uuuuuuuuuu ” bbfifi&b&§hb so. uuuuuuuwuwuwuw HHH O. 86. Intrivotobo'inmnd'wimI-minginam. PAIGE. Plumdoocalobnlwtomthohllaquuutim. 1-notata1111koly 4-quit01iks1y 2-slight1ylikoly Susanna-unholy 3-muhnt1flu1y 87. mulflnulyisittlatywwillactivolylodcbranujoblnthmtyur? l:2:3:4:5 mmmmmmmn APPENDIX B SUPERVISOR POWER RHZIPIENI' WESTICNNAIRE 258 APPENDIX B: SUPERVISOR POWER RECIPIENT QUESTIQINAIRE WWW Thismrvoywillukqumdaaabwtyunrsladmhipwdflnmofmm. Plano thinkabaatyou'rolatimshipwid: ,andmallqwmtiaa referring to this per-an. quu: this can mly. All remmocnplotslyomfidmtialnpluumildesurvoyinfle “lop-ascends itisoanplcto. Mymmymuyourtiml PART}, WW. 1. Infinispornonyom'dlrsctamordimto? yon/no (plus-circlcom) TARI'II. mmnodngqmctimukywabwtthmtofinnmthhgmmm Mmafmuklifo. Plancir'clothmuatbutruprmflamtof influeothispnrmhuowrudndifhrmtup-ct. 1cm 43‘91‘81’.“ 2-alittlo s-cxnploto 3-auodaratomnt \ 2. thawinflumcodoumismmmuatmbywbmyday 1.an7 1:2:3:4: 3. lwainIIueesdouthispcrmhmnowrlowyoudrc-satworm 1:2:3:4: 4. hmmflmdoumsmhwmmpcmnum ormluatim? 1:2:3:4: 5. wwmflmmmmmmmm 1:2:3:4: 6. hadhnfluceedouthispsrsmhavocwrnmyoumacisiomat work? 1:2:3:4: 7. lbwuudninnusneodouthispcrmnavoowrlomeMt—ceiw forbaunuorovortim? 1:2:3:4: 8. mmmnu-eodoumispsramMIovorMywuk-vacatiom? 1:2:3:4: 9. munchinflmdouthismnawmwkmymtakcbraks from work? 1:2:3:4: 10. ”whammmisp-rmhIwmemymmuth? 1:2:3:4: 11. Manninflu-eodouthispnrmhavocvcrymatwork? 1:2:3:4: 12. munchinflu-esdcuthispcrvaoovormttcols,mipn-nt¢ wppliuyouworkwith? 1:2:3:4: l3. mmmnmdoumhmhwomfluphyflalmimt inwhidnyouwork? 1 N u U “ u 14. ”MMWWMMMWMMMW ”'43” 1:2:3:4: 15. hadhflmdossfiismhwmiatmjobmimofl 1:2:3:4: 16. unfininflumsnbuMpcmlavaMywhatdiffm timldn‘imthday? 1:2:3:4: 17. bMinflmdoafiismMomfiapriorityofyou: 61mm? 1:2:3:4: 18. munchinfluencodouthhmlIwovorthIWe-offla difhrmttuksywdoatwork? 1:2:3:4: 259 19. hadhflucecdouthismmmwhatywaytoothc ”catwalk? 1:2:3:4:5 20. hMMmmmmmmmmmamm havotoworkin? 1:2:3:4:5 21. lbwmdninflucesdouthisporsmhmnovarlowstrmmlywrworkis? 1:2:3:4:5 22. munchinflueeudoaflnisp-rmlawcvcrlmymfulwhiloym marking? 1:2:3:4:5 23. mammamdossflusmmmlmoftaymstcptodat ' withfricdsatwork? 1:2:3:4:5 24. thvuaehinfluee-douthispcrmlavoovarlrudiffientyourmh? 1:2:3:4:5 25. lbduadninflucesdouthismmovorthostududsuudtojwdgs muck? 1:2:3:4:5 26. mammflmdouthnmmmwhoyouatlmdnorukc Width? 1:2:3:4:5 27. Etunudninfluoncodocsthispcrmhavoovorlowm-ocializimyou doatwm'k? 1:2:3:4:5 28. bMomtroldouflnispcrsmhavoovorymatwork? 1:2:3:4:5 29. Mmehinfluaecdoathispurmlavocvcrwhidnpcoplcyousocialiu Withatwork? 1:2:3:4:5 30. Ebwmrehinflusncsdouthispsrnmnavocvaryourcarurm? 1:2:3:4:5 31. tbwnnadninflmdouthisp-rsmhavoowrdattimyougsttoor leavowlrk? 1:2:3:4:5 32. mvnndminflue-ecdouthispersmnavoovsrhadnndnyougctdone everyday? 1:2:3:4:5 33. Ibwundninflueesdouthispcrmnavoovorlmyworganiuyumwork? 1:2:3:4:5 34. manehinflumcodouthispcrsmlavoovcrrmmchyoucnatwith oflerpsoplowhiloymaroactmllyworkirg? 1:2:3:4:5 35. Ebenehinfluaeodcuthispsromlavocvarlowinvolvodymmwith yarn-0th? 1:2:3:4:5 36. machinflusncsdouthismlavoovcwhntprocodmuymuuto -gotyom'workdono? 1:2:3:4:5 37. Machinflueesdouthispsrunlavsovarhowatisfisdymmwith yourworkovurall? 