Ti-tESlS [3‘39 IlllmuliMimiIiiiiil‘l’limifliwill 3 1293 01810 3907 This is to certify that the dissertation entitled A Model of Constructive Conflict Outcomes in Exchange Relationships presented by James A. Eckert has been accepted towards fulfillment of the requirements for . . ' t' 8: Su 1 Ph D degree in gfiége gang elm—3%? 3’ Sim K: W Major professor Date74vwzz¢ (RS/ff? MSUiJ an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY Michigan State University PLACE IN REIURN Box to remove this checkout from your record. To AVOID FINES return on or before date due. MAY BE RECALLED with earlier due date if requested. DATE DUE DATE DUE DATE DUE 1/98 animus-p.14 0“ §.. "‘W—. I - 'r‘ H‘ 'i (I A MODEL OF CONSTRUCTIVE CONFLICT OUTCOMES IN EXCHANGE RELATIONSHIPS BY James A. Eckert A DISSERTATION Submitted to Michigan State University In partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Marketing and Supply Chain Management 1999 Au“. v - gov-”u p .. 4 .- _ D —v-§ Q. 'H- .... .n-vao- d- b..- D ‘ . ~~. _ 0 as.- -‘b (i! ‘J‘ ‘ ~.. ~ ~ ...2 *3: ““ i . ‘V--~ n N a ' . ~f~"‘v-- ~- — a u u- . bu.-.. . 'I‘ C C. ‘ .. v- ‘ h :“n ' ..“ “4-... ""~~;.. ‘H ‘. o ‘9 sh -F‘ ~ . - ‘-¢- an‘P‘ a...‘ .- ~§ Vv“- ‘».~" - - s. p. .. - n ~“ ‘\~.‘ \ ‘5 ‘ .- .\ -‘ 7“ v. - ~ 5‘ § s - ' s "‘ ‘.. .- A ‘u ‘0 ‘ c: _ v- ‘v A v. v- .‘s. s by . - -. ‘- ... .‘ . .~ Q ‘~‘ '~ 7-“ .n - ‘ ~‘C~N _‘ u . .- .' .‘ ~._‘ . u y. 5:; u ~‘ ‘ 5 ~— R. \ v— flit u- ABSTRACT A MODEL OF CONSTRUCTIVE CONFLICT OUTCOMES IN EXCHANGE RELATIONSHIPS BY James A. Eckert The intent of this dissertation was to create and test a model of constructive conflict outcomes. A model was created based on the tenets of Social Conflict Theory and theories related to marketing channels. Data was collected from both marketers and purchasers who managed exchange relationships on behalf of their firm. The relationship was the unit of measurement and the responses were based on the perceptions of the relationship managers. Analysis of the data using structural equation modeling showed highly mixed support for both the proposed measurement and structural models. However, a useful set of conclusions did emerge. The central finding of this dissertation is that there does exist a concept classified as constructive conflict outcomes based on the perceptions of the respondents. The cmmcept gained.adequate measurement support and performed suffirxiently within the overall structural model. There werezcather important conclusions. First, there was not strcum; support for the adequacy of Social Conflict Theory in the context of business-to—business exchange relationships. The next important conclusion is that the relationship structure framework proposed by Robicheaux and Coleman (1994), and operationalized in this dissertation, proved inadequate for capturing the underlying condition of the exchange relationships that formed the basis of the data collected. The final major conclusion is that a limited model of constructive conflict outcomes in exchange relationships was supported. Specifically, the research found that low conflict frequency and an integrated problem solving approach had strong positive effects on constructive conflict outcomes. It was also found that constructive conflict outcomes had a strong positive effect on relationship satisfaction and trust. In the end, the dissertation, with its highly mixed support for the original model, provides starting point and a great deal of direction that should allow future researchers to create an improved theoretical model describing the existencetgf constructive conflict outcomes in business-to— business exchange relationships. Copyright by JAMES ARTHUR ECKERT 1999 0" U‘ To Paula, Fergi, Trixie & Merit. Unconditional love in the face of a dissertation is a wonderful thing! ACKNOWLEDGEMENTS First, I have to thank my wife. Her level of patience has been unfathomable. In addition, this task was made much easier because I have a set of family and friends who never doubted me and were always there to help when needed. So, Mom, Dad, Mrs. L, Ann Marie, Mary Beth, Tom D., Monica, George, Ann, Frank, Sheryl, Patty, Jef, Tom G., Jeff, Dave, Duffy, Julie, Kay (and “the girls”) and anyone I am forgetting: THANKS. On the business side of this project, I want to thank my Committee for helping me do this research the right way. Specifically... To Lloyd, thanks for your time, your involvement, your enthusiasm and your friendship. To Dr. Calantone, thanks for pushing me into a corner and thus keeping my butt safe! To Dr. Page, thanks for your much- needed attention to detail. And to Rob, thanks for your easy going demeanor and insightful direction. Lastly, thanks to the folks at Northeastern. Your patience and support was greatly appreciated! vi b! TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES INTRODUCTION CHAPTER ONE - EXECUTIVE SUMMARY OF THE DISSERTATION 1. Introduction 1.1 The research Objective 1.2 The Organization of the Dissertation The Dissertation Model and Research Questions 2.1 Understanding Constructive Conflict 2.2 Creating Theory that Addresses Constructive Conflict Outcomes 2.3 A model of Constructive Conflict Outcomes in Exchange Relationships Executive Summary of Literature Review 3.1 Social Conflict Theory and Constructive Conflict Outcomes 3.2 Relationship Structure 3.3 Relationship Outcomes Summary of Research Design 4.1 Design Issues 4.2 Measurement 4.3 Relationship Assessment Basic Results and Implications 5.1 Theoretical Results 5.2 Managerial Results CHAPTER TWO - LITERATURE REVIEW 1. 2. Introduction Background Information 2.1 The Pit of this Dissertation into the State of Channel Theory 2.2 The Relationship Paradigm 2.3 Robicheaux and Coleman’s view of Channel Structure 2.4 The Conflict Literature in Marketing Social Conflict Theory & Constructive Conflict 3.1 Social Conflict Theory (SCT) 3.2 Constructive Conflict Outcomes wi xi xiii \OCDONUJNN 14 20 2O 21 22 24 24 24 25 25 26 27 28 28 28 29 33 4O 49 56 56 58 .‘F0 n ...-—. am no - cont: 0 I -' ‘- 1 I; V“, In cub oohUb. 's “A... nfih .8. O bytoo..- - Appnby..- ' \ Vvoan. ‘. \ Ric.-' UHsQU.--.. a- '. A‘rqp'. - § Eva.-.‘ ‘- .4 P"“r:>.. ," vb:.u_.. D A Avauv , C-..D,. ‘- l I A_ __ . 3" L.:‘.r ha... a “A _ .- '-~~;V'\n .: o~u~-‘.... ”.- K . . -§- . ~'. ~Y" u: §.~_-~~ I q “ . a In“ . .Ih .‘.. :- u ‘ A I. ’: V'~fi. ' .v..v' d...- n n a Q i . . "‘ “5...--. n I A A? g u.‘ ~~~V:p. s "“_I 2 a A ‘ h . ‘rg‘ -. r. J »"'a.. ‘ ‘n R.. 'l ~~ -‘. “he. ”Q ' Q . I. \ V“~’"Yq‘ U..‘~ " A y‘- ‘ I f‘ - ‘.‘ an"§ 5".“ g A I A yl " ‘ v.“-. IA ' - n -, A "‘ » F; ~ v o.‘-‘ “A“:Q_ s \uv._‘-‘ '\ ‘ f‘ ..5 -,;. 5 1 v..‘-. NA’:. 5 . v..“ I 2., 'fl':. Var ““ F‘;. 'v o.“ '- ‘ u 3, . ' b-n"-‘~ UV“ \ ‘V A ‘ v»..‘ I Q “ I- - W‘ v - C~~v; ~ "“ H 4-. '~. ‘v..: 1" _ 1,, A_ v 'A. v..; ~ - C -v“ s ‘..~‘ ~ I . . A.” A .- ‘ $ Q; a... - ‘ Q | ”u“ ‘ \ ~A___‘ - ~ . " c.._ . V‘n‘, - n "i ' * u u .- .'s "_- ~ ~ A ‘¢-:\ Q ‘ H‘ .‘ Q . 4 ‘v . V- H N s‘ r A b ‘ Q 3' A . .‘-‘ c. A ‘w s TABLE OF CONTENTS (continued) CHAPTER TWO - continued 4. The Model 4.1 Conflict Characteristics and Constructive Conflict Outcomes 4.2 Decision—Making Structure and Constructive Conflict Outcomes 4.3 Operational Integration and Constructive Conflict Outcomes 4.4 Constructive Conflict Outcomes and Satisfaction and Trust CHAPTER THREE - RESEARCH EXECUTION Introduction Research Design 1. 2. 2.1 2.2 Pretest Design Issues Full Study Design Issues Construct Definition & Measurement 3.1 .2 3 3 3. 3 Clan—Like Structure Operational Integration Constructive Conflict Outcomes Conflict Relationship Satisfaction S ructural Hypotheses 3 4 .5 t 4.1 4.2 4.7 4.8 Conflict Frequency and Constructive Conflict Outcomes Conflict Intensity and Constructive Conflict Outcomes Conflict Importance and Constructive Conflict Outcomes Conflict Focus and Constructive Conflict Outcomes Decision-Making Structure and Constructive Conflict Outcomes Operational Integration and Constructive Conflict Outcomes Constructive Conflict Outcomes and Relationship Satisfaction Constructive Conflict Outcomes and Relationship Trust Statistical Analysis 5.1 Measurement Model 5.2 Structural Model Conclusions viii 67 67 71 74 75 77 77 79 79 82 86 87 100 108 112 117 118 119 120 121 122 123 124 125 126 127 127 132 134 TABLE OF CONTENTS (continued) CHAPTER FOUR - RESEARCH PLAN EXECUTION 1. Introduction 2. The Pretest 2.1 Pretest Sample & Data 2.2 Pretest Data Analysis 2.3 Pretest Results & Implications 3. Data Collection Results 3.1 Data Collection 3.2 Response Rate 4. The Sample Description 4.1 NAPM vs. NAMA 4.2 Industries Represented 5. Sample Conclusions CHAPTER FIVE - RESULTS OF THE STATISTICAL ANALYSIS 1. Introduction 2. Details About the Data 2.1 The Distributions of the Variables 3. Testing the Measurement Models 3.1 Relationship Structure 3.2 Conflict Characteristics 3.3 Constructive Conflict Outcomes (CCO) 3.4 The Full-Model CFA 3.5 The Repecified Measurement Model 4. Testing the Structural Relationship 4.1 The Path Model 4.2 The Path Model Analysis Results 4.3 The Support for the Hypotheses 5. Statistical Analysis Conclusion CHAPTER 31:: - DISSERTATION CONCLUSIONS 1. Introduction 2. Measurement Conclusions 2.1 Relationship Structure Conclusions 2.2 Integrative Problem Solving Approach 2.3 CCO Conclusions 2.4 Conflict Characteristic Conclusions 3. Structural Conclusions 3.1 The Overall Effect of Relationship Structure 3.2 Decentralization and Constructive Conflict Outcomes 3.3 Participation and Constructive Conflict Outcomes 135 135 135 136 137 141 144 144 146 150 150 160 162 164 164 165 166 167 170 180 184 186 199 218 218 223 226 229 230 230 230 231 233 235 238 240 241 243 244 0‘ no -! m SIX - conti . A‘ . ~.ra~..,r; " :U..V".‘. 5v v I . >- t ' r>;-r: ” a“! ......_-..a.\. I" - a. ....., F\ v Event-u. in. v -' .An... Wr‘ ‘. ~\ H ‘ "v .yuay .,. '9 “an. ...’ Ubn...-_. ru 1; a-.. ... "' Vis.---~. .I “AF. '5‘ y....-.~_ , 'l. ‘ T ”T'. ‘Av- ca. Ewan..-.' .. A-V‘. .~‘ \uvo-..-_- ‘ D - ‘an'. ‘ '~ ‘ do onv-‘--v.' “TA...“ \ iv..- -- ‘. vr- - ’ 4- ‘ a‘LY': C! LNG-‘:~.-‘- ‘ " A -, 1 ‘F. .a‘ I. yv...--~‘ I“ ‘ I n“ v- - .uh oh. -.‘_~ ~r~-.fl ‘ on... b‘. Us"-‘- I\ A . “s- 3' 'v_‘ A." ' v--‘-'- ‘A “V‘- d ‘ b-l— . H.-u . .. a ‘w.‘ “ - rm..- “u.“'-vcn; . I I v -Q at. h”: V; ..~‘~‘-‘ ' o h “A' ‘ ‘.~ - -'. O‘V.‘ ~.. u C ’ A ‘ ( “A. "" ‘ ' r. o-u--“.. D .l u.‘ A N i.‘ _ v o..‘ ~‘ . . \ --.- A» v. . ~ \.. ..b s... , I " : "v-A -- "“= n: I.‘-~ ’ a n ‘- U.. - “ ‘u Q --.~ ' A - ‘v- . u.‘ ~ A- ‘~. -\‘ sp‘ ' H u t v.‘ - ~ 'I U " ‘° ~: ‘ , h I ‘ . "‘ ‘va b. A.‘ E O f‘ - n'.. I 'V..‘- K . ‘-_~~ ~'§.- :‘n' ‘u y ‘ \ .‘ ‘Z a H .3 ‘. I N r ) TABLE OF CONTENTS (continued) CHAPTER 811 - continued 3. Structural Conclusions — continued 3.4 Integrated Problem Solving Approach and Constructive Conflict Outcomes 3.5 Issue Importance and Constructive Conflict Outcomes 3.6 Conflict Frequency and Constructive Conflict Outcomes 3.6 Conflict Intensity and Constructive Conflict Outcomes 3.7 Relationship Satisfaction / Trust and Constructive Conflict Outcomes 4. Managerial Conclusions 4.1 Conflict Can be Constructive 4.2 An Integrated Approach Leads to Greater Constructiveness 4.3 Greater Constructiveness is connected to Greater Satisfaction 5. Limitations 5.1 Measurement of Conflict 5.2 Non—Dyadic Approach 5.3 Reliance on Perceptions 5.4 The Cross-Sectional Methodology 5.5 The Skewed Level of Relationship Quality 6. Future Research 6.1 An Interaction Effect? 6.2 An Issue Group Effect? 6.3 Is Relationship Structure a Moderator? 6.4 Are the Results the same at the Conflict Episode Level 7. Conclusions APPENDICES Appendix Al Construct Conceptualizations and the Items Used in Questionnaire Appendix A2 Construct Conceptualizations and Final Items Used in the Analysis Appendix B The Pretest Questionnaire Appendix C The Full-Survey Questionnaire FUHTRENCES 245 246 248 250 253 254 254 255 256 257 257 258 259 261 262 264 265 265 266 266 267 270 271 291 306 308 317 Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table .10 LIST OF TABLES Pretest EFA Results Common Respondent Position Titles Weekly Response Rate Tally Respondent Demographics Respondent Firm Demographics Relationship Demographics Industries Represented Decision-Making Structure Item Reliabilities Decision-Making Structure CFA Results Operational Integration Item Reliabilities Operational Integration CFA Results Conflict Characteristic Issue Groupings The Conflict Characteristic Construct Indicators CCO Item Reliabilities Item Reliabilities Goodness-of—Fit Measures for Initial Measurement Model Convergent Validity Support for the Constructs within the Initial Measurement Model xi 140 145 147 152 153 155 161 171 173 175 178 182 184 185 190 192 196 IO. Cl -' ' IF -1 In. - - “5...... .. I ' I. . A . "a. A ._.._... - C O. I...O .’ ~ a -l .h.-"., ‘ C I O ‘ a- . ~ ’ o... ‘ . ".' a; o _. 'fi.‘ 0 11-. '- ‘§ ‘ . ._.- ‘- s "-2 - 'l‘v ". :- . - ,_ -‘..-‘ o- “ v..‘ n -. § v'~' Table Table Table Table Table Table Table Table Table Table Table .11 .12 .13 .14 .15 .16 .17 .18 .19 .20 .21 LIST OF TABLES (continued) The CCINT and SHARPAR Items The IPSA Construct Items Item Reliabilities for Respecified Measurement Model Goodness-of—Fit Measures for Respecified Measurement Model Convergent Validity Support for the Constructs within the Respecified Measurement Model Summary of the Measurement Results For the Supported Constructs The Paths and the Hypothesis Each Represents The Input Matrix for the Path Model Path Model Goodness—of Fit Measures Path Test Results The Results of the Hypothesis Tests X“ 205 206 210 212 215 217 221 222 224 227 228 vy- ~.-.. . ‘0... Wu" ‘Lfi...*_ L» U Figure Figure Figure Figure Figure Figure Figure LIST OF FIGURES The Original Structural Model The Decision—Making Structure Measurement Model The Operational Integration Measurement Model The Constructive Conflict Outcomes Measurement Model The All—Variable Measurement Model The Respecified Measurement Model The Full—Variable Path Model xm 16 17 18 19 188 201 220 n~ .uga. .... \ eu... c... . INTRODUCTION Firms are creating more and more close relationships with customers and suppliers, and many firms believe that these relationships are key to the firm’s success. Within these close relationships, conflict is inevitable. However, conflict does not necessarily have to be a negative experience, even though that is the connotation usually assigned to it. In fact, the tenets of Social Conflict Theory contend that conflict is a required element for tnoductive and healthy social relationships. This (fissertation is designed to test the adequacy of applying tjmt concept to business-to-business exchange relationships. Ittheoretical model was created using the tenets of Social (kmflict Theory and marketing channel theories. This model “as tested and results reviewed. These results will show tint the model proposed was not fully supported but still unwributes theoretically towards the goal of building an adequate model that describes constructive conflict outcomes hikmsiness-to-business exchange relationships. antthe beginning this research was considered exploratory :htnature because the concept of constructive conflict (the flxwl variable) did not come to this research with a strong fluxuetical background. Therefore, the results, including thermmrsupported hypotheses, provide an excellent source of inflnmwtion that should help future researchers clarify the thanxmical importance of constructive conflict and help to butkia.suitable model that positions this concept amongst lasrelevant antecedents and consequences. And to that end, the research was very successful. Executive .—- INTRODUCTION 0.. .~> g ‘V a... . S .. via an». . -..~. - ' o ' ' 1.. a“... ‘R fir ‘- ‘ ‘ fl~ .. .‘O . Vi... “H b..- i ' . It. v-q-'. .' -“ : ,... - ~ ‘\ o.»\po..-.l “it ‘ o Qua..-’ F C. _ ~_ 95.. you». _..‘ v - ' q HV‘LV ‘5‘- s4 -‘ . - ‘ g' ‘ I.‘ A ..I ~~~ '- a is»... .-.: ‘ ~ C :‘v._ N .5. .' ...'_ v .- ». v‘ - U..-‘l‘ ~‘ 0. . -."-.:; . I A “"~~- can“, ~‘-.. 0‘: V‘-~ D..| I. I ‘ N. v" ' v- "Nn 1:. .. ~~ “huge-w ‘ v-‘ \ ‘n a . .. .- “ ’a . ‘ I. "'~‘c Ra~ ~T u“\"v.. ‘~ A‘ ‘s “ - I I.‘-‘. f.;\ l- A "~.‘; r... ‘ . ‘ts ‘- n a ‘ -0 .. : '5‘! Q Q “E ha‘:. . l'.“ Q fi‘ ‘ ‘ ‘b Q C ‘ I y- ‘ C‘Cn-I fi.‘ n.- § I ‘s ‘- ~ ‘\'I, '. . . ~‘. A" 'n “.~‘~l‘ b.‘ 5‘ ‘ A . i. ‘k \ s. ' ‘ 2 v. "s s '1 C. " .¥ -- & o ~-.." ,, ‘H Chapter 1 Executive Summary of the Dissertation 1. INTRODUCTION Conflict is inevitable between closely interdependent units: units such as the firms that make up channel relationships. And recently, as firms have tried to establish even closer ties with exchange partners, the potential for conflict has grown even larger. How many married couples never fight? Few indeed. The same can be said for the organizations that have chosen to arrange their buying and selling relationships in a way that mimics the closeness of marriage. Unfortunately, this inevitable conflict can lead to inefficiencies and dissatisfaction. It can break down the bond between these organizations. Are these destructive outcomes of conflicts the only possible result? Returning to the marriage analogy, isn’t a little conflict healthy? If spouses don’t “let it out” occasionally aren’t they candidates for a plate-throwing melee? Haven’t many spouses arrived at productive changes via a route that began with a fight over the issue? an ' ... . l ,yuaoo ' eyfi‘f . ..do-.. ,. .o. - n 1.0-. on... \ V to... .. Conflict in this context is constructive for the relationship. If close exchange relationships between organizations are indeed similar to close interpersonal relationships then shouldn’t conflict have these same constructive outcomes? 1.1 The Research Objective The introduction highlights the general question that this dissertation seeks to answer. Stated more formally, the fundamental research objective of this dissertation is: To gain an understanding of the concept of constructive conflict outcomes and to build and test a model that represents a group of the key antecedents & consequences of this phenomenon. By achieving this objective, this dissertation will begin the theory-building process that eventually seeks to arm relationship managers with a useful model that predicts when conflict will be constructive and what impact such outcomes will have on the overall exchange relationship. Relationship managers are those individuals who have primary responsibility for the creation and maintenance of their (II 7' 0‘ . a “A .. '- ->. Ive-.- ftnn’s exchange relationships. Examples would include a purchasing manager who is responsible for procuring needed raw materials from an outside supplier, or an account executive charged with maintaining the customer relationships for her firm. By providing individuals such as these with that theory, these relationship managers would be able to implement the appropriate actions that would lead to constructive conflict outcomes. The aim of this dissertation was to make an exploratory attempt at building such a model. In the end, the results did not provide a complete picture of the phenomena. However, the research did accomplish two key things; 1) it establishes a starting point for future research that can be designed to clarify the model, and 2) it provides a starting point for managers who wish to actively manage the conflict in their exchange relationships. Both results have value. Since conflict is a fundamental part of all close, interdependent relationships, if we, as scientists, are to better construct theory that explains and predicts the relationship building and maintenance process, then we must have a robust understanding of the potential beneficial outcomes of conflict. Without such an understanding, our theories will lack the ability to explain the richness of how conflict impacts relationships. Thus, - an. In. _ ., w. ,_ ~- ~o.._ v. ._ . a o. .‘ ..¢.~ ‘ A "H. . .