1:2:3:4:5 38. madman-thumhwoovcymatwork? 1:2:3:4:5 39. Whohisfiarposittaudnismmalogitimurighttom dathisfiurcuggaflanwinboarricdmt. 1:2:3:4:5 40. ‘nnispsromlasflarighttotullndatmmdoatwtatdm. 1:2:3:4:5 260 41. quallysccsptthsdscisieuthispsrsenwwssaboutlodthingseaght tobsdensatwork. so N N N N N N n U U U U U U .0 H H H H H H U 0: UI U' 4! UI 42. Igeurallyfollowthispsrsen'sidssssbentnmtosolvoerblustwork. 43. Ithinkit'satbtflnispsrsmtotcllmwlnttoughttobstm. 44. IrslyenthispsrsenbrtldmsInssdinordsrtogstmyworkdens. 45. Hitluntthispsrsenlwenldhavssmnardsrtimdoingwjob. HHHHHH N 46. 'mispsrsensolvose'oblmtlatstqmsfreugsttingmyworkdens. 47. Ilooktothispsrsentofindentfinsmstsffsctivswaytodothings at.work. 1:2:3:4:5 48. Thispsrsexlstsmslcmwlat'srssnyinportantintsruofdoinga goodjob. 1:2:3:4:5 49. Thispsrsenoftengstsnoldofrssenrcssformstlatzmsdtobs succsssfulhsrs. 1:2:3:4:5 SO. Mlmintopnoblenswithwe'k,flnispsrsenhs1pmsfindawayto mam. 1:2:3:4:5 51. Ifthispsrsentlonqhtrwasnotdoingsgocdeowghjob,itwe:1dhm mydanessofstayinghsrs. 1:2:3:4:5 52. Ifflnispsrsemwantsdto,l'a/shsoou1dprobab1ybsablsto9stmsfired. 1:2:3:4:5 53. It'sinportantfornatoksspthispsrsennappyforwlengtsm smash-to. 1:2:3:4:5 S4. 'mispsrsen'sopinienofmalasansffsctenhcwmsmyjobis. 1:2:3:4:5 PARI'IV. Thisssctienasksabemnruoftenymandthispsrsenoentscteaamtnar. Plssssanswsr mindsetssdnofflafollcqumtiemmtlefirupsnbycirdingeerupeus, and thsssoedpsrtbywritinginanumsrbstweenenssndten. 55. a. Qanpsrsdtothispsrsex,whiche:sofye:easstossst1aoueruorsoften2 b. Mundane" often? (Ebwvmanytimasmorspsrwssk? s.9., Mcsasoftsn). timssuorsoften. 56. a. E tothispsrsenunidnensofyoucallsthsothsrmorsoften? thispsrsen/thsmmnt(mamm) morsoftsn? (lawmanytimsmorspsrwssk? s.9., Mosesoftsn). timssmoosoften. in b. 57. a. Wmthispsrmuueneeofyoucmtactsmodertogstinfomtimabent work? you/thispsrsenlthssansmnt(mnszamaml b. mundnmorsoftsn? (bunnytinsuorspsrwosla s.9., twicsasoften). timssuorsoften. Partv. mismtimsskssbwtuoomtrolsflaplmicalthimsorflainfiomtimyoumdm .togstyen‘workdeu.Plssumsthmllowlngsca1stcmrtlauqmstiea. 1-Mnorstlansvsryenss1ss 4-sensalatlsssthsnsvsrye1sslss 2-smswhatuorsthsnsvsryeasslss 5-mdnlssstlansvsryensslss 3-abentthssenssssvsryenss1ss 58. Plsassidsntifythsmtsrialsgr; iss sstlatywnssdthsmostinordsrtogstyenr work dens. Qmpsrdtosvsryonsassyofi' mutsM—an—a'iispsrsen a.9stsccssstotnll:? 1 2 3 4 b.9stthenfotyou? l 2 3 4 l 2 3 4 noses ssssss «no. s00... “M“ c. influeesdscisiewssboutwiogctstm. 261 S9. Plssssidentifythsm,too1sor thstymnsdthsmstinordsrtogst 2 workdens. Qanpsrsdtomryensslssyenwork ,towtatsrtsntcsn psrsen a.9staccssstoth-n? 1:2:3:4:5 b.9stthellforyou? 1:2:3:4:5 c.1nflueesdscisiensabentwlo9ststh-n? 1:2:3:4:5 60. Plssssidsntifythsssnvicesghsl fienothsr sflatyounssdthsuostinordsrtogst yarn-workdens. (hipsrsdtocvsryons ssyoum' towtatsxtentcantm-psrsos a.g¢taccesstotheu? 1:2:3:4:5 b.9stthlnfioryw? 1:2:3:4:5 c.influsncsdscisienssbentwtogststhu? 1:2:3:4:5 61. Plsass identifythsaspsctsofths ical environ-Imam: (e.g., roemlighting,spscs . stc.) thatyounssdflnsmostmommmtmmmtmmmuufmmmm yenfremgsttingyon—E'workdons. Qmpsrsdtosvsryensslssyouworkwith,towhatsxtsntcanthis parse: a. getaccesstothem? b.9stthenfmyou? c. influencs docisiensaboutwtogststhen? NM” “WU ssssos HHH O...“ HUI“ 8 8 8 8 8 8 62. Pleassidentifytnawaysinwhichyennssdmtopsyforgoodsorssrv yen‘workdeo,oruatnotnavin9muuymi9htstopyoufremgettingyezwork everyonnsslssymworkwi‘,tcwhatsxtsntcanthispsrsen a.9stsccssstoit? b. gstitfioryou? c. influence docisiens abentwko gets it? in 3 s 3‘3 n? i ”N” sssoss UUU oossss HHH osssss UIUIUI 8 8 8 Whenproblensoccuratwork, sensinformatienisusuallynesddtosolvsthemortomaksdscisiens amutlwtodealwiflntl‘nen. mmgmassgqmwnm, nomacmnr: 63. dcssthispsrsennavsthssxpertisstosolvsproblenstlatcanstopyou frenyourgsttingworkdens? 1:2:3 64. dossthispsrsenlelpyentogstsxpsrtisstosolvomajoreobleasat work? 1: H 1.” N ss U ss . ss 0' 65. dossthispsrsenhsvsaccssstoinfomtientlatisnssdsdtosolvsmajor problematwork? 1: N so U ss H ss UI 66. dossthispsrsenpsseninfornatienthatisnssdsdtocopswithmajor problmatwu'k? 1:2:3:4:5 67. canthispsrsengsthslpfrenoflnsrpsoplswholavsthssrpsrtissto solvsnajorprobluatwork? 1:2:3:4:5 68. dossthispsrsenfindentbsfiorslandabente'oblmthstnyoensnp stark? 1:2:3:4:5 69. csnthispsrsenesdictdangsstlatarsgoingtomksproblensfior youatwork? 1:2:3:4:5 70. isthispsrsenwillingtolstyoulcwbsforslandofdangsstlatars 9oin9tomaksproblufe'ymatwe'k? 1:2:3:4:5 71. isthispsrsensblstounticiptseoblubsfiorstllylappen? 1:2:3:4:5 72. dossthispsrsenusnnllyforuwsrnyouofproblubsfontluyensup? 1:2:3:4:5 WWWWMMMI APPENDIX C SUHDRDINATE PCNER I‘DIDER QUESTIQQNAIRE 262 APPENDIX C: SUmRDINATE m PDLDER QUESTIGNAIRE askquusstienssbentyermstlendnipwithensofyerms. Plssssthinksbent rustlenshipwith ,sndmsllqusstiensrsfsrringtodnispsrson. mmmmdtmsmeuy. mmmowlstslymfidentm: aminthseelossdenvslopssssoenssitismstsd. WNW-mum an E: g: 1 E E w E | M5. WWW. 1. bileghsvsyenbsens‘srofthisorpnintien? yssrs/unths (plsssscirclsons) 2. unstisyourssx? ulslflells (pisssscircls ens) 3. Isthispsrsenyonrdirsctnpsrvisofl yss/ns (plsssscirelsens) PARTI_I. mmmmmmmmmuWMMWuMup-cuo: 333 'sworklifis. Plsssscirclsthsmllosrthstbsstrmths :3." mmofworkhrthispsrsen. 1-ncns 4-s9rsstdss1 2-s1itt1s 5-m1sts 3-ssodsrsumsnt 4. mammnueesdoyenhsvoovsruattsdcsthlspsrsenbssmdsyin his/YEW l: 5. mammnueesbyunmovsrhowthispsrsendrsssssstnoda 6. Muhambywmmthhm'smnumcmlmum? 7. mummiueumywmovsrthispsrsen‘ssslsm HHH ss NNNMN O. uuuuu - HHHHH . 8. benchinnuencsbyenhvsmnovthisptsenukssdscisieustwrzk? p-s so 9. mummnumcsdoywmmnomethismrscsimbr beusssorersrtin? 10. humanesaymhvsmimthispsrsenmm? 11. lbw-minnuesdoymhvomimthismmmfruufl 12. ibumdninflueesbyouhswmwlusthispsrsenworkswith? H H H H H . N N N N N U U U U U H H H H H ss 13. bflhflueesdaymhvsovsrthispsrsenstwe'k? 14. bummnueesdomrmnovsrmttodls,senimtor.mlissthis psrsenworkswithl' 1:2:3:4: 15. mmmflmbmhmowthsmsmminwhidnthis ptsenwekfl 1:2:3:4: l6. hadninflueesdopnmenrflnrsthispsrsenorbwmdsy? 1:2:3:4: 17. mummflueesbyenlmommtflnispsrsen'sjobmistsofl 1:2:3:4: 18. hadnhflmdammmflatmhmmstmdm ‘ minded-y? 1:2:3:4: 19. hmmbyouMUMdoprie'ityoffinism'sdia-rutude? 