‘DO 1.... . u__‘. ‘ .._‘~ ". (A the results represent theoretical advancement. In addition, the current state of business finds many firms engaged in close business-to-business relationships that are bound to be face conflict. This means that there is a current managerial need for guidance on how to best handle the conflict to achieve the most constructive outcome possible. Therefore, the basic model of the antecedents of constructive conflict outcomes that was supported provides the foundation for pertinent managerial prescriptions in this area. The concept of constructive conflict outcomes is not new to the marketing literature. Assael (1969) explored this concept in automotive distribution channels almost thirty years ago. However, while many authors since have proposed that conflict could have constructive outcomes (e g. Dwyer, Schurr and Oh 1987; Morgan and Hunt 1994; Weitz and Jap 1995), none have attempted to establish a full conceptualization of this idea. Additionally, there exist no empirical findings that position this construct in a theoretical model. Therefore, it was the intention of this dissertation to tackle both of these tasks. Social conflict theory provides the basis for a full conceptualization of this notion and also provided .qub-l-ns as t: . a h,...‘vuoult o i...— ~;'- ;e5: P“ - ’0 M..- n-‘AnAQ knac- s .a-qu- e‘kvnp:v-w r u - than hIIU‘IO: .5 .. a AOIAQ. A. .V'"“."V' "' . nonvo- v. ucd~vuln e ““~D-R '3‘ .p‘u- a "‘ ASS c" we'vvn-yu . Q A A-.. A - ~— v-H .4 .v~.~nua ..u -v- u - . , )r‘ 'y her-n b-‘p~.‘ .‘~-o"- U...» g-.. ‘ r “I... - -"a w- -. ""°"¥ Hue “.v-.. ' . vu. - :’~.Q : . '- - ?~--...-. .. ' .v- - e . «- Pfin. An ’ twin. v. o...‘ \ V‘;-6-. . . r. ‘.‘ ‘ :- .AV‘Q’. V‘ .. .-.1.-v_‘:..‘d C 5A-. e ‘3‘,- . ...h . "Q s..: an";‘- .".“"‘OI 1.2 The Org - u a '« ~‘A~ iv,“ ‘2“ . Q ""-.- A \r‘ -u 4" u. -‘ ix: . guy a ‘A. . _ 4 f."‘- ‘ I . I t“~. * a“ t -:-' _:,~ ‘ "y A . ‘L:\A ‘ 5 ea ”.2 s h ‘ ~ ~. V‘. . ‘ - s..“ :‘n; . “‘\A ‘p‘ . v— 5..‘- - O s h . ‘ ‘a. '- ‘- i . .v. . . . - . \‘ .v‘A -‘ 5 s “~ Q, I 5“! F‘ ~‘ - e‘ P ‘ ~ H.- ~ “ N .- . n,:~sh ~b A § s v- 4._Q 5. ‘ .‘1 a b, . p. \ \‘. ¥ ~ \~:'~u \ ”u "mfl ‘4 Aux .‘\~s ‘ v ‘ F’ \ a U predictions as to when constructive outcomes will result froulnanifest conflict. Additionally, since relationships between exchange partners can take on many structures, the effect of structure on this concept of constructive conflict outcomes was explored. The structure framework proposed by Robicheaux and Coleman (1994) guided this exploration. Lastly, since firms are entering into relationships to achieve business objectives that are unreachable without the competencies of the firm’s exchange partners, an assessment of how constructive conflict outcomes effect overall relationship outcomes (such as satisfaction and trust) was undertaken. 1.2 The Organization of the Dissertation The document is organized to walk the reader through the relevant stages of the research. First, the theoretical argument as to why the proposed relationships were hypothesized is reviewed. Next the research plan is outlined and then the results of the research are considered. In the paragraphs below, the specific chapters are briefly highlighted. Chapter One. As this chapter continues, the specific research questions, and the model that represents the is» r5» .u.. ‘V proposed answers to those questions, will be presented. An executive summary of the theory behind the model, the proposed research design, and the basic results will follow. Chapter Two. The literature that provided the background and direct theoretical support for the relationships proposed in the model will be fully reviewed. A theoretical argument will be built. Chapter two concludes with a review of the core logic behind the theoretical argument made. Chapter Three. The plan used to execute the research is cmtlined in this chapter. Special attention is paid to the neasurement of each of the constructs presented in the nwdel. Additionally, the sampling plan and the intended statistical evaluations will be reviewed. (Impter Four. This chapter reviews how the data was mfllected and the basic condition of that data. (kmsideration is given to the adequacy of the data for the theory test proposed. (Rapter.Five. This is the chapter that reports the sundstical analysis of the data. It first reviews the rmmsurement tests and then the structural tests. . ,...v»w . n— ...a- v v‘ a l-leD. nu...- .._ . u D '9. \ \ ‘s '5 Chapter Six. The final chapter considers the implications of the research. The theoretical, as well as the managerial conclusions are reviewed. In addition, the limitations and directions for future research are documented. When the reader has concluded reading this dissertation they should be left with a richer understanding of the concept of constructive conflict outcomes and its existence within the context of exchange relationships. 2. THE DISSERTATION MODEL AND RESEARCH QUESTIONS There are two sets of core research questions that this dissertation sought to answer. The first focused on providing a rich understanding of the concept of constructive conflict. The second set of research questions dealt with seeking to understand constructive conflict’s relationship with a set of theoretically and managerially Significant constructs. Through the process of addressing the specific research questions, this dissertation was able to achieve its overall research objective. In addition to these core questions, there was an additional QUEStion for which an answer was sought. Specifically, the dlsSertation provided a measurement test for a set of -. . .ro-au brgv we ‘ . ....~ ‘vh iv v - . ;’~Arcr .MvovOoua... - pa 5 . - H: I: .-~.- I .vUi .0! ho- ‘ P Q - npnA-o.ooe In" 9”. ,. up . u. vv-o' v. baa- . 2.1 Understand . ....,.....,” fl- ;. ----~-..._ 8 ,. . _ .. ¥ ~"‘--_ . |. V9. i n.” o- his,“ ..u‘~¢c ¥v. :‘Y"f‘o- ’ a-..» .. n ‘ a. a I-’ - ~A' ..- - cgoo...‘:. an.- a .- -,- - . . ~~ . .. ‘na F '-.. a... - . n a.._ . u v. a I -b.. I. :p,‘ d A-.. “‘H ~‘v- & .— s‘. . .._ h ".5 ' _..Q '5. w “A - c '9' c. c..- 0» Cl..c- a- a x _. r. ‘x. I. ‘ I- “‘ Q. p. constructs that were proposed to represent the structure of channel relationships. These constructs had yet to face such a test in the published literature. Thus, a secondary objective of this dissertation was to provide that test. 2.1 Understanding Constructive Conflict To create and test theory in this area, the notion of Constructive Conflict Outcomes had to be captured as a vmrkable construct. Theory starts with constructs (Kerlinger 1986). Thus, before the concept of Constructive Conflict Outcomes was related to other concepts it had to be beined and conceptualized (Hunt 1991). Therefore, the first set of research questions outlined below, deal with the conceptualization of constructive conflict outcomes. Specifically the questions are: (1) What constitutes Constructive Conflict Outcomes? (la) What does theory provide in the way of a conceptualization of constructive conflict outcomes? (lb) Do the perceptions of relationship managers support this conceptualization? '- :v .-a n.. a 0.4. o -.. -.. .- -. a n. '0 \ 0- Combined, the answers to these questions approached the concept of Constructive Conflict Outcomes from a deductive and inductive approach. First, the dissertation reviewed Social Conflict Theory, which provided a rich conceptualization of the notion of Constructive Conflict Outcomes. Based on this, the concept was operationalized and the subsequent empirical test of that operationalization provided an answer to question lb. Thus, the first question is a logical question and the second an empirical one. The combined answer provides a richer understanding of the concept of Constructive Conflict Outcomes. 2.2 Creating Theory that Addresses Constructive Conflict Outcomes There were two objectives related to building theory concerning Constructive Conflict Outcomes in exchange relationships. First, there was the primary objective of building a model of the relevant constructs that included the order and structure of the antecedents and consequences. This, in effect, would be considered a fully specified theoretical model. The secondary objective was to, at minimum, identify and verify the set of constructs that make up the nomological net surrounding Constructive Conflict Ontcomes. These two objectives are addressed in reverse Order below . 10 . t a e I I. -ner OIA"~" 'u u..- -.oe‘ area .‘b. n u cyn n..-¢b ’Aw- . I ~ ~ ‘IVQU"OIU IONI-UQb . .. ‘ I‘i-OII-a: ‘S °-pa~-‘ '."‘" '5 . \p. ‘ w . . . . . ._ a '. _ ‘ I:p-p .,,_ .v.¢.. d.a~o.-h 5.. O a... ‘ . en: ‘~b ..'~ o."_ . v... vsu:.' “ ‘ ‘v.~ ‘ " . .v. : Vs R; U“ b v... ,_ ‘ ‘ 5 . ‘e. . . § .I‘ ‘ \."_‘.§ .z" ‘ ~ ‘ Vu, fl" '1 ‘ a...“ - . " “ s ‘e P n. Y . ‘HE ‘Q:.. ‘Q ‘~~u . I 3““: “" :3 C H“ c. ' ~~ ‘ b.‘ v‘ 5. Va ‘h~-:V~.. U.‘. ‘— ‘ Ac. n 'V‘ h a “ ~-.. 3” \“v.““ . . ”i ‘ 'e t '1‘ ‘Av _ v““ ‘ ‘ § The Ndmological Net for Constructive 2.2.1 Conflict Outccmes ENen though a full theoretical model was not supported by at a minimum, did identify this dissertation, the research, a relevant nomological net in which Constructive Conflict Outcomes is embedded. A nomological net is the web of relationships that positions a construct in the universe of related constructs (Kerlinger 1986). The nomological net includes the construct’s antecedents and consequences, as However, the nomological net does a full theoretical model. does not position individual constructs in any type of Casual order, while a full-theoretical model does attempt to dO that. This research used a cross-sectional survey method (Versus a longitudinal or experimental design), and thus, from a purist viewpoint, no temporal precedence was securely identified, and therefore, no causal order can be inferred fITnn the results. Thus, the concept of a nomological net IIrovides a legitimate context for the results of this reSearch. Furthermore, since there was little conceptual or en113irical work in this area prior to this dissertation, the 1dentification and support for constructive conflict's noInological net still constitutes a significant contribution tilert can form the foundation for building a subsequent full theoretical model . ll .,.-. no -.'4. OA.‘ ~...‘ u.-. u... 2.1.2 The Full Theoretical model of Constructive Conflict Outcomes As stated, the primary objective of this research was to create and test a fully specified theoretical model involving Constructive Conflict Outcomes. Therefore, the literature reviewed and the logic applied was with the objective of creating such a full model that included the proper order and specific structure of the relationships between the relevant constructs. (This goes beyond a nomological net, as such a net only identifies the relationships without necessarily predicting the order or Structure of those relationships). The specific constructs that comprise the model tested in this dissertation include the various characteristics of perceived conflict, the Variables that represent the structure of the exchange relationship, and two constructs that represent overall relationship outcomes. There are other constructs that C3Ould have been expected to comprise a fully specified set ‘Df relationships involving Constructive Conflict Outcomes (€3.g. Dependence and Commitment). However, since the (jiISSertation represented one of the first research projects trhat focused on this concept, the number of relationships ex:IDlored was reduced to those that were most directly relevant to an initial understanding of this concept. The C3Onstructs were chosen for the following reasons; (1) the 12 n...- ‘I . 5‘... o... -... ,. '“v- . u I... '- ~.. characteristics of perceived conflict because they are suggested to be related to constructive conflict outcomes by Social Conflict Theory, (2) relationship structure because this it allowed for the diversity of real relationships to be systematically considered, and (3) relationship satisfaction and relationship trust because they are theoretically and managerially interesting outcome variables. The second set of research questions (presented on the flfllowing page) provided the guide to building and testing the model of Constructive Conflict Outcomes in exchange relationships. 13 (2) What relationships comprise an appropriate model of Constructive Conflict Outcomes in exchange relationships? (Secondarily, what variables constitute constructive conflict’s nomological net?) Specifically: (2a.) What is the effect that the characteristics of perceived conflict have on Constructive Conflict Outcomes? (2b.) What effect does relationship structure have on Constructive Conflict Outcomes? (2c.) What effect do Constructive Conflict Outcomes have on relationship satisfaction and trust? 14 2.3 A Model of Constructive Conflict Outcomes in Exchange Relationships To answer the questions proposed above, a model was conceptualized that predicted answers to those research questions. It is this model that was explicitly tested during the execution of this research. The model is illustrated in Figure 1.1. Figure 1.1 represents the structural relationships that were examined. However, measurement issues were also highly relevant. In addition to the structural model, both first, and second-order measurement models were proposed and tested. Three of the model’s constructs were operationalized as being comprised of multiple dimensions with each dimension being indicated by multiple measures. These complex measurement models are presented in Figures 1-2. 1.3, and 1.4. In addition, the measurement properties of the other constructs (the conflict characteristics, satisfaction and trust) will be considered, although no model is presented. Chapter three will outline the measurement assessments that Will be made for the collective group of CODSthCtS presented in this dissertation. 15 Conflict Frequency Decision- Making Structure Operational Integration Conflict Conflict Conflict Intensity Issue Focus Importance - + +‘ Satisfaction Constructive + Conflict +' Outcomes Trust ————>' Figure 1.1 - The Original Structural Model 16 Decision-Making Structure Decentralization Informalization Shared Participation Paradigm Figure 1.2 - The Decision-Making Structure Measurement Model 17 Operational Integration R Joint Actions PM ll H I” DEE] Figure 1.3 - The Operational Integration Measurement Model 18 Constructive Conflict Outcomes Activity Adjustment Norm Adjustment Hostilities Release HUD Figure 1.4 - The Constructive Conflict Outcomes Measurement Model 19 3. RECENT .u-r-p a P F “n: .o»v-d-.‘ . 1*‘h-rhqb. -, \\D N “thus. 3“.-. - ~o . "F" ‘Wn-u al) (I) f 3. EXECUTIVE SUMMARY OF LITERATURE REVIEW The literature that provides relevant background and specific support for the relationships proposed in this dissertation is fully reviewed in chapter two. However, the highlights of that review are presented below. 3.1 Social Conflict Theory and Constructive Conflict Outcomes Social conflict theory, based on the writings of Lewis Cosar (1956, 1967), provides strong theoretical support for the <33ncept of constructive conflict. Specifically, Cosar kMalieves that all healthy social systems require conflict so tfllat hostilities can be released before they build up to exaglosive levels. In addition, he believes that conflict JJBads to positive change. Exchange relationships embedded ir1 marketing channels represent systems and thus should also rENquire conflict if these relationships are to be healthy. SEMecifically, Social Conflict Theory predicts that when CCHlflict is frequent, of low intensity and importance, and Cloes not focus on central issues, the likelihood of Ccnlstructive outcomes is enhanced (Turner 1986). These ITalationships will be operationalized and empirically tested iII this dissertation. 20 7‘ . . ev.' «n" . e. ‘ v ~.. ~- ‘ 1 v.1 n. .fin—e... __._——~_l . __.--\ ._-,d_——~ 3.2 Relationship Structure Robicheaux and Coleman (1994) propose a model of relationship structure that is rooted in the political- economy paradigm (Stern and Reve 1980). (These models are discussed and illustrated more thoroughly in chapter two.) The Robicheaux and Coleman model carefully separates the antecedents of relationship structure from the relationship structure itself and the consequences of that relationship structure. This disciplined division is valuable because it identifies the core elements of the structure itself, thus allowing for a more judicious use of this concept in the study of exchange relationships. This is important because there exists a wide range of structures that characterize ébmchange relationships, and any theory proposed that is irfllended to apply to a full range of exchange relationships mufst take into consideration this diversity of structure. Thfia Robicheaux and Coleman (1994) framework provides this diEssertation with a theoretically sound way to SEWStematically consider such diversity, and thus, extend thfimuy related to the management of exchange relationships. R£flflcheaux and Coleman (1994) propose that relationship Structure has two components, (1) Decision-Making Structure and(2) Operational Integration. This dissertation measured 21 ...r u. .n I9. eon-v e a... ’P. ~... both of these components (via their underlying dimensions) and assessed the impact of structure on Constructive Conflict Outcomes. It was predicted that clan—like decision making structure (decentralized, informal, participative decisions based on shared norms) would be positively related to perceptions of constructive conflict. Additionally, it was predicted that high levels of operational integration (characterized by significant joint actions, information sharing, assistances and monitoring) would also be related to high levels of perceived constructive conflict. There was mixed support obtained for those contentions. A secondary objective of the dissertation was to assess the measurement qualities of Robicheaux and Coleman’s (1994) structure conceptualization. They only provided the rationale for the relationships, but did not directly test the implied measurement model themselves. This dissertation xfill attempt to provide the needed empirical support. 3 . 3 Relationship Outcomes One Of the keys to making the exploration of Constructive ConfliCt Outcomes interesting and relevant is to explore its COnnections to important outcome variables. Relationship Satisfaction is one such relevant variable. Brown, Lusch 22 and Smith (1991) provide a meta—analysis that concludes that conflict has a negative relationship with satisfaction. Yet, none of the research reviewed considered the potential mediating role of Constructive Conflict Outcomes. It was hypothesized that Constructive Conflict Outcomes would be positively associated with relationship satisfaction. In addition, since two characteristics of conflict were hypothesized to be positively associated with Constructive Conflict Outcomes, the model proposed in this dissertation hypothesized that conflict with certain characteristics would actually be positively associated with relationship satisfaction. There was mixed support for the hypotheses in this area. Thus, besides satisfaction taking the role of pertinent outcome variable, its empirical Instory combined with this dissertation’s hypotheses make its inclusion interesting in the manner highlighted by Davis (1971). 23 ol- 4. SUMMARY OF RESEARCH DESIGN Chapter three outlines the execution plan for this research. However a brief review of that plan is highlighted below. 4.1 Design Issues A cross-sectional survey design was employed. This design allowed for the relationships proposed in this dissertation to be put to a relevant and practical test. A pretest of certain measures was conducted to make up for the lack of an empirical measurement history on the part of some of the model's constructs. After appropriate measurement refinement, a “modified” Dillman (1978) approach was used to Gather approximately 220 responses from relationship managers who were fundamentally responsible for managing one 0ftflneir'firmfs key exchange relationships. 4 . 2 Measurement The reliability and validity of the measures employed in this Itasearch was assessed via procedures such as Cronbach's alpha, EPA and CPA. The goal of these assessments was to eStabJJLsh quality measurement of all of the constructs in the ITlOdel. Without quality measurement, the results related 24 n.4- ~0- 'a. I... 's to the structural relationships would become suspect. Between the pre-test and the full—model test, the proposed measurement models were adequately assessed and refined, thus allowing for confidence in the structural results. 