1: 20. ”MMWbymmmthdmmmms pe-sendossstwek? N ss U ss . ss 21. ”manuuamm-mmuu-mmmmmum: 22. ”mumbyouhvsmflamofqaosthismhsstomin? H H H H ss N N N M ss 0 u u u H H H H ' 23. hashinflueosdsmhwomtowmmsm'smm mutatinnueesyouhsvsovsr “GUM 263 24. ummammemmmmm/mum 1:2:3:4:5 25. newnehinfluesdoyenhvsovsrlowoftenthispemensuopstodatwith triedsstiurk? 1:2:3:4:5 26. mmmflueesdoyenhvsersrhevdiffimltthispsrsen'smm 1:2:3:4:5 27. hedninflueesbymhsvsovortbstmmsdtojndgsthispsrsm'swork? 1: N ss 0) ss 0. u 28. mammauesdoymlavoovruothispsrsenssuleehortshs trsskswith? l 29. hummflueesdoyunhswmtowmmfldsmdoustwork? 1 30. hMWdopnmowrflnispsrsenstwrxn 1: 31. mammnueesdoyenmenrmfioplsthispcsensocialis-vithstwek? 1 32. munchinflueesdoyenlavsowrthispsrsen'sarssrm 1 33. hMinflueesdoymhvsovrwluttinsthispstsengststoe‘lmum 1 34. bundninflueesdoyoumowrnownrhthispsrsengstsdenssvsrydsw 1: l. NNNNNNNN .0 uuuuuuuu -00 hHHHHH‘h COO. uuuuuuuu 35. Heweehinfluencsdoywlavsovsrlowthispsrsenegsnizsshis/hsrwwn 1: 36. ibunudninnueescbymlavscvsrlnvmdnthispsrsendatswithothsr psoplswhilshs/shsissctmllywe'kind? l: N Is U 0s . M 37. mummnueesdoyenlavswsrlowmolwdthispsrseniswdthhisfiarwork? 1:2:3:4:5 38. ”Minnueesdoyenmovcflatprocsdm'uthispsrsenussstogst his/hsrworkdens? 1 so N so U so ‘ so In 39. munchinflueesdoyenlavsowrtnvsstisfisdthispsrseniswithhis/Mr workman? 1:2:3:4:5 40. MMmdoyenlavsersr-thispsrsenstwek? 1 O. N ss “8 ss & so 0! FAME. MbllwimmudeuaskamutlouymviumjobaMyemuhtieuhipwiththhm. PIsassussthsscalsbslowtoanswsrssdnqusstien,andcirclsthsnunbsrtlatcorrsspedswithrowmenyou agrsswrithsschstatenent. 1-strequsgrss 4-disagrss 2-agrss S-strenglydisagrss 3-nsithsragrssnordiss9rss 41. Qansidsringmypositien, Ihavsa lsgitimots righttoannsct flatmysuggsstiens willbscsrrisdout. l:2:3:4 ss UI 42. Ihsvsthsrighttotsllthispsrsenwkatusletodoatwtattim. l: 43. Thispsrsenusuallyaccsptsfladscisimslmnksabentlowtldngseohttcbs donsatwork. 1: N ss U ss ss 0: 44. MspsrsengensnnyflollowomyidsssabmtTmtosolvspmblustwork. 1: 45. ’nnispsrsenthhfleit'smmrntotsllmmiathldhmghttobatwork. 46. mispsrsennliuenmsfe'dningsb/shmsdsinordctomthisflurworkdeo. 47. Witlonntms,unisp¢senwe:1dlavssn:dnhudsrtindoimhisflurjob. 48. Isolvse'oblutnatstopflnispsrsenfrmgsttimms/Mrmm. _49. ‘nnispsrsenlodetouatofkdentmmstsffsctivouytodothingsstwork. 50. Ilstthispsrsenluothat'srssllyisportmtinteuofdoingsqoodjob. 51. Ioftsngstloldotrssenrcssmrthispsrsenthsths/dnsnssdstobs manners. 1:2:3:4:5 HHHHHH O... NNNNNNNN N uuuuuuuu O. HHHHHHHH ss uuuumuuu 264 52. “thismmhtoewluwiflnmtmhhmtfindsuytc hudlsthll. 1:2:3:4:5 53. IfItheghtthispsrsenwssnoteningspodeo-ngnjob,itwouldmrt his/lurchmcssofstsyinghsrs. 1: 54. quntsdto,Ioe:1deensblybssb1stogstthispsrsm£ir-sd. 55. It'simpmuntforthispcsmtoksqammppyfa'hisfiu'leymmhrs. 56. mopinienotthispsrsenhassnsffsctonnmssenrshisfinsrjobis. 57. Iwillproublylookfie'smjobinflnsnsrtyosr. 58. Ioftenflninksbcutqnittiny. H H H H H O. N N N N M :0 so u u u u u u a O H H H H H H fl 0. u: u: u: u: u u: Pam-Iv. mmmmmommmmmmmm. Plssssansusrboth psm—stsdnoftrefollednqqnstiemmdaflrstprtbydrclkgeum,sndthssseedprt bywritinginsnnbarbshnenensandten. 