4.3 Relationship Assessment Structural equation modeling (employed using EQS) provided the method by which the proposed structural relationships were tested. Model fit evaluations combined with the significance of the parameter estimates provided support (or non-support) for the relationships proposed. In addition to the overall assessment of the model, each hypothesis was represented by a structural path in the model and thus each relationship was assessed individually. 5.0 BASIC RESULTS AND IMPLICATIONS The results of the statistical evaluations are fully nadewed in Chapter Five, and the dissertation conclusions EH9 presented in Chapter Six. However, in the sections kelow, some of the key results and conclusions are reviewed. 25 --. 5.1 Theoretical Results The main theoretical contribution of this dissertation was to establish the concept of Constructive Conflict Outcomes as a significant construct in the study of relationship management. The results of this dissertation establish the initial nomological net for this construct and provide guidance for future research. Social Conflict Theory was also tested via this dissertation. The tenets of Social Conflict Theory, when applied to exchange relationships, were not supported. This lack of support casts doubt as to the efficacy of Social Conflict Theory as an important theoretical base from which further marketing thought could be developed. The model presented and tested also incorporated the framework presented by Robicheaux and Coleman (1994). This framework was modeled after the Political—Economy Paradigm (PEP) suggested by Stern and Reve (1980) and extended by AChrol, Reve and Stern (1983). The constructs suggested by RObicheaux and Coleman (1994) were operationalized and measured and the dissertation data suggested that this Configuration of constructs is inadequate for describing the Underlying condition of exchange relationships. 26 ~ on...l, ...¢ ‘ I. A . . -...,. n- n “ ....‘~u¢ o. obodvobllhel. ’- ‘."'.~. A.- p U - noanv Von ., .‘ O u... 4... . \ r—l\ ”my ex... ‘. - C'e. ... .- ~;V.. "' nu . ‘~. ‘~ .‘e- —' , l “u... a- --‘ N C; .. Lastly, through the use of some of the Robicheaux and Coleman (1994) framework (that which was supported), relationship structure was considered and shown to have an effect on the constructiveness of the conflict outcomes. This result provides a valuable starting point for future theory creation. 5.2 Managerial Results Since many of the hypotheses embedded within the dissertation model were not supported, combined with the fact that the model as a whole fit the data poorly, the managerial contributions of the dissertation are very limited. However, the basic conclusions that are valuable to managers are: (1) constructive conflict does exist, (2) an integrated problem solving approach creates greater levels of constructive conflict, and (3) these outcomes are strongly connected with other valuable relationship outcomes such as satisfaction and trust. 27 1. INTRODZ - I "- ”’Rvni ‘ v— .- u... ..__-‘ . . ‘cn- o. - r~ .- ‘ “ be..- a e -‘ I..‘ ~ ‘0 ‘ x u . ~-“' ~ ‘1 o \ ‘-~.‘ V e': ~’ \.‘ Chapter 2 Literature Review 1. INTRODUCTION This literature review will go through three basic steps: (1) a review of the relevant background related to this area of study (SECTION 2), (2) a review of Social Conflict Theory and the concept of constructive conflict (SECTION 3), and (3) a structured tour of the model where social conflict theory and marketing channel theory will be used to demonstrate the precedent and rationale for the model’s constructs and their relationships (SECTION 4). Literature ‘Will be included in this review to the extent that it Iprovides relevant background and/or it supports the logic and structure of the model presented. 2- BACKGROUND INFORMATION This section seeks to accomplish the following objectives; (1) position the dissertation in the stream of channel kil'lowledge that has evolved over the years, (2) review the empergence of the relationship paradigm that has come to the 28 ‘e “4 If, forefront in the past years, (3) show how the Robicheaux and Coleman (1994) framework (especially their conceptualization of structure) addresses some of the shortcomings in the relationship paradigm, (4) fully review the constructs that form Robicheaux and Coleman’s structure conceptualization and will also form this dissertation’s conceptualization of structure, and (5) review the conflict literature to identify the gaps that this dissertation seeks to close. When.completed, the background information that is required Us understand the model’s constructs and their relationships shoufld.be available to the reader. 2.1 The Fit of this Dissertation into the State of Channel Theory Thee first task of this literature review is to place this dissertation in the flow of channel research. Channel reSearch has progressed from early economic based deSczr'iptive research to the theory—laden, combined economic aTKi behavioral approaches that currently dominate. Umierstanding this progression of channel research becomes reelevant when answering the question; “where does this disSsertation fit?” And this question is relevant when pc>Sitioning this dissertation as a contribution to that prog'ression. Therefore, this brief history of channel research is intended to position the dissertation, not fully 29 ., . A ”v. ~v‘“ 'V‘.‘ .uvoorob .00» ~ .e;y:>‘.ro sorted-ne- y u . . I‘ I a I-R'peyvl anus nu-.~-.. v ~<-.‘, . . h ~ ve-. u.'_ \ a- v ~e-‘J ....‘ \ "-e, . n 5 ‘\ ' . ~ - . \ OV“ ‘~¥ ‘( u._ ' r‘. '~~'- outline the status of channel research. However, as the literature review continues, many of the topics touched on in this section will receive expanded treatment as the topic becomes specifically relevant to understanding the rationale and.background of the proposed model. (:hannel research has its roots in the Functional and Lnstitutional schools of thought. The early scholars in these schools identified the functions performed by channels (e.gu Shaw, 1912; Weld, 1917; Ryan, 1935) and the hastitutions that performed those functions (e.g. Weld, l9lrh Converse, 1949; Alderson, 1954). The research was Imostly descriptive and was explored from an economic Exarspective. Theories based on economic variables attempted t1) make predictions about channel characteristics such as Cflmannel length (Bucklin, 1965) and channel structure “Mallen, 1973). Currently, the economic perspective Ccnitinues to appear in channel research mainly in the application of Williamson’s (1975; 1979) transaction cost theory (e.g. Anderson and Weitz, 1989). In the 1960’s, a group of scholars started to consider the behavioral components of channels. Mallen (1963, and 1967) unmidered variables such as conflict and cooperation. mmever it was Stern’s (1969) book of essays on the 30 e CAI"... ‘1': E A .:.,“I owed- . all e ‘ v“‘;v‘ fl ‘ . .23... use an - . I Q. . i': 'C \_ ‘u-oo'uoen I I e ”A q'q a. .2! ~' fl u-‘ue ‘Aav uu . u e . a v- - A and oowvon u. ‘ I n.. . A _ev-~ A ,. '“““‘V-O us... e ..... h. .- .- d“ “>b. A a. ‘ Fee... ‘9 , 'znup‘,' -. ‘ nun.» v ' ‘9. -‘ c, v...~ . "e I.-- .~A- -\_.Va A a in- “v .. In. . I“. P~"‘-1 ' ‘v'- ’- — Um. e g a 3“ a. v-~~ . v-.. V. ..~ . s.‘-.. . ., . V l a e- ~,."‘ . . "- C g,‘ ,r a.“ '..: ‘A \_‘-__a e‘ Np" . C e I .‘ i .'e 5. . .. ‘ - b‘-\. fi . .4 87‘s“ ~ . 5 \ F . .‘~-e." a“. ‘4- .‘n.“ - ~ ”v- ‘v. e u p. n n“ "°~ c. . .. '- \‘.‘ .- .- u u e'. u] R-..‘ , \~. - . . ‘1 i.‘ § 5 ‘u n‘. ‘ A ‘ v~ ‘ 0‘ ‘ ' ~ ‘q behavioral elements of channels that pushed channels research away from economic analysis and toward behavioral analysis. Quickly two key topics dominated such research; power and conflict. Authors such as Hunt and Nevin (1974) anCILmsch and Brown (1982) considered how power effected channel control and structure. Lusch (1976a) and Brown and Da)r(1981) expanded theory concerning conflict in a channel setting. Gaski (1984) provides an excellent literature review of these topics. Additionally, section 2.4.2 of this Chapter will revisit the specific findings of the researchers who focused on conflict. TTMs logical extension for channels theory, which began with ar1 economic perspective, and then weathered a flurry of aCH:ivity from a behavioral perspective, was to attempt to CCHnbine these perspectives. Stern and Reve (1980) made a Senninal contribution in this area when they proposed the Political-Economy Paradigm.as the framework to consider both 'perspectives. The perspective has lead to very insightful combinations of variables that have expanded and strengthened channel theory. For example, Heide and John (1992) combine the economic-based ideas of transaction cost analysis (Williamson 1979), with the behavioral concepts of nfletional contracting (Macneil 1980) to show that it is the 31 'A eve! (In relational norms that enable firms to overcome the fear of opportunism and commit to transaction specific assets. The most recent change in channel research perspective has to do with the specific type of channel structure being considered. Initially, a great amount of attention was gdnven to corporate vertical marketing systems, and/or conwentional channels controlled through authoritative power structures. More recently, this perspective has begun to be repflaeed by research focusing on non-hierarchical relationships that are characterized by non-authority based Warns of behavior. Arndt’s (1979) “domesticated markets” Ifflpresents the vanguard of this shift. Most recently, a112icles by Nevin (1995) and Weitz and Jap (1995) highlight tKNN relationship—orientated channel research has become a Si&;nificant, if not the dominant, research focus. (Section 2.22 of this dissertation traces the emergence and findings ‘ . ‘ F. cub “Hit 2.3 u ' - . s- A -. 'eu. a. I'VE-vcuvu‘ '1 R _ 0" :h‘ V” hi. ‘_‘_‘.v ' I he; “ .- a- » 5 I"¢u ... - . \I- F C an. v..-~ nh~ ‘ _ D.~ R ube...h_-. '- e- ”'O- A? L .-‘.-~- ‘- ”A y— e._- RA,- ( v...‘ Q‘v. h,__ . A! fife. q...“ 1 ’e a“. s.“ ~ I. . t~-. . w H”, a_“ - e_q ' . in y - a.. ‘ “ h- .‘5 1 s . ',‘. ‘I‘ ~. . ‘I a .‘v... n e I‘ A x \._ ‘ ~_-.v-, 5“. R“ N ‘H I relationship stages) it fit better into the intent of this dissertation. A review of the R&C framework follows. 2.3 Robicheaux and Coleman's view of Channel Structure Robicheaux and Coleman (1994) provide a research framework for addressing channel exchange. Their model is based on the Political Economy Framework (Stern and Reve 1980). They use this framework to identify polity (behavioral) and economnc antecedents, channel structure, and polity (behavioral) and economic outcomes. The level and interaction of the antecedent conditions jointly determine the relationship structure that is likely to emerge in a particular exchange relationship. This structure, in turn, influences the degree of the various economic and polity performance outputs of exchange. (Robicheaux and Coleman 1994, p. 42) Their key contribution to the study of exchange is their conceptualization of channel structure. They refer to Reve and Stern's (1986) definition of structure as comprising an emonomic component and a polity component. 40 The structure of the economy is defined by the extent of vertical coordination between the interacting organizations ... The structure of the polity is defined by the power—dependence relations between the two organizations Combining elements of both the economy and the polity results in a picture of the structure. (Reve and Stern, 1986, p. 75). R&C use this basic structure when they conceptualize channel .structure as having two components: (1) decision—making structure and (2) operational integration. TTue R&C framework is careful not to include what the authors Ccnlsider antecedents or consequences of structure in their CCuiceptualization of structure. Based on this, R&C did not Sianly borrow Noordewier, John and Nevin’s (1990) IKElationalism construct as their assessment of structure lmacause they considered it to tap domains that were more (appropriately considered outcomes of the relationship Structure, not elements of its structure. They specifically Considered “attitudes toward the relationship that emerge from operations as polity performance outcomes” (Robicheaux and Coleman 1994, p. 47). As relationship structure was operationalized via the R&C framework, a thorough review of the dimensions of R&C’s channel structure follows. 41 . . I ‘nn- n- -v‘_ .....'.VII ‘ pnyzzn av,- Id. vim». u u . .. ..- 'I”: '— .u..-«.- I . Q flaky-p . .P I- 7,. Ovvue-vohfi‘ o ‘vq I A... 1 I s .‘~~vsr- ~ ~ F 'V-ueo_-‘ - "I. . . - r_ .. ."."“.¢§. .0. _. _r_ . '~'."-»go . ‘ F.~"'v- ~-.¢._“: O. . VA.~ “‘~.':"\£.- “s.-.“ ... v, . .‘ ha- “V- ‘b ‘V ‘- ‘u e,_ \ “p. . A ~ ”V. _P‘ § ‘ bu -| ..‘~ . ...- 'l. . ‘ c \ v.'~ . - “‘: ‘9'- V“: o .." ‘ u h . "~ “as. 0 ‘~‘ ‘ ’1 0" IIP (I) (I) 2.3.1 Decision-Making Structure Decision—making structure ranges from clan—like to bureaucratic. This concept was operationalized as a formative construct made up of four components. (1) decentralization, (2) informalization, (3) participation, and (4) shared paradigm. Each of these are discussed below. (Decentralization and informalization will be discussed as centralization and formalization as they are most often operationalized that way in the literature) Centralization has a rich history as a variable that taps Structure. Robicheaux and Coleman (1994) predict that increased centralization would be more associated with bureaucratic decision-making structure. John (1984) Supported this prediction when he used this construct as one COmponent of “bureaucratic structure”. Dwyer and Welsh (1985) also used this concept when they considered the relationship between environmental conditions and structure. They defined the term to mean the “extent to which decision making is concentrated” (p. 400). Dwyer and Oh (1987) also uSe centralization as a structure dimension and provide an exCellent review of the reliability and validity of the measures of this construct. Both were shown to be at acCeptable levels. Additionally, it was shown that 42 I . c u.- ""= : ....‘0--. "‘ .q... - 'I-- Y- .‘."‘V§‘n a...'~‘ ... "' \ ""0 in. ‘o. :‘V‘-nu‘.v_ he -~su-_‘- ~ ' . v. “n.. ‘ e...” .71 r r I [I l l centralization had a negative effect on relationship quality. Formalization is defined by Dwyer and Welsh (1985) as “The extent to which decision making is regulated by explicit rules and procedures” (p. 400). It is predicted by R&C that increased formalization would be more associated with bureaucratic decision—making structure. It has been a dhnension of both John’s (1984) “bureaucratic structure”, as well as, Dwyer and Welsh’s (1985) “channel decision structure”. Dwyer and Oh (1987) review and confirm the reliability and validity support for the measures of this construct. Additionally, they show that formalization has a positive effect on relationship quality: a result that was Counter to its hypothesized relationship (Dwyer and Oh, 1987). ~Participation is predicted by R&C to be inversely associated with bureaucratic decision—making structure. Defined by IDwyer and Welsh (1985) to be “the degree of actual INirticipation in decision making” (p. 400). Participation ‘Nas also conceptualized as a component of structure by Dwyer and Oh (1987). Together, these works establish strong Support for the validity of this construct by demonstrating discriminant, convergent and nomological validity. 43 ..- flu. Io. Q‘- Shared Paradigm. Unlike the first three dimensions of decision—making structure, shared paradigm has no empirical pedigree in the marketing literature. R&C consider this dimension to be their contribution to the concept of decision-making structure. The concept derives from the thoughts of Wilkins and Ouchi (1983). Nor is ... “paradigm” meant to imply a sharing of goals, as typically defined. We have in mind the sharing of general assumptions and values that Kuhn (1970) described among scientists. This term suggests that clan members may share general orientations” ( p. 471). Wilkins and Ouchi (1983) go on to state that “the paradigm Suggests how members determine what is in the interest of the clan" (p. 471). This paradigm concept implies norms or “GXpectations about behavior that are at least partially shared by a group of decision makers” (Weitz and Jap, 1995). Thus, the concept of shared paradigm implies shared norms, (and thus the operationalization of ‘shared paradigm” will COnsider the parties perceptions of how closely the norms of ‘the relationship are shared. This concept does not imply specific norms. In other words, it does not expect that clan-like structure must show a high level of the norm of mutuality (Macneil, 1980), but instead, “O ~- . A ‘a- ..‘o-~‘\,v‘ 9 0!. . ‘N \‘Z’ "~'.~--‘eo -‘ “‘ru... 4 n. "5 ~ ‘. o i":’ “ A ...... ..U II! (Y, 't1 (1’ U) Q. ~ A I._ v-"‘ ‘5 :":-—‘,. Q. h ‘\ - ‘y. ‘» : ‘R' 'Q ... N...‘- ~ “g N. ‘\ \‘. ‘- L '--.; “u «“ §‘~ ...‘c “a" “ ‘F ‘s .- \‘ ‘ 9‘. e ‘g ‘5". \ 9 M5 ~ V-_ “' I— ‘\ n TTLiS concept implies that in more “integrated” relationships tkue traditional arms-length roles of buyer and seller have benen replaced by more complicated roles. Heide and John (15990) found this construct to be related to the presence of triansaction specific investments. Zhssistances is defined by Noordewier, John and Nevin (1990) tr) capture the giving of help to an exchange partner even “dien no help has been specifically called for, and/or no iJnnediate or explicit compensation is provided. Assistances Eire present when the exchange parties go beyond the a priori level of conduct called for in the relationship. This concept was operationalized by Noordewier, John and Nevin (1990) as one (of five) dimension of their “relationalism” variable, which was considered to capture the level of relational governance in the purchasing relationships. No individual empirical results were reported for this construct because it was combined into the second—order construct before analysis. Amnitoring consists of the actions undertaken to ensure exchange partner performance. At the market end of the continuum, there would be little action taken, and at the integrated end of the continuum the level of monitoring “muld mimic a corporate vertical marketing system 47 r!- (Noordewier, John and Nevin 1990). Again, since Noordewier, John and Nevin (1990) treated this construct as a dimension of a second—order construct, no specific empirical support was generated for its potential effect on outcome variables. Information Exchange, as considered here, deals with three elements, (1) the frequency of information exchanged, (2) the type of information exchanged, and (3) the level of information exchange capability between the firms. The first two components represent the conceptualization of this construct by Robicheaux and Coleman (1994) and Noordewier, John and Nevin (1990). The third component is most directly borrowed from Bowersox, et al (1995). Integrated exchange is expected to be characterized by high quantity information exchange (Noordewier, John and Nevin 1990). First, a more integrated exchange relationship would share a greater amount of information about the exchange basics (price, specifications, delivery terms), however, in addition to this there would be more information exchange because new classes of information would be exchanged. In an integrated relationship it would be expected that more long- term focused, proprietary and strategic information would be exchanged (Palay, 1984). Thus, we have a change in the type Ci information exchanged also. 48 l" .1 I Exchange capability is defined as the capability of one firm to receive information from their exchange partner in a timely and usable form (this is adapted from the connectivity concept from Bowersox, et al, 1995). This element of information exchange is being added to the R&C conceptualization of operational integration. The ability of one firm to receive the other’s communication in a usable form is a relevant consideration for a construct measuring information exchange’s role in integration. It is proposed that in more integrated relationships the firms will exhibit a higher level of exchange capability. Similar to decision making structure, Robicheaux and Coleman (1994) conceptualize operational integration as a formative construct. However, once again, there is no empirical support for the construct validity of this conceptualization. A goal of this dissertation was to provide an initial assessment of this construct’s validity. 2.4 The Conflict Literature in Marketing This section will address two major areas related to conflict: (1) conflict’s definition and conceptualization, and (2) a review of relevant findings related to conflict 49 -v- «. n‘lh V a : and other constructs in the model. For two good general reviews of conflict see Gaski (1984) and Brown, Lusch and Smith (1991). 2.4.1 Definition of Conflict There are two different ways to conceptualize and define conflict. The first is conflict as a series of stages. This conceptualization is based on Pondy's 1967 work. In this conceptualization, conflict begins as a latent sentiment, is then perceived by the parties, becomes felt as tension or disaffection, moves to manifest conflict where it is expressed as a behavioral attempt to block another’s goal achievement, and finally passes on to a conflict aftermath stage. This stage model, while well accepted, is not the only conceptualization of conflict. Stern and El—Ansary (1977) provide another conceptualization: Channel conflict is a situation in which one channel member perceives another channel member to be engaged in behavior that is preventing or impeding him from achieving his goals (p. 283). Gaski (1984) argues that it is the perception of conflict, regardless of the actuality of behaviors, that is the critical element of conflict. 50 .- nyAwnc ... g‘.vUv~ . . N. Moo-An rj;r' Lu.- . -ba ‘9': b a... O o .... ~n ...- \ \ .. _ «a - ‘h‘An n‘-.-».. ‘ - ... ‘ '~».. ‘\ u (I) I). y ‘ - He proposes what he calls a “consensus conflict definition”: Channel conflict will be the perception on the part of a channel member that its goal attainment is being impeded by another, with stress or tension as the result (p. 11). In this dissertation, that is the definition of conflict that will be relevant. Measures tapped perceived conflict. Additionally, the concept of hindered goal attainment was applied broadly. Specifically this dissertation considered “impeded goal attainment” to imply opposing goals and/or opposing strategies and actions used to obtain similar goals. Thus, conflict in this dissertation was conceptualized as a perception that captures elements of the perceived, felt and manifest stages of Pondy’s model. While manifest conflict is conceptualized as actual behaviors, the fact that “goals are being impeded” implies that this is being done actively, or through actual behaviors (although there could be the case that no actual behaviors do exist, yet a party to the relationship still perceives conflict). This approach avoids the problem of assessing the Conflictual nature of specific activities or behaviors. Assael (1969) argues that attempting to conceptualize 51 .; an»: ~ fir V‘V wvuuoob b V iaa “'2' it ac: car. ‘uuU v S n. , ““Wfl- >9 fiv- 5Ual‘...‘u‘- 3"" “r- bob: 0 . Pv- be. UV. ' a I.' . . . .~..‘F\V~A . ~ A , ..ungu.b¢eu--1 I .A H .- I u‘.‘ ‘ Q A.“ - MY De ..VI‘.. ‘ ‘ n n In. . hur‘. y.. 9 Ah“ V5“ (1 '1 C) 5 takes a- be 1 c. a v‘ \- 0 “n ... has! L‘9e Crac s V. Y. ~' c‘ . n ‘ e- ‘H ! C“"» v - t ...-V. ~'P ' I ~.\_~. Q““Y fi‘G“s. ' ‘1 ' t \- -"".v- V - CCaueVb Ce ”5;... My. “\Vfi U‘U eSS ‘ § ‘ C A p... u . Ya- I V \Vd‘l aha ‘ ‘-u ""na: » \V-.u‘: A.“S ‘- su v— Q. .. ‘. —_‘~ ~Q' C0“‘7; C‘ ‘ . «‘«C~ ‘ '5 “‘- - V 3-‘Qan‘: ~5~~ A.“ ~ ~ ‘V“ “a . “]\A F'- , “‘a! ‘- I‘.:V;. .s‘ .A ‘ ‘ h b‘d“a conflict via actual behaviors can be difficult. He argues that an act can be perceived as conflict when economic pressures are strong, but a simple irritant when economic pressures are low, thus, the act itself is not conflictual per se, but the perception of the act under certain conditions makes it a conflictual behavior. This view supports the conceptualization of conflict as a perception. Additionally, Deutsch (1973) shares the view that conflict is a purely a psychological phenomena and actual behaviors can only be considered conflict when they are filtered through the perceptions. Thus, the presence or absence of conflict is never rigidly determined by the objective state of affairs.... the importance of “real” conflict cannot be denied; nonetheless, the psychological process of perceiving and evaluating are also “real,” and they are involved in turning objective conditions into experienced conflict. (p. 11) Thus, conflict in this dissertation was defined as a perception and the Pondy stages model was not specifically Operationalized. 52 a" ‘n .l ' ,. Cl: “ (MG MK Ct It: ywuuee a t ' , '1 5" “V ‘.' :nuob 'u'uee EGoO J A . .. ~ . has! fi!fl‘fif‘fi-F.Il .- ‘._‘~“ ustfrd...c-a V . n ‘ nan-a. npw-A uwfi“ ‘1 F A. “1‘..er L¢u¢¢ L. a ‘ ‘ gar-‘wu h'-A|~ ..,- ..-...,...t ..S at a c :‘V‘vnb..r “massage as an . pan 0 AvO‘-rf aviA ‘F; bhuafichL .Au '5‘.‘ ’7‘?" Q l ‘ “haven - w. ~ .. . venue u ‘6‘ y... “Avhl . \ . ~ “g? ”Fervs "“~§\d.. ‘V A .— .04.“... . . ‘ ”TI" ‘9‘ " "' Ao-- ‘c‘h. ' In Iav’I: . . "fiqg‘: . A, V'““‘\rs rt ': -U .c. ‘h ‘ v ‘W .y R“ ‘ . ‘Lsya-‘J P" Hi A a :‘x ‘. 8 vua.‘.~b ch‘a‘ Q ‘ :th .A’ 's\‘.s “FA ‘ H P ‘dvd‘ h . v- ... v . *‘s.e Q‘ x...“ . :‘~.~ u“::° I v :. \ . F ."t..| Ev” . 5.. a“ 'A u...‘ . IQ 2.4.2 Relevant Past Findings While conflict received a flurry of attention from 1969 until the early 1980’s, the accumulation of findings has been disappointing. Findings relevant to the model conceptualized in this dissertation are presented below. Of conspicuous absence are any studies that consider channel structure as an influence on the relationship between conflict and the outcome variables that make up the model presented in this dissertation. Schul and Babakus (1988) do consider channel decision structure as an intervening variable. However, they consider its influence on the power — conflict relationship; a relationship that is not specifically operationalized in this dissertation. Conflict received great attention after the release of Stern's 1969 book of readings on the behavioral aspects of channels. Stern and Gorman (1969) first addressed the causes of conflict and positioned interdependence as the fundamental cause of conflict in all distribution channels. Rosenberg and Stern (1970) picked up this approach and identified three propositions that linked goal disparity, domain dissensus and perceptual differences with conflict. The causes of conflict were not the focus of attention in this model, instead, based on Stern and Gorman (1969), 53 ...5' 3 "‘ was 5 ......0V ' . Q ' y~'-~~An "“"‘ .1 a o‘o.uv.V“ “‘5" a. ‘ ... «Av H .- ‘nfi. eeV‘b H c" . . a - '.:;'p AF..- 1 _ nuuveahnvvnu .-.. u- v ape:r.‘~ WV“.V“ «....-eau U- vb. d a o ., . “'H AFRVI ' -~ov-\.‘¢\,l I a CID -.Ione“, ( I It: (I) 9"" Q.‘ conflict was assumed to be a standard component of exchange relationships and its characteristics, and thus its causes will not be addressed. Rosenbloom (1973) contributed to the conflict literature by offering propositions on the effect of conflict on channel efficiency. Lusch (1976b) took this perspective and attempted to assess the effect of conflict frequency on retailer performance. Unfortunately, his study showed mostly inconclusive results. There has been no major follow-up attempt to assess this connection. Most conflict studies after Lusch’s (1976b) piece began to use “satisfaction” as a proxy for economic performance. Brown, Lusch and Smith (1991) provide an extensive meta— analysis of these studies and show that conflict has a negative effect on satisfaction that averages —.5 and ranges from -.11 to -.89 when assessed with a 95% confidence interval. This finding is important because the model conceptualized in this dissertation attempts to show a positive, indirect relationship between perceived conflict and satisfaction. The concept of constructive conflict outcomes acted as the mediating variable. The rationale for this proposed relationship will be discussed in section 4.0 of this literature review. 54 I'D- ,— hhv vuo‘ h»!- nvu la.- v v.- .u ‘\ \ \ Only one major “conflict” study addresses the concept of constructive conflict. Assael (1969) identifies five constructive outcomes to channel conflict. The specific connections between his findings and the dissertation model will be addressed when this literature review considers the conceptualization of constructive conflict. However, of relevance to this work, is the fact that his study made no attempt to (l) relate conflict characteristics to the emergence of constructive conflict, (2) assess the effect of structure on constructive conflict, or (3) connect constructive conflict to measures of relationship outcomes. All of these tasks will be addressed in this dissertation. Conflict measurement has also received attention in the literature. Rosenberg and Stern (1971) made the first attempt at suggesting a standard for conflict measurement. However their “difference score” method gained little acceptance. Brown and Day (1981) provide the most accepted view of conflict measurement. They concluded that a multiplicatively combined measure of frequency, intensity and importance best captures conflict. However, they indicated that other combinations of these elements of conflict could provide acceptable measurement under the correct conditions. This measurement model for conflict has 55 I I Q Q am n rNa " 32:5“ t, .y¢ .-" ..vg-‘npefl S; "K‘ \A nan-ha... -..» .. . . . . Dunc "U have ‘ A ....u‘ “Rugbstl. ‘ k '1 | - n-vb ~ 6 by.- I‘ve-0“ L'I ban. 5 " . .‘ (...: 'f‘. Ffi’fiy fiv¢¢g¢¢¥y 0.1:..- A "C‘. A; u". '“5 y" ‘ ha '0. V». ~ aw: . =6- C v, ' “ °«-y C a M.."‘-h . ""m‘ .- f“n.- ‘ w 5‘ >- O“ O c‘-"- .I- N ‘ ub‘.‘-b av-Q “.- .a b“ ‘5 " 5 ‘~. '. ”J wfla av; ‘V'I .. 5~:"1 J} k": . ‘ “| .- .R‘ "~_.\ w. . e v Q ‘ ““1 ‘h l e, - “\S I" \A - \~‘ ~ . ‘el 5'. “~e V \ I“: 'v .A‘ .. l ‘ A I V“"."\‘ 'v \ ~‘ . n. e e A ‘~"~¥ \' . . ex“ 5. fi‘ \‘. . V been partially accepted in most studies that have been conducted since its publication. While there is support for this multiplicative combination of these characteristics of conflict, this dissertation addressed those three elements as separate in order to remain true to the tenets of Social Conflict Theory. Further explanation for this decision is addressed in section 4.0 of this literature review. 3.0 SOCIAL CONFLICT THEORY & CONSTRUCTIVE CONFLICT One of the goals of this dissertation was to execute an empirically based theory test of the tenets of social conflict theory. Therefore, this section outlines those tenets and positions the concept of constructive conflict as arising out of this theory base. This section is divided into two areas; (1) the general tenets of social conflict theory will be discussed, and (2) the concept of constructive conflict will be developed. 3.1 Social Conflict Theory (SCT) When this dissertation refers to “Social Conflict Theory” (SCT) the references are to the works of large group sociologists such as Georg Simmel and Lewis Cosar, although other scholars, such as Karl Marx and Max Weber, contributed 56 :3 this area as we. . l to 1“war 5 ‘deee . . . . ‘5- -Afes my (19 CF ‘ ‘5: eye 0: n y. u- a- v ‘ I q. n s e '5' 9r ‘rfiu‘y\f‘~‘ '- .ve vu LAAM. V -uu . :- .....u mu“ Mn aver ..:a...S DE " . ' : .m- -.~ ,. .. bug‘.. V. CE. :13} EYE: , H". ‘I 5‘! h 1...: Q a_ \ .- “-.' ‘ .v b“ LL. “8...: a l "van-qr ; A ‘ l ‘ '\ -~.»..vus C5 VCQGA .I,‘ h . m“:- 4M9 ch” ‘ ‘ .. -u‘....\,-sl 80“ f‘ g! V‘ §U-J h C H‘U— 0 '- I ... ~ ”A ' (h- ‘Wv. C ...“- * “KL 6 . ... ‘ See.“ ¥‘. 1. ".:~C" _‘. ~b~ . 'K‘Q ZR; ‘ \. V.‘ ,‘.r~t; ‘ ““Q‘f‘ a V ~‘ Q“" Q “\4 C. s.‘ ‘ 5--;‘. V .'\ ~+ \ s V). ’7‘- “-63?" I. v.‘ ~ _ "‘a‘ h . b V :H‘ 4“, “D *R ‘ 51‘) < . I'D-..c \ \“3fi‘cd ' ‘ "w 1* A a. . "V \. ‘ .:‘ “Q. ‘(‘~ ~h ‘5“ g V o Ch F‘“! ‘§ “‘s C: .Q 5'- ‘~ “Ne ‘1! . Q to this area as well. For a thorough of these works please refer to Turner’s (1986), “The Structure of Sociological Theory.” SCT makes predictions concerning the functions of conflict for the individual parties and predictions concerning the reasons behind increased levels of conflict violence and duration. However, what makes SCT most appropriate for this work is that it makes specific predictions concerning the functions of social conflict for the social whole. With marketing channels, and the relationships that make up these channels, solidly positioned as social systems (Stern and Brown, 1969), SCT becomes a relevant source of theory by which to make predictions concerning when conflict within these systems will have constructive outcomes. There are two relevant considerations to using SCT within the channel setting. First, the definition and conceptualization of constructive conflict needs to be clarified, and second the specific predictions that Social Conflict Theory makes about the role of conflict in relationship to constructive outcomes needs to be transformed into the relevant setting and language of Imirketing'channels. The first task is addressed below, whilje the second is covered in section 4.0 where the 57 :uvv.s.\y a- . ana~§:;p 'ra: titer!“ :3? 3.2 V:o.. ”‘v‘fah . ......v ...c‘u" _- ‘ .:.'; R-A" .va- .u-v Vv-.~_- ‘ ‘ . ...;VQA" \ ~,.~ G...“ ....l in! ...i ' 97-; fib'fi‘v~« ”“5 \" ‘=v§ ‘ § ‘ viz. “ ‘1 ‘ vua‘. ..“Sbr § Q" I ~ '“M a: .5 Qt, IV ‘I e c. _ a k' ' A ‘fix‘ “‘v‘,. f. VV‘Vu‘ 3. . "~. - \~ ..(0 AAA‘ ~ a» bay v s u ., 2 - .'-».v :v.‘ “-. r- 4" .‘ u":5‘-" ‘ “‘ “ V “."Q ~, “ V d .£. y ‘A‘ *\ unvA "~-=s v. 5.. C Q In 's‘;Y ‘ ‘ C en M specific transformation of those propositions to the channels setting is covered. 3.2 Constructive Conflict Outcomes Many marketing scholars have suggested that conflict can have constructive outcomes (Dwyer, Schurr and Oh 1987; Anderson and Narus 1990; Morgan and Hunt 1994; Weitz and Jap 1995). However, there exists no rich conceptualization of this construct, nor an empirically supported nomological net that illustrates its antecedents and consequences. The creation of such a conceptualization and relevant nomological net were fundamental goals of this dissertation. 3.2.1 Definition of the Concept Lewis Cosar (1956) considers conflict to be neutral phenomena. It is the outcome of conflict that can be potentially destructive or constructive. Additionally, he believes that a healthy social system needs conflict in order to survive, adapt and grow. 58 Cosar expressed this sentiment when he wrote: “No group can be entirely harmonious, for it would then be devoid of process and structure. Groups require disharmony as well as harmony, dissociation as well as association; and conflicts within them are by no means altogether disruptive factors. .... Far from being necessarily dysfunctional, a certain degree of conflict is an essential element in group formation and the persistence of group life” (1956, p. 31 - Cosar’s paraphrase of Georg Simmel (1955)). “Conflict is not always dysfunctional for the relationship within which it occurs; often conflict is necessary to maintain such a relationship. Without ways to vent hostility toward each other, and to express dissent, group members might feel completely crushed and might react by withdrawal” (1956, p. 47). While social conflict theory certainly supports the concept of constructive conflict, marketing scholars have not always agreed. Anderson and Narus (1984) for instance, ask respondents to assess the quality of the working relationship that they have with their exchange partner by considering the intensity and frequency of disagreements. According to their scale, the highest quality, or most cooperative, relationship would be recorded as “free of any conflict” (p. 72). This illustrates that conflict is conceptualized only as a destructive element. Interestingly, their measurement model presented many Lnoblems, and they were unable to use this conceptualization 59 1: F‘”“"f‘t a L‘ 5V0.--.\r ‘ ' ' . F n a why. 7‘ C39.“ nave MC . 'qsnb‘P‘op. ~ vve ‘VUVOHV u... I «_~_ __‘ ."33 {EStaIC “"‘N'Inb '\. . \v‘uo‘.L.. al,! I',Avlnbru.;‘. . " w \ , . .“i...v_-.e . l 'A q“fln—‘ t c . ,. . .IV‘VQQ“. N o -O_‘ ."‘CH “by“ I U . ‘ "a.“ . \~“ n ““ . ‘ ‘ ..a“s, i . . ~ .- of conflict as a distinct construct. These difficulties could have been the result of the shortcomings that a “destructive only” conflict conceptualization would present. Other researchers have embraced the idea of constructive conflict. Dwyer, Schurr and Oh (1987) considered the constructive role of conflict to be an area that begged for research attention as the discipline moved towards the study of relational exchange. Anderson and Narus (1990) answered this call when they used “functionality of conflict” in their model of working partnerships. Their conceptualization was based on conflict providing improved productivity to the relationship. Unfortunately, the fruitfulness of this study is limited because the authors had difficulty with the measurement of their construct and were forced to drop it from one of the two models they estimated. Morgan and Hunt (1994) operationalized a “functionality of conflict” variable in their model of relationship marketing. However, their conceptualization of this variable was based on future intentions, instead of past or current sentiments and thus had limited relevance to this dissertation. "a-‘DR 7"; Ta ('13:: Cgu‘ an“ ‘1 p " ‘ I a"; O v'L h ’h " '5' fl *9084‘ ' evuoao¥b V“ ‘ ... r‘”‘.‘f‘.* “n r“ \y’.t..¢\v5 CO“ H.» P‘ n '. - . I r ‘Av‘r‘r‘ knee .e-ah¢u«;:--. ‘ - u u. bn-t. ~§ a,“ ".3.” ‘5 .G..u-' ' s 1.1: :persona- Fr“ 9 ”...“... ... u‘u ., CO fizuu‘aa- \ :Ou’ sgqcp. A“ “V. -Q \. -Vco - .- p .n . 3:42 "MFR be » .....ors 27.6“ .. Rn ' ‘ ‘ . - ‘ “a v """‘"---.. *w- n F r “ ‘0 \a‘.. ‘(a H 7- ‘. ' F u v: ~0 0'. .1: E .‘w- ~n..‘.‘ ' a. I . ' ".~'“G‘ “e h as l E . . - ‘9“ ‘ ~. F v- \ “we! d r‘ I” ‘~ a.“ ““b I ....» w. . | t (e NR , .‘ °"~filr,{\—‘ ‘ r § --. Weitz and Jap (1995) also took up the cause of constructive conflict when they wrote: Conflict can often act as a source of novelty for the relationship, forcing it into new terrain that, if handled successfully, can strengthen the interpersonal relationship and cultivate greater trust, communication and relationship satisfaction. (p. 315) These authors then proceed to make elaborate propositions concerning when constructive conflict resolutions are more likely to occur. Unfortunately, the piece is simply a conceptual one, thus no empirical test of the propositions is made, and thus, there are no results to discuss. Thus, the marketing literature provides scarce direction (and almost no empirical support) on (1) a rich conceptualization of the concept of constructive conflict and (2) the identification of supported antecedents or consequences. 61 \- 33u‘e<*°' Ccsa Ibo! I I any-n nOn-t vn-a .L...e.y-ad.-..d . p «O afar ‘7'"f‘" 1‘. intro. weave. TV 16.Av\ lnaa .ygb '. f‘f‘f“ -;- VVbe§finv 3 Dr: .‘NEET H» ‘1 ~':.‘ g r--.- ‘-'a- < a. V"~-.. Ora“ 5“vr~,\ u“\'r- \— . f‘hv-o ' I up... -- . . E ""v --...-“ e 5-; ‘ Jig. ~ “hrs... b“‘IV“-. - s Ar... . V“§¢. 0"- { Hus '-’v (3\ However, Cosar’s (1956) writings can provide a robust conceptualization of constructive conflict. The dimensions of constructive conflict are hinted at in Cosar’s writings. In loosely structured groups and open societies, conflict, which aims at a resolution of tension between antagonists, is likely to have stabilizing and integrative functions for the relationship. By permitting immediate and direct expression of rival claims, such social systems are able to readjust their structures by eliminating the sources of dissatisfaction. The multiple conflicts which they experience may serve to eliminate the causes for dissociation and re- establish unity. These systems avail themselves, through the toleration and institutionalization of conflict, of an important stabilizing mechanism (1956, p. 154). 3.2.2 Dimensions of Constructive Conflict Outcomes Three dimensions of constructive conflict outcomes can be inferred from the writings of Cosar (1956, 1967). These are; (1) the release of hostilities before they build up to dangerous levels, (2) the revitalization and/or realignment of norms within the relationship, and (3) the innovative and/or creative reassessment and reorganization of relationship activities (Turner, 1986). The conceptualization of these dimensions of constructive conflict will be addressed below, while the operationalization of these dimensions will be addressed in Chapter Three. 62 ’ .h It.)' _ A...