59. s. mdtothispsrsen,whichensofyenmsstossstluothsrnorsoften2 yen/thispsrsen/thssensmnt (meme!) b. lawman-soften? (mumtimsmorspsrwssk? s.9.,t:d.csssoften.) ' tinssmrooftm. sdtothispsrsen,midnensofyoucsllsthsothsrsorsoften2 Ithispsrsen/thssensmnt (manger) minor-soften? (thumnytimssmorspsrwssld' s.9.,cdicsssoften.) tinssmorsoftm. in b. § 6 61. s. Ctrpsrsd you/ and: thispsrsm,unieneaofyenoentsctst1aotnartogstinmumaboutwork? psrsen/thssensmsnt(mc1mm) often? (bunny times mors psrwosk? s.9., varies as often.) timssmorsoften. f- b. n 9 Part2. Thisuctimuksaboutuocenuolsuaptysialflnimsormmfomtimddsprsmmdsm toqsthis/‘hsrworkdens. Plssssussthsbllowiuascsisuomthsssensstiens. l-nundnmthsnsvsryensslss 4-se-whstlsssthsnsvsryenss1ss 2-senutatnorsthsnsvsryensslss S-nndnlssstlanmslss 3-sbentthsseloassvsryenss1ss 62. Plssssidsntifythsmatsrialsgg; issdatthispsrsennssdsgguostinordsruogsthisflmwrk dens. Qupsrsdtosvsryonsassyouwor towinstsxtentcanyou a.9stsccssstotheu? 1:2:3:4:5 b.9stthenflorthispsrsen? 1:2:3:4:5 c.1nflusncsdscisiensabentwlo9ststh-n? 1:2:3:4:5 63. Plssssidentifythssmipusnt,toolsor thstthispsrsennssdsthsmostinordsrtcgst his/havorkdens. Wtosvsryenssl's'symwo wiflutowhstsxtent _§ou a.9stsccssstothea2 1:2:3:4:5 b.9stth-uflorthispsrsen? 1:2:3:4:5 c.influsncsdscisienssbe:tw0o9ststh-n? 1:2:3:4:5 64. Plssssidsntifythsssrvicssghslgfresod'ar sthstthispsrsennnssdsthsmostinordsrto hisflnu'workdens. moratosvuwnssWye: todatsxtsnntosnyen—— 9“ s.9ctsccosstothen? 1:2:3:4:5 b.9stth-s8orthispsrsen? 1:2:3:4:5 c.1nflusnesdscisienssbentwto9ststh-7 1:2:3:4:5 65. Plssss identifythsaspsctsofths icsl mire-mt (s.9., moonlighting, spscs, stc.) thstthis psrsennssdsggmostinordsrtoqsthis Wmtmtlubsdiffsrentmightstcpms psrsenfrenngsttingfiis/hsrwcrkdens. Wmmsmyenworkwith,touatsxtentcanym s.9stsocssstothen2 1:2:3:4 b.9stthlsh'thisprsen? 1:2:3:4 c.1nnuencsdscisienssbentflo9ststhn? 1:2:3:4 ssss “in“ 265 § § psrsennssdthtopsyfloreaodsorssrvicssin hishuwukdens,e‘thtmthvingmnsynientstm psr'senfrt-gsttinng/Mwuk to ye: n E HHH ssssss NNN “0.. some» sass. HHH sossss 0|qu 67. dopahsvsthssspsrtisstosolvoeobluthtcsnstopthispsrsenfres his/hsrgsttirqwukden? 68. doyenhslpthispsrsentogstsspsrtisstosolvsnjormlustwork? 69. doyenlavssccssstoinfcuatienthstismsdsdtonlvonjcproblu-stm? HHHH fl nun» “ u N HHHH . u: 70. doyenpssseninfiorntientlatisnssdsdtoowswdthnjorpenle-stwork? 71. osnyengsthslpfruothsrpsoplswtohvsthssrpeflsstosolvs-je- metals-stark? p so N . 8.9 ~ H ss (I 72. doyeafindentbsforshsrdsbextprohlutlatnymupstm 1:2:3:4:5 73. mMWWMmMmmMmhrflfi-matwork? 1:2:3:4:5 74. arsyenwiningtolstthispsrsenlcowbsfordwdofdangutlatsrsgoing wmm-lbt'mstM? 1:2:3:4:5 75. myensblstosnticipstsproblubsfiorsthyhsm? 1:2:3:4:5 76. doywmallymmsmoffldalubhrsfllymw? 1:2:3:4:5 PARTY; Thisssctienssksabemlowyenlikstcm. Plssssussthssalsbslevtodsscribsyenrsslfst l-aluy- 4-ss1nhs 2-usnn11y 5-nsvsn: 3-seutims 77. “Inmsdnoics,1trytoworkins9repinstssdofbylysslf. 1:, 78. Isssksnsctivsrolsinthslssdsrshipofsgrep. 79. Ipsyawoddsslofattentientothsfsslingsofothsrsstwork. 80. Iswoidtryingtoinflusncstlosssresdmstosssflningsmyuy. HHH ~ 81. Ie‘sfsrtodowenworksndlstoflnsrsbthsirs. 32. Ifindmysslfe'ganizinganddirsctingthssctivitissofodnsrs. 