— 1 Q pAbAf“! co > . ~ Fur-vuyc on»: b . . Ami-nt- Q-Ahuiv .. x r L ~ ....Uv.‘ v- c - mere .13-IS S ...-...A F‘Y‘ ~ h- Hdioaoi 3 MG. .. o :“vchs~ .... C distruye, .. -..: u . . . ‘CC‘Pfinbuc. r“ “buvcdvbfil ~ ' l ‘ | V: ;“ an an . .v“'¢\v’.eb O‘-p ~- A no" I we ‘ Ir av .u».. ..e S W _ . u "‘N “*-...\4‘ v I . Ye u "C': ‘ ‘us . l Vu -. ac“‘fl.“"fl‘ {ID H v...\“ | r. ~‘ ‘ . d.“ 1R“V' Releasing HOstilities. Constructive conflict allows the system to release hostilities before they build up and potentially tear the system apart (Cosar 1956). This constructive outcome is highly relevant in today’s climate where firms seek close relationships with many of their key business partners. Close relationships are more prone to suppressing conflict and are thus more likely to have these destructive blow—ups (Cosar, 1956). However, if the relationship allows conflict to occur and works to resolve it then the system can be spared such violent outbursts. Damming up of unrelieved or only partially relieved tensions, instead of permitting adjustment to changed conditions, leads to rigidity in the structure and creates potentialities for disruptive explosion (Cosar 1956, p. 47). This negative or destructive outcome can be avoided through conflict’s role as the mechanism of hostility release. 63 1? I" A V...”- ‘ x ‘ i rune..- \— u n by. Fr .v -v-A ... ...C ”Fueer.. \ ~.... ... a . I ""‘ca Qua-y‘,‘ ...“ " was: " n as, “A;- \ ...7 a..“" a. - " o u- ‘ "‘5 s. .‘ 'I .' I ~ c~‘ . N. "~ .... -. AA’: .. ”Vu‘ ’ . .1 - . \L - “'~» ‘ ' ‘.. CA V . qv Q‘ fa by AA “V W. ‘I K . a. . x ,. u f- ~ ~‘§ \ y .. .Q ”a.“ rxw . ‘N “ ‘fl‘ ‘ A ‘ ‘Q ~‘: '«Iv- ‘1“ -,_. h ‘5 Again Cosar: Conflict is thus seen as performing group- maintaining functions insofar as it regulates systems of relationships. It ‘Clears the Air,’ i.e., it eliminates the accumulation of blocked and balked hostile dispositions by allowing their free behavioral expression (1956, p. 39). In the only study that directly attempted to assess the constructive outcomes of conflict, Assael (1969) found that indeed the release of hostilities was a constructive outcome of the conflict in the automotive distribution channel he studied. Thus, the release of hostilities is positioned as the first major benefit of conflict in a social system. Adjustment of NOrms. The second major constructive function of conflict is the adjustment of the norms that govern the relationship. In the course of conflict new rules are continuously created and old rules modified. By bringing about new situations, which are partly or totally undefined by rules and norms, conflict acts as a stimulus for the establishment of new rules and norms (Cosar 1956, p. 124). Cosar sees this role as fundamental to long—term success of the relationship. And again this seems a particularly relevant role of conflict as we create and test theory related to the establishment, maintenance and dissolution of ...,- I long—term relationships. Cosar argues that the adjustment of the relationship norms are necessary because circumstances change and old rules and norms of behavior cannot possibly be permanently established to handle the diversity of circumstances that a relationship will face. As a stimulus for the creation and modification of norms, conflict makes the readjustment of relationships to changed conditions possible (Cosar 1956, p. 128). Thus, for the long—term health of the organization, these norms and rules will need constant updating. This logic is analogous to Macneil’s (1979) arguments supporting the need for relational norms to support written contracts. Additionally, Assael (1969) provides empirical support for this contention. His results indicated that the updating of norms was a basic benefit of conflict. The adjustment of the norms of the relationship thus became the second dimension of constructive conflict that was considered in this study. .Reassessment of Activities. The third dimension of constructive conflict is the reassessment of relationship activities with a focus on increased innovation and creativity. While norm re-evaluation is one component of constructive conflict, the adjustment of the actual 65 In y. activities of the relationship is another. This distinction is between what is acceptable behavior (norms) and what are the specific events that will produce that behavior (activities). Cosar (1956) believes both will be adjusted by constructive conflict. Morton Deutsch (1973) agrees with this assessment. He considers conflict’s constructive role to mimic the creative process and considers conflict to be the starting point of positive change. Specifically, he contends that conflict will enhance creativity in activities by forcing parties to assess problems that might otherwise go without attention. Thus one of the creative functions of conflict resides in its ability to arouse motivation to solve a problem that might otherwise go unattended (Deutsch, 1973, p. 361). Thus Cosar (1956) and Deutsch (1973) agree on this role of conflict. Additionally, Assael (1969), in his study of the constructive impact of conflict, found that the reassessment of past actions was a basic benefit of conflict. Therefore, the reassessment of relationship activities is the third dimension of constructive conflict that was considered in this dissertation. 66 -—,,-__ 1r .fl". .-. ..- I -‘_ 1 ‘ . A at. VIA, 'u a P‘ \ ...).lvoyun... ...u. .v.. . Fv;b, I . :‘vu .“ \— ‘01 .. a. v;- ..,:.~ :y: \- u‘\ u ‘n y. ""‘A. “ u— 5"- a Ie. 1,. :s r» I ..5 .... s ‘§ c v.‘ V U ‘ " -‘QA “. ‘u l‘. . - -_..‘ ‘ Q '-. A; “ § v . n‘ ‘ '\ .'A I h .N: \C ~~ 'M \ a, I. 'a‘ ‘ n . "J‘ Pv- V§ Va 4 . THE MODEL This section reviews the logic that supports the relationships proposed in the model illustrated in Figure 1.1. Each relationship is addressed in the sub—sections that follow. The specific hypotheses represented by each path in the model are presented in Chapter Three. 4.1 Conflict Characteristics and Constructive Conflict Outcomes There are four relationships between certain conflict characteristics and constructive conflict that were operationalized in the model. These included conflict frequency, conflict intensity, conflict importance, and conflict focus. Brown and Day (1981) found that the most robust measurement of manifest conflict was a multiplicative contunation of conflict frequency, intensity and importance. IHowever, in this study, these characteristics of conflict xmere purposely measured and operationalized separately Ibecause social conflict theory indicates different relationships between each of these characteristics of cxnnflict and constructive conflict. Combining the measures <1f the separate conflict characteristics would create a sitiuation where distinctions could not be made. Each of 67 ‘ A A: an . ""8 by U‘Uv. "r M" ‘ .n , ""‘O—V I. RI... ‘ A mun“- v 0. . AA... ‘ a vin.‘.‘ V . .v.. a- ...“ ‘ . ,‘ -.~.- kv~_‘-‘ a.‘:. V ‘1 ‘ :D“ \.‘ 9“,. ‘. u ._ H. '. x .l 'V ‘1 . s v p.“ ‘5 e ‘- these constructs and their linkage to constructive conflict is discussed below. 4.1.1 Conflict Frequency The frequency of conflict is a very common characteristic of conflict that has been measured (e.g. Lusch 1976b; Schul and Babakus 1988). Social conflict theory indicates that high conflict frequency is likely to produce constructive conflict (Turner, 1986). The reason for this connection is that frequent conflict will allow for the release of hostilities before these hostilities polarize the parties. Hostility releases of this type should allow the relationship parties to avoid a severe blow—up of intense conflict. 4.1.2 Conflict Intensity The intensity of conflict has also often been assessed (e.g. Assael, 1969; Rosenberg and Stern, 1971). Social conflict theory defines intensity to include people’s involvement in, and commitment to, pursuing the conflict. SCT predicts an inverse relationship between conflict intensity and constructive outcomes (Turner, 1986). Constructive outcomes are considered to be more likely when conflict 68 ;'.A',~Q Div " r eubvtc oh. 4 u ay- 7:. H. .ta.- (n (Y a . . , I’M-u... n w _ i . teeth». ' r- . . - - Q IV“'- ‘Fn “'8“ an.“ a. ‘7'..." 'l "-"Vc- ~ a \AT‘A NF U'V.c‘ s,“ ‘a . ”AI-o .....--L»E A-.“A A“ v 1 ~‘ "Vv.._‘_~ - I O ~A‘A-Y ‘ ‘ ‘F ‘ v.~“ - . . Q w-‘Fn '— hob.‘ u..‘. a.‘ ' . ., Cr n; a“ y« ,. u. ‘ "‘4. “F“ \V.‘ ". ‘ a ' b s. ‘QF \ “~‘. a intensity is low because the parties will stay more focused on realistic issues, the parties will communicate more effectively and the parties will be more likely to allow norms of conflict resolution to work effectively. 4.1.3 Conflict Importance Brown and Day (1981) found that conflict importance was a critical characteristic when assessing manifest conflict. Social conflict theory postulates that low violence conflicts will be characterized by more constructive outcomes (Turner, 1986) and conflict issue importance is closely connected to SCT’s concept of conflict violence. Cosar (1956) defines conflict violence as the degree to which the parties are seeking to injure or eliminate the other party. Conflict violence is said to rise when the parties are fighting over a centrally important issue because their emotional arousal and involvement related to the issue will be increased (Turner, 1986). Assuming this relationship holds true allows us to make the inference that when conflict is over more important issues, the conflict will tend to be more violent and then the likelihood of constructive conflict outcomes would be reduced. 69 a ' h to u an,” '12» Gnu u» in vuc . .ae- t ' 1;.» h” I Q I F ‘ "H F 'Ffl ' bee-aU aau _ v : ‘Iflfia RYC Jawu-Sc.. - “T. a. t~~ ... .. dude ~_ . . . "" Q’RA .( I.v....~ . ..Y ”Ar. 4 h. uvao“.\' ' a .Y~,a .": ‘.“l" iraou S‘- { V: ;.,:" ‘.‘U' .~' “ - at. ‘AA.,: ) a‘o‘ ~" I "A ‘v.. hVQ,‘ -V,‘ . . ~ I . A‘ ‘5‘. .. ~4’ -.. ~K . 1 n3? ,, Id~ ‘ l -1 -n ‘ $1: I.“ . i. I & v.1“- . 5 Q‘. “ “’2'. -“ -.. . I ”A ' 1 “"3 =:~. U‘U‘ Dant and Schul (1992) support this logic when they argue: “the links between high stakes issues and corporate well— being and/or ideological commitments are likely to prompt franchisers to seek favorable resolutions on such issues at any cost” (p. 42). The “at any cost” mentality would suggest high violence by Cosar’s definition of conflict violence. 4.1.4 Conflict Focus Another linkage in the model is between the focus of the conflict and constructive conflict. Stern and Gorman (1969) argue that most conflict would be over procedures on how to achieve goals, yet Etgar (1979) positions goal congruence as a fundamental cause of conflict. In light of these competing contentions, assessing this characteristic seems relevant. The focus of the conflict captures whether the conflict is centered around different goals or different methods of achieving similar goals. This linkage is highlighted by Cosar (1956) when he states “Conflicts arising within the same consensual framework are likely to have a very different impact upon the relationship than those that put the basic consensus in question” (p. 73). Dant and Schul 70 (1992) attempted to capture this same concept when they measured “issue size”. Their conceptualization saw some conflicts to focus on policy issues, that is, issues with high precedent setting potential, while others focused on operational issues, which related to disputes over the performance. Cosar (1956) argues that when conflict is related to issues that do not call into question the basic consensus (goals) of the relationship then the conflict is more likely to have constructive outcomes. However, not all conflicts are positively functional for the relationship, but only those which concern goals, values or interests that do not contradict the basic assumptions upon which the relationship is founded. (1956, p. 80) 4.2 Decision-Making Structure and Constructive Conflict Outcomes Decision-making structure runs the gamut from bureaucratic to clan-like. When the structure is clan-like, decisions will be made in a decentralized, informal, participative structure that shares similar norms. While there is no empirical support in the marketing literature for a positive link between these components of decision making structure 71 o Ara new .- ‘enu 5 Vet ‘F" "a " (1 .uk 5 \v ‘IIA. h V ooa.’ ‘ .. o q A ‘s c Thu: ‘5 - 1. F“! v.‘ ‘ .oa‘ ‘ ‘5‘ "Ju‘ . V ‘l .‘ 8“. n .I ‘— fl ‘4‘. , 3"? ..b“. .0 . N a..‘bl I! c ‘ A gh‘ ‘n: B..\- 1 a “ch and constructive conflict, there are logical connections 'that can be made. Because the parties share similar norms and values there is a greater likelihood that they will rise above the unpleasantness of the conflict episodes and achieve positive change in the relationship. Thus, it is more likely that they will perceive the conflict to be constructive. If a relationship has a centralized decision-making structure then decisions are concentrated in a few individuals. It is unlikely that these few individuals would have the time and resources to make decisions concerning small disputes that occur between the parties. Thus, these type of disputes or conflicts will go unattended and the relationship will be more likely to be characterized by hostility accumulation. Without the hostility release that is a result of the attention paid to these more minor disputes, the conflict in the relationship is less likely to be percieved as constructive. Therefore a more centralized decision-making structure is hypothesized to contribute to ihigher levels of destructive conflict. Mflumn a relationship is governed by formalized rules and the rxxrties put high value in those rules and in a sense are 72 n‘n-rn .- »..Suv \- yng. yo .vu‘..\. o a 1 re ’1': Av- o cit-AV. v apafi~a '- 14 ‘ "nanny, ‘ U {... t t- A V" luv. . v- . . ' A - “ol ‘, .AYAA V .‘i-w: & e,‘ LA "v” ”as F, » V‘h ‘ . a "i ~— 5‘ Q ...» ( ~ I- \. ‘married” to those rules. Therefore, there is a greater chance that change will be hindered since change will often require the changing of the rules and procedures of a relationship. Thus, when conflict resolution requires a change in rules and procedures there is less likelihood that the formalized relationship will (1) achieve this change, and (2) be pleased with the outcome (because any change of rules will be traumatic). Therefore, the highly rule—bound relationship is less likely to perceive conflict that often forces necessary rule adjustment, to be constructive. In a participative structure, decisions are made with multiple inputs. This implies that the change forced by conflict will be perceived to have come from the collective action of the relationship parties, and thus, is more likely to be perceived as positive change. There will be ownership to the conflict outcomes, and that ownership should produce a more positive perception of outcomes. 73 V . " F R I~l‘- \1 . 9: IV U51“- g :‘y‘nhburp ivodyyu. - 'w. IAIO . H5 ht-o...\.\- . Rana.r._'- o b- DUAOV‘L’ _‘G- e .~ ‘A eh. I (1’ (I) u. (D “a p- 9‘ v P s‘ v V:_e a} 5‘ - “Ca-1.. ‘b‘:‘ Q IF -‘V ‘9 . v- " ”Cw- * “s. c 4.3 Operational Integration and Constructive Conflict Outcomes No studies were found that addressed the effect of channel structure variables on the perceptions of constructive conflict. However, logic provides a powerful conceptualization and prediction of the relationship between these two constructs. A benefit of conflict is the release of hostilities before they sunder the relationship (Assael 1969). Highly integrated relationships involve a high degree of joint action and information exchange, thus, the parties should become aware of the problems at an early point. Early detection makes early resolution more likely, and this should lead to the releasing of hostilities. It would be illogical that an intense blow—up would occur because the problem went unrecognized when the firms share large amounts of information and engage in significant joint actions. In an.integrated relationship, the operations, or activities, of the relationship will be more closely :hitertwined between the parties. If this is the case, it :fidllows that the parties have a greater stake in the operations of both firms, thus, they should be more likely to gnarceive the adjustment of those activities to be 74 “vagnfl‘ 3 V" 5.1-Ha» be 7‘ c . . » q . p-purfi V : Utuguil u ' I «not. Drab v Iov* v4.G ' \ I I .F is 4"- F- ‘“ oo.‘\l.-C v - . (32" “R Q Mun-...u .. 9 d . :"‘V“ ’V‘H 'Vvovu... , o. c "'f‘ A. L «ur‘t: . ‘la . ‘ r ‘C‘ tug. ' Vb- BA 0,," ,._YA; 9‘5“..- ‘.4 g \‘u. p .I- ... "‘\D :f‘ ‘u .‘ ‘~ Arr! .“ V ‘ e .- .I ‘w 1 a." I ‘5; 'v 1 “a. '"¢~f}n‘ V ' V s a. ‘9‘.c productive change. Since conflict can often be the impetus behind such change, it is more likely that the parties will view that conflict as producing constructive outcomes. The higher level of assistances means that the firms are not keeping score. They instead have a long—term win—win approach. Thus, even if a productive change resulted from an unpleasant conflict experience (one where a party lost a short term battle), the parties to the relationship would be more likely to recognize the constructive change that occurred due to the conflict instead of tallying up their win—lose record related to the conflict episode. Thus, in this situation, there is a better chance that conflict will be perceived as constructive. 4.4 Constructive Conflict Outcomes and Satisfaction and Trust Satisfaction is an important variable and is often an appropriate proxy for performance (Brown, Lusch and Smith, 1991). Therefore, the relationship between satisfaction and constructive conflict outcomes will be considered. Satisfaction has often been the outcome variable in studies <1f channel conflict (see Brown, Lusch and Smith, 1991 for a nmfixi—analysis of these results). Interestingly, the general relauzionship between conflict and satisfaction is considered 75 8" V‘ 'V :C U “E: I oranogg‘m u..bv~vvfi 5 A ".~r~- h\ onvvo4b. \. ' ~ a'qtv Ab Quito. ‘. h‘r~.y..fi \ vvug,. I", . A6 . . “Lab- 7‘ .5»st ‘ I 'VAQ..AA~ b P ..Vu‘byu I. .. V "Cr ~-‘\'- ha. ix“. ‘4'.“ . \p ”‘ a...»- . . "" Ud\ \- .715. 9‘ , ...H.“' Q ‘ U 'a n. P Q ‘7‘: G ”V u ‘4. . ..‘w ‘ ( A “'V~~’“ V a Vs o a V V , h \ a- s. ~. g ‘5 V. . ‘ u“: MAflfi to be negative, however in this dissertation, the indirect effect of conflict frequency and conflict focus were proposed to have positive indirect effects. Conflict intensity and importance are still proposed to have negative indirect effects. Constructive conflict should be characterized by low levels of pent—up hostilities and a recognition that conflict has produced positive change. These should make the party more satisfied with the relationship’s performance. If the party considers the relationship to have weathered some storms, and subsequently gained in strength, then they should be more satisfied with the relationship. Relationship Trust and Constructive Conflict Outcomes is proposed to have a positive relationship for many of the same basic reasons that are outlined above. Trust is a highly studied construct and is included so that the model has a greater diversity of relationship outcomes incorporated into it. This should strengthen the support for the relevance of Constructive Conflict Outcomes should the model be supported. 76 . ' fiVH;y .6. ‘J“," ...e 06 ‘5‘? “7'; my‘y‘ Us \ ~“V.A,~’ \ Vi u.‘\' 4 n‘y c. v.. ‘ ~‘ ~ . "A a..: 11.- Chapter 3 Research Execution 1. INTRODUCTION In order to test the propositions that were presented in chapter two, a research design was identified and carried out. This chapter reviews that proposed design. The intent of the design was to provide a sound theory test for the model proposed. This test included considerations of the measurement of the model’s constructs, as well as assessment of the effects between constructs (the structural model). The highlights of the design include a cross—sectional survey design that assessed respondent’s perceptions about their firm’s role in an important business relationship. The key statistical method used was structural equation :modeling and the results were interpreted for their lnanagerial and theoretical significance. Two stages of the .research took place: (1) a pre—test designed to assess the rmaasurement properties of the constructs in the model that ‘were newly proposed, and (2) a full test of the model's measurement and structural parameters. 77 '. .- .tnv .u. -»n ...v .- pro: Vow u I 'do a“ vu- u. ' a ‘i l'l The pre-test stage of the research attempted to provide initial support for the reliability and validity of the model's newly proposed constructs. A pretest survey provided the vehicle for assessing the reliability and validity of the measures for these constructs. Exploratory and confirmatory factor analysis, reliability assessments and other techniques were used to establish the reliability, content validity, discriminant validity, and convergent validity of these measures. This stage was essential because seven of the model’s first—order constructs did not have pre-established (through the literature) empirical support for their construct validity. The second stage of the research was designed to assess the measurement attributes and the structural significance of the model proposed. Specifically, the structural hypotheses that are presented in this chapter were tested. Structural Equation Modeling provided estimates of the path model paraneters, and these acted as the test for each hypothesis. The measures of model fit and measures of parameter significance outlined in this chapter provided the evaluative criteria for the inferences made in the final chapter. 78 The discussion of the research execution will be organized to include the following sections; (1) the research design and sampling plan for both the pre-test and the full—model test, (2) construct definition and measurement, (3) The measurement pretest, (4) the structural hypotheses, (5) the rationale behind the statistics employed, and lastly (6) the analysis and interpretation framework. 2. RESEARCH DESIGN This section reviews how a rigorous, and practical research design that empirically tested the measurement and structural relationships, was developed. To this end, the following subjects are addressed below; (1) pretest design issues, (2) full study design issues. 2.1 Pretest Design Issues A pretest was executed to allow for an initial set of measurement tests to be conducted. This section reports the reasoning behind that and the design employed. 79 II' ~' A ‘9 7"". .- .u» nnL‘Mv , In“? tn u».v AK. “’1".v~ f“ ak§vo5§yc . _. w... ““‘“l \- J. ‘-.~ “I ""5 w. . I . “*«-_ . o. ~- ' ‘r'l \- “M. .A‘ a. 5g \ v .... ‘- ‘ .. .. “'5 .._: A ‘- 1‘.:. ‘ l ‘ ‘ A A. .‘..t t .- "» 95“. -‘.. 5. 2.1.1 Objectives of Pretest The model conceptualized in Chapter Two, and operationalized here in Chapter Three, incorporates constructs that had no empirical support for their reliability and/or validity. Thus, the pretest was designed to allow for a limited amount of empirical evidence to be gathered and used to guide measurement improvements prior to incorporating these constructs and measures into the full study. In the end, this process enhanced the measurement success of the full model. The Constructive Conflict Outcomes construct that was not only new in of itself, but the dimensions that were proposed to make up this constructs were also new, thus, this construct became one of the focal points of the pre-test. Its measurement model was presented in Figure 1.4 in Chapter One. Additionally, the “shared paradigm” dimension of clan- like structure was included in the pre—test since there was no empirical evidence for its reliability or validity in the literature. 80 Y. Y. . . . - .. . ‘ .w. 3.. 3. » ... ... ... ... ... .- . - - - :. V.» r. .\~ I: ..s 9. v. .F‘ . . ‘5 .s.. .. .\ ..~ v s z . . ~ 2.1.2 Pretest Sample In order to pretest the measurement model a limited size survey was used. The required respondents for the pretest were individuals that were responsible for managing a key relationship for their firm (same as the full study, see section 2.2.2 below). The desired sample size was approximately 75 respondents. This number was chosen because it would allow for sufficient power for the various measurement purification procedures proposed. The target respondents included business contacts assessable to the author. 2.1.3 Approach Strategy The questionnaire was a one—page document that was designed to be able to be completed within ten minutes and mailed or faxed back. Contacts were approached personally or via ‘phoneeand arrangements were made to fax or mail the pretest tub these individuals. This approach strategy was pursued 'until a sufficient number of responses had been gathered (see Chapter Four for the specific results). Those who respmnnded to the pretest survey were not re—targeted for the full.:study to avoid introducing a history threat to the internal validity of the full study. 81 ‘v o. ' i 2.1.4 The Pretest Questionnaire The questionnaire included all the measures proposed for the dimensions that make up Constructive Conflict Outcomes (CCO). Additionally the measures for Shared Paradigm were included. The specific questions are presented in the appropriate parts of section three of this chapter. Additionally, five single item questions were used to represent other related constructs to allow for an initial assessment of the nomological validity of the model and constructs. A copy of the pre—test instrument is presented as Appendix B. 2.2 Full Study Design Issues The following sub—section reviews the full study design issues and provides the rationale for that design. 2.2.1 General Issues A cunoss-sectional survey design was employed to test the relationships put forth in the model. While this design fail£xi to capture any dynamic change that could characterize the zxalationships in this area of study, it did allow for an :hmitial test of the relationships in a practical and 82 ' Duznr I .vuu.u- I 1' A p oh \ "can” .C-vus ban) 3"‘70‘ AA‘ wlfooolha. ' I ’V’;,"'~ he hum-52», . . 298’. 'NEIE . «a... a \n-V" ‘ .~~~.. ‘ ' ‘ “.’ ‘n .. vs ...e S». ll 1,. A . . an H 1’ n v . aA.' 'r1 relevant manner. As this area lacked significant prior research, and this dissertation represented the first empirical test of some of the proposed relationships, the practical considerations of conducting a relevant theory test were considered very important. Building on this dissertation, a future study can explore the dynamic nature of the supported relationships. 2.2.2 Sample Who? The required respondents for this study were individuals that were responsible for managing a key relationship for their firm. The dissertation collected data from both buyers (purchasing) and sellers (marketing) as both were considered to be relationship managers. Additionally, the position in the channel of the represented firms was not a concern within this sampling plan, thus all levels or tiers of the channel were targeted. Lastly, while it was predicted that the model would hold true regardless of the product characteristics, this dissertation focused on the exchange of tangible goods (versus services) in order to avoid measurement complications. Therefore, the targeted relationship managers were those that manage the exchange of tangible goods . 83 o ‘ ‘ A A ~nn. n' .3} db” '1: was ceszg: :svvsnv- ‘ ...».an . y a. a K). a) v ‘ t (D .. ‘V'r'v-u-u ' \ u..~a-e C I A V . ”‘H‘ Q‘s... ""3 Uevuc . 1 Y; a."\wq 0.. -cw..:,. . .:~QY‘ ‘. ‘ \ .~"'. 5G. A I x311... . ~.u..v.‘ .’ANI| "‘V=: E: Ocj -. “‘5":- ...: :-.; . ‘M‘ ‘IA 2.“; J- ‘ ‘ .FV!» . ‘.. C. "- no . ‘9 ‘N ' A 5 u a“: To achieve a successful sampling plan, an approach strategy was designed to get the questionnaire into the hands of the appropriate relationship managers. These individuals acted as the key respondents for their firms. While there is dispute over the efficacy of a key respondent approach, in this situation it did offer the following advantages; (1) it was practical, and (2) if the respondent was indeed the lead relationship manager, then it was their opinion that this dissertation sought to tap. As the constructs are operationalized as perceptions, it was appropriate to gain those perceptions from the specific relationship managers. Size? Because the method employed to assess the relationships was structural equation modeling, the standard consideration of statistical power was not appropriate. However, Bagozzi and Yi (1988) suggest a minimum of five respondents per free parameter within a SEM design. The ;proposed model had 28 free parameters (if the entire measurement model is supported) and therefore, would require earninimum of 140 respondents. However, Bollen (1989) would suggest a sample size of at least 150 in order to remove any nytential power problems. Thus, to follow Bollen’s advise, auui incorporate a larger margin of error, this dissertation targeted a sample size of 170. A response rate of 10% was assumed (a reasonable response rate considering the job 84 ('9. I. . ( ' . . CD (I) (D D '\ s - ‘FARV ‘ ...‘o g- . v 3‘1'3""'o Nnuy‘vy " Q ‘s-~ ~ . _. I~ N. .K I ..W . I I . n1: _. 5" “~ . FA 1‘ A'C §U~~v.‘ ~‘. in, q -..: eo ‘ l p .....V's _ h- .‘s ‘\-. A ‘ ‘ .‘ . v‘ “. h duties of the target respondents) and thus the approach target was set at 1700 relationship managers. 2.2.3 Approach Strategy A “modified” Dillman (1978) approach strategy was used. To execute this the survey was administered in three stages: (1) an initial survey mailing, (2) a postcard follow—up, and (3) a repeat of the full first mailing. Each of these stages is further discussed below and a copy of the survey instrument is presented as Appendix C. First Mailing. The first mailing consisted of the survey (see Appendix C), a letter requesting participation, and a postage-paid envelope. The survey was designed to enhance response by being visually effective and easy to complete. The letter was on MSU letterhead and requested participation. The respondents were promised a report outlining the results of the study in return for their participation. The return envelope was postage-paid to encourage response . Postcard. Two weeks after the first mailing went out, a postcard was mailed to all sample members. The postcard 85 ‘ . .a "A,” \‘C Lynn .. ‘ “"c‘fif‘n ry- : ~54.“ ‘1' .A" AA». ‘ ~ 0.. ‘Vao b“‘ 7:8: ~..,‘ ....» *c: i ’Q‘»..,‘ 'b‘.u“ ‘ i Q ‘ “M. ..Y V‘ A‘ ‘ II . v...” -(‘P”A :‘vv reminded the individuals that they should have received the first mailer and encouraged their participation. Second Mailing. A revised cover letter accompanied the second questionnaire mailing, however, the remainder of the mailing was identical to the first mailing. The mailing was sent out approximately four weeks after the first mailing. The cover letter indicated that this was a follow—up mailing and thanked respondents who had already responded. With this three—prong approach strategy the target response rate was achieved (see Chapter Four for details). 3. CONSTRUCT DEFINITION & MEASUREMENT This section addresses the proposed measurement of each construct incorporated into the model. The definition of each first—order construct is presented and the measurement history of that construct addressed. Final measures are proposed. When there is a robust measurement history for the construct in question then the final measures most often reflect this pedigree. However, in some cases there is no measurement history for the construct and thus, the measures proposed represent an extension of the construct’s conceptualization. In addition to the text that follows, 86 "fl“:rw'rw < wrung. «V b .. ‘ .‘f‘u‘ifl‘ M n-Uttuu‘ . ‘ . .- .. P‘; me: My. ...». ... uh,“ av;cy‘ ave ‘I‘n‘y. | :v‘ rug~ “HL4 h V, ' - ..“"AY.' .... ....Lq‘ ‘- O 5., \ VAN “'w- 4:: Q ~ A‘FAQ: "':V~H‘ . ‘ s n ,_ h N ‘A' v.‘ . c. 6:. ‘o. . \ y“ v.: F“ 3.1 v§_ "4:" n ‘ v n .‘ 5;.A ' .A in A "“3”: v .- . ~ ~- 5 N . ...tt-C "s .‘ ~ ~__ h» t ‘5‘:- N... \- .4,‘ ‘ s \ ”A s‘ , 7.. U- n,“ 5.52.: sqs construct conceptualization and measurement is presented in Appendix A1. The measures for each construct (both first and second— order) are addressed below. The proposed tests of the quality of the measures presented are addressed later in this chapter. An important note needs to be made. The discussion of the measures proceeds as it was written for the dissertation proposal. Thus, even though the final results were often different, the original language was kept so that the reader could assess the initial reasoning and understand the starting point for the measurement choices made. 3.1 Clan-Like Structure In this dissertation, clan—like structure will be a second- order construct indicated by informalization, decentralization, participation and shared paradigm. Each of these first-order constructs will be directly measured. The measures for each of these first—order constructs will be summed to form construct scores, and then these scores will be used as the measures for clan—like structure. This process will occur with reliability and validity checks at 87 1 ' V PU“! 5" ( LIL-n4 u.h\ . ,. CIA!!! O. f ~~~'v-.a\ h a! . gnu ‘Hfl.’ ' ‘C UI-oott \ Va. .: ‘:-V“-\. .".D:‘ -. Va 0.“:9 0"“, \ . l'y § ...8 T I ..A “LI 2‘ U. . .a ‘A‘ . ._ A“ both the first—order and second-order construct level. The specifics of the first-order construct measurement is outlined below. 3.1.1 Informalization Formalization is defined as the extent to which decision making is regulated by explicit rules and procedures (Dwyer and Oh 1987). In the context of being a component of clan— like decision-making structure, this variable will be reversed in order to capture informalization. Three works provide suggested scales for this construct. First, Dwyer and Welsh (1985) provide four questions that in their study provided an alpha of .78, thus showing adequate reliability. The question posed of the respondents in the 88 p filer a" u («i-(”r V V Jib . ‘i o I: “fla‘ ..w 3L6 .IA : ~ ‘ ~ mt . I . “~- ~ u‘.‘l'\"“n 'H~‘V“ Q s ..Q ‘u “"h L ‘1‘.“ i- .I D “ n n... «A a e ‘v- Q q | ‘- A 4. - ~- “ .‘ A I-.‘ ‘. 4" A'fi‘ ‘H. s \H p Dwyer and Welsh (1985) study was as follows: “Thinking about your relationship with suppliers, would you say...” (p. 412). 1. You follow strict operating procedures. 2. You have a high regard for existing rules and procedures. 3. Your responsibilities are clearly specified. 4. You follow previously written and verbal instructions. The scale for these questions was a five point scale with the following categories, (1) never, (2) seldom, (3) occasionally, (4) rather often, and (5) nearly all the time. To add to this set of measures, John (1984, p. 287) provides the following sample items: 1. My dealings with the supplier are subject to a lot of rules and procedures stating how various aspects of my job are to be done. 2. My contracts with the supplier and his representatives are on a formal, pre-planned basis. John (1984) used a 5 point Likert scale with anchors of strongly agree and strongly disagree. The author reported 89 an internal consistency coefficient of .63, thus his scale showed poorer reliability than the Dwyer and Welsh scale. Lastly, Reve and Stern (1986) provide these two additional items. 1. The relations between the supplier and our firm are governed by written contracts. 2. Complaints and returns from the retailer to us are handled through standard procedures. Their scale was an 8—point scale that assessed the truth of that description concerning the relationship. The authors reported a Cronbach’s Alpha of .701 for their five—item scale. 90 is . . . . a 2‘ «.4 at Nu m .. .2 J . a: -~ - .u a a .4. it . . N v un u {bin . V Based on these studies the following combination of items will be used to assess formalization: Considering the relationship you identified, please respond to the following statements: 1. When decisions are made concerning this relationship, the parties rely on clearly specified operating procedures. 2. Existing rules and procedures are often ignored when relationship decisions are made (R). 3. The parties refer to previously written and/or verbal instructions to guide relationship decision making. 4. Formal, written contracts guide relationship decision making. 5. When faced with a relationship decision, the parties fall back on standard courses of action. 6. When faced with a relationship decision, the parties rely on behavior dictated by formalized roles and responsibilities. The scale for these questions will be a seven point Likert scale anchored by strongly agree to strongly disagree. 3.1.2 Decentralization Dwyer and Oh (1987) define this as “the extent that decision—making authority is concentrated” (p. 349). Within this dissertation an important distinction will be made between centralization and participation. The distinction 91 . vs h-R a: MOD! 5...... quv‘ ‘5 new y: vHo.y.u RA'RA' : . u- want... vi 1'»; f nut" Mk W . *‘v. ,. r- own. AA...“ vy,." RA,“ v..v ‘-‘ I “Q ~. \A V- is based on the research of Hage and Aiken (1967). These authors carefully distinguished between participation and centralization. When decision making authority is concentrated in a few individuals, this is centralization. When decisions are made based on the input of both parties to a relationship, this is participation. Prior to reviewing Hage and Aiken’s (1967) conceptualization of centralization, other research works will be reviewed. These works measured “centralization”, but not in the manner conceptualized in this dissertation. Dwyer and Welsh (1985, p. 412) provide the following items to measure centralization: (The question posed was “Thinking about your relationship with suppliers, would you say....”) 1. You go ahead with actions without checking with your supplier. 2. You refer marketing matters to your supplier. 3. You yield to the recommendation of your supplier. 4. You rely on your suppliers for an answer. Imnfortunately, while these measures showed strong reliabdlity, they do not capture this dissertation’s conceptualization of centralization. They are more 92 :ny‘y/‘nr- Ubh. VU‘ . O. . (xi) ‘- (I Q . J ”v- *. n‘P‘v.Y .- ~‘.~; “M appropriately participation measures. This distinction will be addressed later in this section. Reve and Stern (1986) also measured centralization. They defined this construct as “the perceived degree of influence on dyadic actions” (p. 82). Two sample items they included are: 1. Our advertising campaigns for the supplier’s products are determined in detail by the supplier. 2. The retailer determines which ordering procedures he is going to use. Again, this conceptualization matches this dissertation’s concept of participation. Thus, these measures will not be replicated for this dissertation. Lastly, John (1984) also provides two sample items for his centralization scale. This scale achieved a reliability of .79 and was conceptualized to be measuring a component of “bureaucratic structuring” of an exchange relationship. 93 n‘ 9 p... . “.431 U a ' | ”‘53. (hp; “.01.“... hi) . n"- I"‘-u (I) J)- (h 1): Thus, John's conceptualization is closely analogous to Robicheaux and Coleman’s (1994). 1. I have to ask my supplier's representatives before I do almost anything in my business. 2. In my dealings with my supplier, even quite small matters have to be referred to someone higher up for a final answer. The second question gets directly at the concept of centralization and an analogous question will be used in this dissertation Hage and Aiken (1967) consider centralization to have two components. The first component is akin to participation, and is not the operationalization that this work will use as participation is considered a separate construct. However the second component is defined as “how power is distributed among social positions” (Hage and Aiken 1967, p. 78). It is this second component that captures the concentration of decision—making authority in a few individuals in the relationship. 94 The measures suggested (and supported) by Hage and Aiken (1967, p. 78) include: 1. There can be little action taken here until a supervisor approves a decision. 2. A person who wants to make his own decisions would be quickly discouraged here. 3. Even small matters have to referred to someone higher up for an answer. 4. I have to ask my boss before I do anything. 5. Any decision I make has to have my boss’s approval. Based on the conceptualization outlined above and incorporating the measures presented above, measures of centralization are proposed below. Considering the relationship you identified, please respond to the following statements: 1. Within this relationship, decision—making authority is concentrated in a few individuals. 