83. strrsssmydissgrseu-ntswithcthsrsopenly. 84. Istriwto9ainuoroentrolcvsrthssventssresdmsstwrxk. 85. Ifhdmysslftsikingtotlossuesdusboutmninsssrslatsdnttsrs. “86. Istrivstobs'inmd'whenxuwekinginagrep. H N N N N N N N N N N U U U U U U U U U U as H H H H H H H H H H 0. ss M U u: M u: M 48 U 88 UI HHHHH ss PART I. Plssssmsthsscslsbslcwtomthsblledngeassden. 1-notsts111iks1y 4-enitsliks1y 2-slight1yliks1y S-utrealylilwsly 3-seuwhstliks1y 87. Ebwlikslyisitthstyenwillsctivslylodnbrsnswjobinthsnsxtysan 1:2:3:4:5 mmvmmmmmn APPENDIX D SUEDRDINATE m REXZIPIEN'I‘ QUESTIMAIRE 266 APPENDIX D: SUHJRDINATE 1:0me REIPIM QUESTIONNAIRE mmmm E envoywillsskyenqnstieusbentyerrsladendupwitheaofyeroo-worksrs. Plssss Mahatyen'rslatieuhipwith ,udmanusstiens rsfsrringtothispsrsen. mmdsscribsyerrslatiefliipviththis enly. A11 rsqenssssrscouplstslyce:fidentis1:plssssnilthssurvcyinths mlopssssoenss itisoenplsts. Myenvsrymdnmryenrtinsl PART}, Backgresdinforntien. 1. Isthispsrsenyen‘dirsctsupsrvisor? yss/no (plsssscirclsens) 9mg. flufonwinngqmstieusskyenabenttlamntofinnueesthismhssovsr MMaspsctsOfyerworklifs. Plsssscirclsthsnsnbsrthstbsstrsprssentsthsmsntof influeesthispsrsenlasovorssdndiffsrentaspsct. 13:“. “.mm 2-slitt1s 5-enp1sts 3-suodsratsmsnt 2. Ibuuehinflueesdossthispsrsenhsvsovsruattssksyendosvsryday inmwurk? 1:2:3:4: 3. mummeummmemm-um 1:2:3:4: 4. munchinflueescbssthispsrsennavsovsryerpsrfiennesnting ormluntien? 1:2:3:4: 5. ibvnehinflusncsdcssthispsrsenlavsovsryen-sslsm 1:2:3:4: 6. Ebameninfluencsdassthispsrsenhavsovsrnowyouwwsdscisieaat work? 1:2:3:4: 7. mmmueescbuthispsrsenmovsrlodmdnuensyyenrscsivs forbeusssorovsrtinns? 1:2:3:4: 8. tbwusehinfluencsbssthispsrsenhawersrunsnyentaksvacatiens? 1:2:3:4: 9. lbwerdninflueesdossthispsrsennavsovorwhenyentaksbrssksfrcn work? 1:2:3:4: 10. mumeuemmthispsrsennavsmwkmyenwe'kwith? 1:2:3:4: 11. hedninflueesdossthispsrsennavsovoryenstwork? 1:2:3:4: 12. hashinfluenosdossthispsrsenlavsovsrwtattools,sqaipmtcr supplissyenworkwith? 1:2:3:4: 13. beehinflueesdossthispsrsenlavoovsrtlaenysialewim inwhidnyouuodd 1:2:3:4: 14. beehinflueesdossthispa‘senlavsove'mrsyenworkflueohent thsdny? - 1:2:3:4: 15. mwnndninnueesdossmspsrsenhsvoovsruatmjoboeoistsofl 1:2:3:4: 16. mun-unmannedossthispsrsenhwovsrmtyenbstditfsrent tin-immay? 1:2:3:4: 17. burdninfluencsbssthispsrsenlavsovu'flnspricrityofyer dimw 1:2:3:4: 18. hummfmencsdossthispsrsenhvoenrthsimmsoffla ummmbuw 1:2:3:4: 24. 25. 26. 27. 32. 33. 34. 35. 36. 37. 38. Pm PI 267 ”winnuencsdossthispsrsenhsvsovsrwhstyenssytoothsr psoplsatwu'k? hedinflueesdossflnispsrsenhavsmflnsmtofspcsye: hsvstowe'kin? Ibwnndninflueesdossthispsrsenhawovsrhwstrsssfulyenrworkis? minnhinIIusnesdassthispsrsenhsvscvsrrmyenfsslmilsye: arsworking? munchinflueesdossthispsrsenlavsovsrlowoftenyenstoptocrat withfriedsatwork? munchinflueesdossthispersonlavsovsrlowdiffientyetwrrkis? nbdnndninflueesdossthispsrsenlavsomrthsstandudsmsdtojndgs yen'wcrk? “winnueestbssthispsrsenhsvseerwioyensstlmdnortaks broskswith? mammnuemdossmisprsmesrmmmunimm doatwe‘k? Mennoentroldossthispsrsenhsvoovsryouatwork? ibumehinflueesdossthispsrsenhsvsovsrwhidnpsoplsyensocializs