2. If a decision needs to be made, the decision is often made by those directly involved in the situation, and does not need the approval of superiors (R). 3. A small percentage of the individuals involved in this relationship set the relationship policies. 4. Superiors must review all the important decisions before the decisions take effect. 95 5. The frontline employees involved in managing the relationship are free to make decisions involving this relationship (R). 6. There can be little action taken until relationship decisions have been approved through proper channels. The scale for answering these questions is a seven—point Likert scale with anchors of strongly agree and strongly disagree. 3.1.3 Participation Hage and Aiken (1967) develop the concept of participation as a sub—component of centralization. However, authors in marketing have considered it a separate component from centralization (Dwyer and Welsh 1985; Dwyer and Oh 1987). Ihnyer and Welsh (1985) define participation as “the degree of actual participation in decision making” (p. 400). Thwyer and.Welsh (1985) employ a slightly different operationalization, in that, they also have the following set.<3f questions to measure participation: 1. You play an active role in decision making 2. Suggestions from you are encouraged. 3. Supplier decisions are made without you. 96 Additionally, Dwyer and Oh (1987) offer the following sample item from their participation scale. 1. My ideas for ordering, selling, and servicing are welcomed by the manufacturer. Question #2 from Dwyer and Welsh (1985) and #1 from Dwyer and Oh (1987) both deal with ideas and suggestions being welcomed in the decision making process. This is an indirect method of assessing participation. As long as the assumption is made that the collection of suggestions and ideas is a fundamental part of the decision—making process, then these measures would capture participation because the offering and welcoming of ideas would constitute “actual participation”. However, it can be argued that the welcoming and encouraging of ideas does not constitute real participation as the decision—making process really is focused on the evaluation and selection of ideas and suggestions. In this case, a party could contribute ideas, but still not be directly involved in their evaluation. This case seems to represent “partial” participation. 'Ehe measurement strategy employed by Hage and Aiken (1967) offers a way to avoid.this potential measurement confusion. Their strategy involves identifying areas of “organizational" decision making and directly asking about 97 ‘ 5 :3e .req. *"s ”Fe“ PAY. ‘ In. “w. VG. b.\-.~I ‘ s p 3? “Rina Vt Q‘s.“ .( - "YD” “E: ~‘- s... a” fancy "‘"““I” »\ ‘ 4 ~ I A;- ~,, r— ~t.vfi' IA . , [A 2‘ ’1 ”Vogy‘u u 9 P ...e if: 5... “20“.. the frequency of participation in those decisions. Since this method captures the spirit of the definition of participation as defined by Dwyer and Welsh (1985) (“degree of actual participation” (p. 400)), and avoids the potential measurement confusion outlined above, this method will be employed. The specific questions proposed are presented below. “Consider the following decision making areas, and indicate the amount of mutual participation that is present in each decision in the relationship.” (7 point scale anchored by: Decision is made entirely by one party and Decision is made equally by both parties) 1. Delivery arrangements and terms 2. Product/service design changes 3. Length of contract/agreement 4. Ordering procedures 5. Product/service prices and/or margins 6. Scheduling of operational activities 7. Demand and/or supply requirement forecasting The scores from each of these decision areas will be combined to form an overall participation score. 98 j i I A v t. R vac v . u a 3.. a» fl: ~\U ~.\~ . . .. . a: L. G. .n .. C; v . .6! a . ~ ... n h. r“ "I ‘th ‘ ‘ A \ ‘u 9 5‘4 3 . 1 . 4 Shared Paradigm Shared paradigm considers whether the exchange partners share general assumptions and values related to conduct in the relationship (Robicheaux and Coleman 1994). It is the paradigm that they share that helps the participants determine what is in the best interest of the relationship. This construct has no empirical history and thus the measures suggested below are based on the conceptualization presented here and in chapter two. These measures will be assessed in this dissertation’s pretest. A seven—point scale with anchors strongly agree and strongly disagree will be used. Considering the relationship you identified, please respond to the following statements: 1. The parties often agree on the appropriate course of action for this relationship. 2. Both firms share the same values. 3. The parties often disagree on what is in the best interest of the relationship (R). 4. The parties have different expectations for this relationship (R). 5. The parties often agree on what constitutes appropriate relationship behavior. 99 0“ Uh. “'_R‘\r U \r‘v. 3.2 Operational Integration In this dissertation, operational integration will be a second-order construct indicated by joint actions, assistances, monitoring and information sharing. Each of these first-order constructs will be directly measured. The measures for each of these first—order constructs will be summed to form construct scores, and then these scores will be used as the measures for operational integration. This process will occur with reliability and validity checks at both the first-order and second—order construct level. The specifics of the first—order construct measurement is outlined below. 3.2.1 Joint Actions The construct, Joint Actions, is defined by Heide and John (1990) to represent the “degree of interpretation of organizational boundaries" (p. 25). Heide and John (1990) measure joint actions by identifying areas of operations that would be typical in a relationship and then assessed the level of joint actions in the relationship by asking the respondents to indicate whether there was “minimal joint action" or “extensive joint action” for that operational activity. Since there is no other available literature to 100 . aehan. ..wvuvi ~ ~ ...... v‘I-‘u . .N’ 5;... \ b...‘ or 1 (I: (I) 1 V . I‘V“ ‘ 5 1 ‘V‘ ‘ n .A! \ VA J‘ (1 “A 1 kn .. ‘:\"A VV' W \ V guide measurement on this construct and the measurement method employed by Heide and John showed strong reliability and validity, this method will be employed in this dissertation. To operationalize joint action via this method the first step is to identify a set of operational activities that would characterize the typical relationship. Through discussions with some relationship managers and calling on past experience the following list has been generated. 1. forecasting 2. component / product testing 3. personnel training 4. product design and development 5. promotional activity 6. inventory control Based on this list, the respondents will asked to assess the list based on a seven point scale with the anchors “no joint action" or “extensive joint action”. These anchors have been slightly modified compared to those used by Heide and JOhn (1990) in order to allow for a “no joint action” reSponse. 101 AArr I'VVHV. 5 4A. U‘v Is,‘ 3.2.2 Assistances Noordewier, John and Nevin (1990) define assistances as the giving of help to an exchange partner even when such behavior is not specifically called for by the preset standards of the relationship, and/or, when there is no immediate compensation given in return for that help. Based on this, the measures used by Noordewier, John and Nevin (1990, p. 92) are appropriate for this dissertation. l. Supplier calls in advance to advise us of shipment problems. 2. Supplier makes an effort to help us in emergencies. 3. Supplier recommends stock substitutes when delivery troubles develop. 4. Supplier helps us in value analysis ideas, cost reductions, problem solving, etc. 5. Supplier advises us of potential problems in meeting our needs. ZNoordewier, John and Nevin (1990) showed strong support for this construct. It was assessed as a first—order construct .and.each of the measures above loaded significantly on the eassistances construct. This study did not report .reliability. There are no other uses of this construct in tflne literature, thus there are no other measures to draw front With this in mind the measures above will be used 102 ( g - ”no 9| a». VI flAflinp: hwacs‘u I by. ...: Ln: ...... . "v (I) t f '- s 9.. (with the appropriate adjustments. The final proposed Jneasures are listed below. Chansidering the relationship you identified, please respond t1) the following statements: 1. The parties in this relationship advise each other in advance of potential operational problems (such as delivery delays or stock outs). 2. The parties help each other in emergencies. 3. Unless payback is immediate, help is not offered to the other party (R). 4. The parties help each other in value analysis, cost reductions, problem solving, etc. 5. The parties advise each other of potential problems in meeting each other’s needs. IFor these questions a seven—point Likert scale with anchors :Strongly agree and strongly disagree will be used. 3.2.3 Monitoring 1Voordewier, John and Nevin (1990) define this construct to Capture the monitoring or supervisory actions that the Iparties undertake to ensure partner performance during the execution of the exchange relationship. The authors contend that a more relational realization of monitoring will 103 . Q . "'95 034' #5 . npv'n 11v. .v . .. R ~\ Vs. .y. involve “active supervision ... to ensure specified performance" (p. 84). qflie measures used by these authors are as follows (p. 92): 1. We advise supplier of its performance in relation to that of other suppliers. 2. We monitor supplier’s inventory levels. 3. We assess supplier’s performance through a formal vendor evaluation program. 4. In this arrangement, supplier must provide summary usage reports, tally sheets, or some similar kind of report (on a quarterly or monthly basis). 5. We conduct quality training for vendor personnel. 6. The relationship we have with this supplier makes use of many controls. Tluese measures all loaded significantly on the monitoring annstruct and thus showed convergent validity (Noordewier, Jtflni and Nevin 1990). No reliability results were reported ill this study. As the Noordewier, John and Nevin (1990) Stnmdy is specific to purchasing arrangements, some adjustments of these measures is necessary. 104 ... 5‘). .l r AVA INC. : 1 At ..SS {1) e (I) f.) Therefore the following measures will be used in this dissertation in order to capture monitoring. Based on the relationship you have with this party please respond to the questions below: 1. The parties spend little time checking on each other’s performance (R). 2. The parties monitor each other’s inventory levels. 3. The parties assess relationship performance through a formal evaluation program. 4. The parties share performance reports regularly. 5. The relationship makes use of many controls. 3.2.4 Information Sharing As outlined in chapter two, there are three areas of information exchange that require operationalization in this dissertation. First, the frequency of information exchanged is considered important as high frequency of information exchange indicates higher levels of operational integration (Robicheaux and Coleman 1994). Second, the type of inrformation exchanged is relevant. Noordewier, John and lflevdji (1990) predict that more strategic and planning iJrfornation will be exchanged when the relationship is 105 ‘A a ~‘v‘....C ‘f‘w; y“vn.~ q a...» ‘V‘F 5,. . characterized by a high level of operational integration. Lastly, it is logically consistent that highly integrated relationships will be characterized by high level of capability to exchange information in a timely and useful format. The three sets of questions (after measure purification is completed) will be combined to form an overall assessment of the level of information sharing in the relationship. That combined score will be used as an indicator of operational integration as conceptualized by Robicheaux and Coleman (1994). The scale used for all the questions below is a seven-point Likert scale anchored by strongly agree and strongly disagree. Frequency of Information EXchanged. Measurement of the frequency of information exchange will be done with the following three items. 1. Communication between the parties happens often. 2. The parties frequently share information. 3. Information / data is exchanged infrequently (R). Type of Information EXchanged. Noordewier, John and Nevin (1990) offer the following measures that attempt to capture 106 :Ite t p. 0. , e-se\ Anna. vVuuu .¢‘~ (I..- ‘ fie.“ DH“ . "A, .N a”; e.A u..." s p h 'vE g \F“ ~..:: the type of information exchanged. The measurement effectiveness of these four items was supported by a CFA conducted by the authors. 1. We give the supplier usage information to help him plan for our needs. 2. We keep our supplier informed of production plans. 3. We regularly provide supplier with long—range forecasts of supply requirements. 4. We inform the supplier in advance of impending design changes. While these measures were supported, the domain of this study is different, thus some changes are needed to make these measures appropriate for this study. Specifically, the measures will become more generic and not directly refer to a “supplier”. The following measures of information sharing type are offered. 1. Strategic information is readily shared in this relationship. 2. Production plans are readily shared. 3. Long—range forecasts are readily shared. 4. Impending design changes are readily shared. 5. Financial planning information is rarely shared in this relationship (R). 107 Exchange Capability; Bowersox, et al. (1995) used the frequency of the use of EDI, bar—coding and real—time communication systems to capture connectivity. This approach was relevant for that research, however, in this research the measures need to be more generic and thus are based on the definition of this concept. 1. Information shared between the parties is easily understood and used by both parties. 2. Although information is shared between the parties it is often unusable because it is in a format that cannot be understood by the other party (R). 3. The parties are able to exchange data or information easily and in usable formats. 3.3 Constructive Conflict Outcomes Since Constructive Conflict Outcomes is a newly created construct, measurement for it arises from its conceptualization and not past empirical work. Based on the conceptualization laid out in chapter two, three dimensions of Constructive Conflict Outcomes will be operationalized. First, the release of hostilities will be measured. Second, the adjustment of norms will be measured. And finally, the adjustment of activities will be measured. It is anticipated that these three dimensions will form the higher order construct of Constructive Conflict Outcomes. This 108 1 n r n: V “Ssut ”IQ.".‘ will be explicitly tested when the measurement model is considered. Below, the measures proposed for each dimension are presented. 3.3.1 Hostilities Release Cosar (1956) claims that a constructive outcome of conflict is that it allows the system members to release hostilities before those hostilities build up and lead to a destructive blow—up. Assael (1969) found that hostility release was a benefit of conflict in an automotive distribution channel. Unfortunately, Assael did not operationalize this component, thus the questions proposed below represent a logic driven operationalization, not one based on past literature. Based on the relationship you have with this party please respond to the questions below: 1. The relationship conflict has often cleared the air between the parties. 2. The conflict that occurs helps to defuse the more serious and destructive conflict. 3. The conflict in this relationship has gotten to be too extreme to handle (R). 109 , .3sz \IJ.,‘~ . i.“ 44 . ""1" . vvu“ . IAN-HQ a“ ‘ . & “A'Hs 1.‘ \ b.‘~ b Jun. ——_—___ 4. Conflict is generally over issues that this relationship can easily handle. 5. Conflict helps the parties release hostilities before they become too volatile. 3.3.2 Norm.Adjustment Cosar (1956) sees conflict as a mechanism by which new rules and norms are created in the relationship. Because changing conditions require changes in the relationship, the adjustment of norms caused by conflict, becomes a constructive function of conflict. The measures below are theory based and will be explicitly tested when the :measurement model is considered. Based on the relationship you have with this party please respond to the questions below: 1. Conflict between the parties has rarely resulted in beneficial policy changes (R). 2. Beneficial policy adjustments have been made in response to conflict. 3. Conflict has forced the parties to rethink and adjust the relationship norms of behavior. 110 “:2‘,‘ ‘ ‘5‘; v 4. When changing conditions have caused friction, the parties have been able to make the necessary adjustments to restore balance to the relationship. 5. In response to the relationship conflict, new agreements have been forged as to what constitutes acceptable relationship behavior. 3.3.3 Activity Adjustment Deutsch (1973) considers conflict to act as the impetus behind creative change. He considers conflict to provide the motivation for change, and thus play a constructive role. Assael (1969) found that the reassessment of relationship activities was indeed a constructive outcome of channel conflict. The measures below are based on this conceptualization and do not arise from an empirical history. Based on the relationship you have with this party please respond to the questions below: 1. Beneficial operational changes have been made in response to the relationship conflict. 2. Conflict often makes the relationship less effective (R). 3. Conflict has pushed this relationship to higher performance levels. 111 4. Conflict has made us review past actions and make improvements. 5. Conflict has forced the parties to come up with innovative solutions to relationship problems. 6. After conflict has been resolved, the relationship returns to the status quo without making the changes required to avoid the reoccurrence of the conflict (R). 3.4 Conflict Ccuiflict is assumed to occur in any relationship that is cluaracterized by interdependence between the parties to the Inelationship. However, as Brown and Day (1981) illustrated, ccniflict measurement is best approached by considering rmiltiple facets of conflict. In this case, the frequency, :hitensity, importance and focus of the conflict will be measured. These characteristics of conflict are predicted to have effects on the level of constructive conflict according to the predictions of social conflict theory. Thus, although Brown and Day (1981) suggest a multiplicative combination of these characteristics, this dissertation will consider these characteristics separately. 112 IF ‘4’“, Via... . ”Hwy/- bbu.\ ' . ‘Fl‘. ._ ...‘fil ‘ "A. “eta! FA“. e.‘ ~P 4,.“ 3.4.1 Conflict Frequency, Intensity & Importance Sindlar to the approach of Brown and Day (1981) the :Erequency, intensity and importance of conflict will be cuperationalized by identifying issues that are standard scnurces of conflict and then measuring the characteristics off conflict for each issue. A combined sum of these irnflividual measures will provide the final conflict frwaquency, intensity and importance measure in this di 5 sertation . The: first step of this procedure is to identify a set of aremas within the relationship that could be common sources of cnonflict. Using past studies that operationalized cxnrflict in a similar manner (e.g. Brown, Lusch and Smith 1991.; Schul and Babakus 1988; Brown and Day 1981; Lusch 197Gb» combined with the sources of conflict approach (Rosenberg and Stern 1970; Etgar 1979) the following list of potential conflict areas is proposed. 113 Q 10. ll. 12. 13. 14. 15. l6. 17. 18. 19. 20. 