withatwork? Meehinfluencsdossthispsrsenlavsowryenrarssrsuccsss? Ibumuchinfluencsdossthispsrsentavsersrwnattimyengsttoor lsavswrx'k? Manninfluencsdossthispsrsenhavsovsrlowwchyengstdens svsryday‘w‘ ”Minnueesdossthispsrsenlavsowrlowyenorganizsyen'work? Munchinflueesdcssthispsrsenhsvoavorludmehyendatwith otterpsoplswhilsyensrsactuallyworking? Ebwmmehinfluencsdossthispsrsenlavsovsrnowimolvsdyenarswith yenrwork? lbwuehinflueesdossthispsrsenlavsersrwlatpmcsdm'ssyenussto 9styenrworkdens? Ibdnehinflwesdoaflnispsrsenlavsovsrneaatisfisdywmwith yenrworkenrall? aewndnpomrdossthispsrsenhavscvsryenatwork? £2. MbllevlmqmsdeuaskaboutMymviwyernlatieuhipwiththispsrsm. Myousgrsswithssdnstatnent. 39. Cbnsidsringhis/tarpositien, thispsrsenhssslsgitintsrighttosxpsct 40. l-strenglysgrss 4-diss9rss 2-s9rss s-mydissgrss 3-nsithsrs9rssnordiss9rss tlathisflarsuggsstienswillbscsrrisdent. 'nnispsrsenhasthsrighttotsllmuattaskstodastwlattim. msflasalsbslwtomrssdnenstim,andcirc1stnammtoorrupedswiflnm 1:2:3:4:5 1:2:3:4:5 268 41. Iusnallyaccspttnedscisieuthispsrsenuokssabenthevthingsenght tobsdensstwork. 42. Igsnsrally ficllowthispsrsen'sidsasaboutlmtosolvs eoblm at.work. 43. Ithinkit'smforthispsrsentctsllmswlatIeghttoeaatwe'k. HHHH O. 44. Irslyenthispsrsenfiorthingsznssdinordsrtogtmyworkdens. 45. Witlmtthispsrsen,1wouldhaveanmehnardsrtimd>inguyjcb. 1: 0. N N N N N N so so U U U U U U 0. H H H H H H .0 U! 0| 0| UI VI 08 46. Thispsrsensolvssproblmtlatstcpmsfrengsttimmyworkdens. 1: : : : 47. Ilooktothispsrsentofindenttl'nsmostsffsctivvswoytomthinm at.work. 1:2:3:4:5 48. Thispsrsenlstsmalcmwhat'srssllyimpwtmtintsrmofdoinga meljob. 1:2:3:4:5 49. Thispsrsenoftsngstsloldofrssenrcssformsthstlnssdtobs snecsssfulhsrs. 1:2:3:4:5 50. Mlmintoproblemswithwork,thispsrsenhslpsmflndswoyto handlsthen. 1:2:3:4:5 51. Ifthispsrsentl'onghthasnotdcingsgoode-onghjob,itwou1dhm mycnancssofstayinghsrs. 1:2:3:4:5 52. Ifthispsrsenwantsdtc,hs/shsce:1dprobsblybsablstogstmsfirsd. 1:2:3:4:5 53. It'sinpcrtantformotoksepthispsrsennappyformylengtsm succsuhsrs. 1:2:3:4:5 54. finispsrsen'sopinienofmehasansffectennmssenrsmyjobis. 1:2:3:4:5 PARI'IV. Thisssctienasksabentlowoftenyenandthispsrsencontactensanotler. Pleassanswsr mfimofsadnoftlefiollwingqnestiens:answertl'efirstpartbycirelingensrsspenss,and tnasscedpsrtbywritinginamnbsrbstweenensandten. 55. a. Cbnparsdtothispsrsen,whieneeo£yougosstossstnaotlarmorsoften? yen/thispsrsenlthssamsaaomt(mcxma~8) b. Ibwnrhuors often? (mamnytimssmorspsrwesla s.9., Mcsasoftsn). tinasmorsoften. 56. a. Qmpsrsdtothispsrsen,whidnensofye:callsthsothsrmrsoften? yen/thispsrsen/thssmmmt(mamam) b. I-bwuudnmorsoftsnflubwmanytimmrspsrwssk? s.9.,tndicsasoften). tinnssuorsoften. 57. a. mwgrdtotnuspsrsen,whidneeofymcmtacufluoderto9stinfiomtimabent wank you/thispsrsen/thsseusmnt(mmwfi) b. Madman-soften? (I-bwmytiuasmorspsrwssk? s.9., twicsasoften). timssnorsoftsn. Part2. ‘nnissctimasksabexdooenuolsdoplysioalthingsortnamfomtimywmdm mumm‘workdens. Plssssussthsfollodimscslstoamwsrthsssqusstiens. 1-erflnmorstlansvsryensslss 4-seuulat1ssstlansvsryensslss 2-senowlnatnorstl'anmryensslss 5-mdn1sssthansvsryensslss 3-sbentthsseusassvsryenss1ss 58. Plsassidsntifythsmatsrialsgg isstlatyennnssdthsuostinordsrto yen-work dens. mlpsrsdtosvsryensfiye: ,touatsxteE—amspsrsen 9st s.9ctaccssstothen? 1:2:3:4:5 b.90ttl'nlnflrw1? 1:2:3:4:5 c.1nf1usnesdocisiensabentwlogctsttm‘i’. 1:2:3:4:5 269 59. Plsassidentifyths ,tcolsgr; thatyennssdggmostinordsrtogstyenr workdens. Whom ye: win ,towtatsxtsntcsntfifs—psrsen s.9staccssstothen? b.9stthemfioryen? c.1nf1uesdscisiensabentwhogststlnes? HHH MN” 3 4 3 4 3 4 GUI“ 60. Pleassidsntifythsssrvicssghsl frenothsr lstlatymnssdgngmostinordsrtogst yenrworkdens. mnparsdtosvsryens pum— todatsxtentcantfifi'psrsen s.9staccssstothen? 1:2':3:4:5 b.9stthentoryou? 1:2:3:4:5 c.influenosdscisienssbentwto9ststln? 1:2:3:4:5 61. Plssss identifythsaspsctsofths icalenvirensnt(s.9.,roemlightim,spscs,stc.) thatyounssdthsmostinordsrtogstyemmmtlavingfleubsdiffsrmtmightstop yenfrengsttin'q—yenrworkdens. Wmmryensslssyenworkwith,towtatsxtentcsnthis sen per a.9staccustothln? 1:2:3:4:5 b.9stthemfe'yen? 1:2:3:4:5 c.1nfluencodscisiauah3utwhcgstsflm? 1:2:3:4:5 62. Plssssidsntifythswaysinwmichyennssdmtopyfiorgoodsorssrvicssinordsrtogct yenrworkdens,ortnatnothsvingmensymightstopyenfremgsttingyen'workdens. Chlpsrsdto everyensslssyenworkwith,towiatsxtsntcanthispsrsen a.9staccssstcit? 1:2:3:4:5 b.9stitbryou? 1:2:3:4:5 c.1nflnnnesdscisiensabentwmo9stsit? 1:2:3:4:5 Whenproblemoccuratwork, sensinformatienisuswallynssdsdtcsclvsthemortomksdscisiens 63. dossdnispsrsenlavsthsexpertisstosolvoprobluthatcanstopyou fremyenrgsttingworkdens? H s N ss U u H ss UI 64. dossthispsrsenhslpyoutogstsxpsttisstosolvomajorp'oblemat work? 1:2:3:4:5 65. dossthispsrsenhavsaccssstoinformtienthatisnssdsdtosolvsuajor problematwu'k? 1:2:3:4:5 66. dcssthispsrsenpssseninfomatientnatisnssdsdtocopswithmajor problmatwor‘k? 1:2:3:4:5 67. canthispsrsengsthslpfremothsrpsoplswdnohavsthssxpsrtissto solvsmajorprobleuatwork? 1:2:3:4:5 68. dossthispsrsenfindentbsforslandabente‘obleutlatmyoensup stink? 1:2:3:4:5 69. canthispsrsene'sdictd'angsstlatarsgoingtomksproblems for youatwnrk? 1:2:3:4:5 70. isthispsrsenwillingtolstyenhnowbsforslandofdnanysstlatars qoingtomaksproblmforyenatwrrk? 1:2:3:4:5 71. isthispsrsenablstosnticipstse'oblubsforsthsyhappsn? 1:2:3:4:5 72. dossthispsrsenusmllyforswornyenofproblubsforsthsyensup? 1:2:3:4:5 MWWWMMMI APPENDIX E CO-VDRKER POWER I-KJLDER QUESTIOWAIRE 270 APPENDIX E: CD-VDRKER RIVER I-KIDER QUESTIGMIAIRE momentum mmmnmmmmmrmwpuemonmm. Plssssdninksbent yerrslstienshipvith ,udmsllqusstiensroferimtothispsrsen. mammyerrflitfindnipwith_this_menly. mrsspenssssrsenlstslyenfidentisln plssssmailthsswminthseelossdmlc—ps—sssoenuitisoeplstsd. Myenrsryeehfe-yenr tin! PART 1. Duogrsphic infomtien. 1. mulenghvsyoubsensI-bsrofthisorgsniatien? yous/ultra (pisssscitclsens) 2. What isyenrssx? nulls/fans (plssss circls ens) MIX. mmmmmmmwymmmwmmmm. PIssss—mstnabuowirqscalstomthsssensstieu,ud¢irc1sths earrsspedstothsbsst dsscriptienofluvclosslyyenwek. 1-netsts11 4-s9rsstdss1 2 - s littls 5 - oelplstsly 3-smodsrstsuonsnt 3. miclosslydoyensndthispsrsenwuktogsthsn 1:2:3:4:5 4. ”mama-workmlmunubyenbothmentbsnmaa enditfsrsntpsrtsofthssnthhngs?) 1:2:3:4:5 5.m-Ivs11