21. Minimum order quantities Payment terms Product quality Level of sales support Level of service support Advertising support Performance objectives Terms of the contract Inventory policies Warranty policies Delivery terms Delivery performance Communication of problems Performance expectations Frequency of communication Strategic objectives Control of resources Sharing of market information Design issues Price and/or margins Ease of communication H4 Frequency. For this list of twenty items the respondents will be asked to identify the frequency of conflict related to each issue based on a 7—point scale with the following anchors; never and very frequently. After checking for the internal consistency of these items and deleting the issues that fail this test, the frequency scores for the remaining issues will be summed to form a final measure of conflict frequency. Intensity. Respondents will also be asked to identify how intense the conflict is on each issue. A seven point scale with the anchors, not very intense and very intense will be used to tap this construct. A sum of the individual issue intensity scores will be used as an overall intensity estimate. Immmutance. Based on the issues identified above the respondents will be asked to identify how important each issue is within the context of the relationship. A seven point scale with the anchors, not very important and very important will be used. An overall conflict importance scale will be calculated by summing the individual issue 1mPortance scores. 115 3.4.2 Conflict Focus Conflict focus is designed to capture whether the conflict in the relationship centers mostly on operational issues or on core or strategic issues. Social conflict theory proposes that conflict will have greater constructive outcomes when conflict is not focused on core or central issues. There is no literature support for this construct, and thus the following measures are based on the construct’s conceptualization. Based on the relationship you have with this party please respond to the questions below: 1. The conflict in this relationship has focused mostly on operational issues. 2. The parties have disagreements over fundamental issues (R). 3. Conflict in this relationship centers on how to achieve goals, not the goals themselves. 4. Much of the conflict centers on disagreements over the basic goals of the relationship (R). 5. The relationship is characterized by strong disagreements over strategic issues (R). 116 3.5 Relationship Satisfaction Conflict and satisfaction have often been considered in conjunction. (see Brown, Lusch and Smith 1991 for a review of these findings). Relationship satisfaction will be captured using the following questions. (seven—point Likert scale) Based on the relationship you have with this party please respond to the questions below: 1. We are very satisfied with this relationship. 2. This relationship meets or exceeds our expectations. 3. This relationship has performed poorly (R). 4. Based on the objectives we hold for this relationship, we are very satisfied with the performance of this relationship. 5. This relationship has not performed up to our expectations. (R) 117 4. STRUCTURAL HYPOTHESES There are eight structural hypotheses. The relationships were illustrated in Figure 1.1 contained with Chapter One. The hypotheses correspond to the logic laid out in Chapter Two. Support for the hypotheses constitutes support for that logic, given the way the logic was operationalized. This section is organized by reviewing each of the specific hypotheses being offered. The specific hypothesis will be preceded by a brief review of the theory and/or logic that supports the hypothesis. U8 4.1 Conflict Frequency and Constructive Conflict Outcomes Social conflict theory indicates that high conflict frequency is likely to produce constructive conflict (Turner, 1986). The reason for this connection is that frequent conflict will allow the relationship parties to release hostilities before they polarize the parties. Thus, the following hypothesis is put forth: H1: Conflict frequency (CFREQ) will have a significant and positive effect on constructive conflict outcomes (CCO). 119 4.2 Conflict Intensity and Constructive Conflict Outcomes Social conflict theory defines intensity to include people’s involvement in, and commitment to, pursuing the conflict. SCT predicts an inverse relationship between conflict intensity and constructive outcomes (Turner, 1986). Constructive outcomes are considered to be more likely when conflict intensity is low because the parties will stay more focused on realistic issues, the parties will communicate more effectively and the parties will be more likely to allow norms of conflict resolution to work effectively. Therefore, the following hypothesis is put forth: H2: Conflict intensity (CINTS) will have a significant and negative effect on constructive conflict outcomes (CCO). 120 4.3 Conflict Importance and Constructive Conflict Outcomes Social conflict theory postulates that low violence conflicts will be characterized by more constructive outcomes (Turner, 1986) and conflict issue importance is closely connected to SCT's concept of conflict violence. Cosar (1956) defines conflict violence as the degree to which the parties are seeking to injure or eliminate the other party. Conflict violence is said to rise when the parties are fighting over a centrally important issue because their emotional arousal and involvement related to the issue will be increased (Turner, 1986). Assuming this relationship holds true allows us to make the inference that when conflict is over more important issues, the conflict will tend to be more violent and then the likelihood of constructive conflict would be reduced. 33: Conflict importance (CIMP) will have a significant and negative effect on constructive conflict outcomes (CCO). 121 4.4 Conflict Focus and Constructive Conflict Outcomes The focus of the conflict captures whether the conflict is centered around different goals or different methods of achieving similar goals. In the context of this dissertation, a high level of conflict focus will indicate that the respondent perceives that the conflict in the focal relationship is not over the strategic goals of the relationship, but is instead, over the operational choices designed to achieve common goals. Cosar (1956) indicates that this is an important distinction when he states “conflicts arising within the same consensual framework are likely to have a very different impact upon the relationship than those that put the basic consensus in question” (p. 73). Cosar (1956) argues that when conflict is related to issues that do not call into question the basic consensus (goals) of the relationship then the conflict is more likely to have constructive outcomes. H4: Conflict focus (CFOC) will have a significant and positive effect on constructive conflict outcomes (CCO). 122 4.5 Decision-Making Structure and Constructive Conflict Outcomes Decision—making structure runs the gamut from bureaucratic to clan—like. When the structure is clan-like, decisions will be made in a decentralized, informal, participative structure that shares similar norms. These characteristics form the basis for why clan-like decision making structure should lead to higher levels of constructive conflict. Specifically the following should occur; (1) the sharing of similar norms and values will improve the chance that conflict outcomes will be agreed upon, (2) the informal structure will be less rigid in the face of conflict, (3) the participative nature of decision making will help create perceptions of positive change, and (4) the decentralized structure will allow small grievances to have a forum for resolution. These factors lead to the following hypothesis: H5: Clan-like decision-making structure (CLAN) will have a significant and positive effect on constructive conflict outcomes (CCO). 123 4.6 Operational Integration and Constructive Conflict Outcomes High levels of operational integration would be characterized by the presence of significant joint actions and information exchange. These attributes should bring problems to the attention of managers quickly, and thus there is a lower likelihood that the hostilities brought forth by the conflict will become unmanageable. Additionally, since the parties have a high level of assistances they should avoid a scorecard mentality and should recognize positive change even when it has been induced by conflict. Lastly, the integrated relationship should exhibit a greater concern for the joint operations and thus will make the parties more likely to seek out and recognize productive change. Based on the logic outlined above, the following hypothesis is made: H6: Operational integration (OPINT) will have a significant and positive effect on constructive conflict (CC). 124 4.7 Constructive Conflict Outcomes and Relationship Satisfaction Satisfaction is an important variable and is often an appropriate proxy for performance (Brown, Lusch and Smith, 1991). Constructive conflict will be characterized by low levels of pent-up hostilities and the recognition that conflict has produced positive change. These should make the party more satisfied with the relationship’s performance. If the party considers the relationship to have weathered some storms, and subsequently gained in strength, then they should be more satisfied with the relationship. Based on this logic, the following hypothesis is made: H7: Constructive conflict outcomes (CCO) will have a significant and positive effect on relationship satisfaction (SAT). 125 4.8 Constructive Conflict Outcomes and Relationship Trust The inclusion of trust as another relevant outcome variable was done because Trust is a highly considered construct in the relationship marketing literature. The proposed relationship is positive because it is assumed that the “local” positive outcome (CCO) would be directly associated with the more “global” positive outcome. Therefore, the following hypothesis is made: H8: Constructive conflict outcomes (CCO) will have a significant and positive effect on relationship trust (TRUST) . 126 5 . STATISTICAL ANALYSIS The analysis employed in this dissertation was structural equation modeling (SEM). This section is designed to lie out the criteria that will be used to determine if a specific model has been confirmed or disconfirmed by this statistical methodology. The criteria will be applied to two classifications of models: (1) the measurement models, and (2) the structural model. Those two classifications will be considered below. 5.1 Measurement Model The measurement models were assessed separately from the path model. Additionally, the original, full measurement model was very complex. Four of the constructs incorporated in this model had multiple dimensions that had multiple measures. Thus, the intention was to test the model at the first-order level, then sum the measures of the dimensions and have those calculations act as the dimension score. This, in effect would allow for the second-order measurement model to be actually tested as a single-order model. This was desirable so that the overall model complexity could be reduced 127 The plan was to initially conduct any measurement model assessments using traditional statistical procedures such as Cronbach's alpha and exploratory factor analysis (EPA). This would be followed by the use of confirmatory factor analysis (CFA) procedures. It is the confirmatory factor analysis procedures that are addressed below. 5.1.1 Measurement Model Fit There are two basic procedures that were incorporated into the design in order to assess CFA model fit: (1) assessment of preliminary model fit, and (2) statistical and other assessments of model fit. These assessments are both recommended by Bagozzi and Yi (1988). They are discussed below. Preliminary Fit. According to Bagozzi and Yi (1988) the mOdel should not exhibit any of the following: 1. Negative error variances 2. Factor loadings that are smaller than .5 or larger than .95 3. Very large standard errors. 128 The preliminary fit of the measurement model was assessed against these criteria. Failure to meet these criteria often indicates identification or specification problems (Bagozzi and Yi 1988). Statistical and Other Assessments of MOdel Fit. The second assessment that was made was the overall fit between the proposed measurement model and the sample data. Since there is no entirely conclusive way to assess model fit, a series of tests were considered as recommended by Bollen (1989). Listed below are the tests and test values that were used to assess model fit. The majority of these are based on the recommendations of Bagozzi and Yi (1988). 1. A non—significant chi—square value 2. A BBNNFI value of .85 or greater 3. A RMSEA value of under .5 4. A CFI of .85 or greater 5. No standardized residual greater than 2.0 ‘Bhere1was a high probability that the models tested would :notLIneet the criteria on all of these tests. Thus, the final ccun21usions made concerning the fit of the various 129 measurement models were made based on a preponderance of evidence and not a single test. 5.1.2 Construct Validity Tests There are two main assessments that were made. First, the convergent validity of the model’s measures was assessed. Second, the discriminant validity of the constructs was considered. These are discussed below. Convergent validity. Defined as “the confirmation of a relationship by independent measurement procedures” (Churchill 1979, 1991 ), convergent validity is sufficient support for the internal consistency of a construct (Anderson and Gerbing, 1982). When using SEM, if the factor loadings of the indicators on their posited constructs are significant, then convergent validity is supported (Anderson, 1987). Discriminant validity. Peter (1981) considers discriminant validity to be supported when a measure does not correlate Very highly with another measure from which it should differ. Discriminate validity supports external Consistency, that combined with the convergent validity - internal consistency confirmation from above, will support 130 the unidimensionality of the construct under consideration (Anderson and Gerbing 1982). Using SEM, the manner in which this dissertation assessed discriminant validity was the following: using a pairwise series of nested CFA models the discriminant validity was assessed by considering if a model that had a pair of constructs with their covariance constrained to one had worse fit than the model with no constraint (using a chi—square difference test). If this was the case then the disciminant validity of the constructs was supported (Anderson 1987). 5.1.3 Conclusion Before the structural tests proceded the measurement models needed to pass the tests outlined in the sections above. Chapter Five documents the actual process of applying those criteria to the measurement models. Models that failed to gain support were either eliminated or altered in theory consistent ways and re—tested against the criteria. Any model that was respecified is explicitly identified as such in Chapter Five. A.two-stage assessment approach was used as recommended by Burnkrant and Page (1982) and Anderson and Gerbing (1988). ThUS. the final confirmed measurement model was not 131 explicitly re-assessed while the structural model was assessed. 5.2 Structural Model A structural, or path model was examined using SEM with the purpose of assessing the quality of the proposed model. Two main considerations addressed were: (1) the overall fit of the model, (2) internal structure fit. These are discussed below. 5.2.1 Structural Model Fit. The same model fit standards that were applied to the Ineasurement model were relevant for addressing the structural model fit. However, there are the additions .6 and no cross—loading over .4) consisted of the expected items and could correctly be labeled as (1) Integrative Conflict Resolution Approach (ICRA) and (2) Constructive Conflict Outcomes (CCO). Table 4.1 below documents these results. Table 4.1 - Pretest EFA Results Factor Factor Item 1 2 Common Goal 1 .721 .177 Mutual Interest 2 .765 .179 Mutual Interest 3 .849 .121 Positive Sum Orientation 2 .789 .002 Positive Sum Orientation 2 .811 .191 Activity Adjustment 3 .252 .722 Activity Adjustment 4 .193 .649 Activity Adjustment 5 .006 .803 Hostility Release 2 .340 .647 Norm Adjustment 4 .000 .820 M0 2.2.3 Initial nomological Validity A correlation matrix of the principle variables was run. This matrix showed support for a positive and significant relationship between the respecified ICRA construct and the respecified CCO construct. It also showed a positive relationship between CCO and both Satisfaction (3 items with alpha = .9556) and Trust (single global measure). These results all support the hypotheses identified in the proposed structural model. These results lend nomological validity support to the project but fall far short of supporting any causal paths. That task will be left to the full test with the full data set. 2.3 Pretest Results and Implications The pretest results were a mixture of support for the measurement model proposed. While in general, the constructs were supported, the exact proposed specification of those constructs was not. The data indicated that certain respecifications would be appropriate. The pretest data analysis suggests two significant respecifications. First, it suggests changing the three component (factor) Constructive Conflict Outcomes construct to a single factor 141 construct. This new construct would contain questions representing all three proposed sub-factors of the original second—order construct, but not possess the proposed measurement model complexity of the original. Second, a similar change would be suggested for the construct titled Integrative Conflict Resolution Approach. The suggested respecification would also be to make this second-order construct into a single factor construct with questions from each of the original three components. These two new single factor constructs (CCO & ICRA) were shown to be distinct via an exploratory factor analysis (no cross loading over .4) that included all the questions from both of the suggested constructs. Both of these simpler constructs also showed adequate reliability. In addition to these measurement issues, the pretest provided support for the nomological net that forms the basis for the proposed structural model. The proposed associations between Constructive Conflict Outcomes and Integrative Conflict Resolution Approach, Relationship Satisfaction and Relationship Trust were all supported via a simple correlation analysis. 142 Due to the results of the pretest, changes were made in the final set of questions asked on the dissertation questionnaire. Although the original three-factor, second- order models of Constructive Conflict Outcomes and Integrative Conflict Resolution Approach were not supported via exploratory factor analysis, the final instrument will still attempt to test that configuration. Because the pretest sample was small it was decided that a full-scale measurement respecification based on the results of the pretest was premature. However, certain items (questions) were either eliminated or modified in accordance with the pretest result (see Appendix A1 for the final set of questions that appeared on the questionnaire). The potential for a single factor solution in the final analysis exists, and will be looked for if the more complex, originally proposed measurement model fails to be supported again. 143 3. DATA COLLECTION RESULTS This section documents the collection of the data for the dissertation model test. Included in this section is a description of the data collection procedure and the results of that effort. In addition, evaluations will be discussed as to whether the presence of two distinct sub- samples poses a serious threat to the model test proposed. Other evaluations will be made, including general descriptions of the data obtained and evaluation of potential biases in that data set due to situations such as non—response bias. 3.1 Data Collection Data was collected from the members of two organizations: (1) the National Account Management Association (NAMA), and (2) the National Association of Purchasing Management (NAPM). NAMA members are individuals who manage national accounts for their firms and would generally be considered marketing representatives. The NAPM members surveyed were from.a classification of membership that was most likely to include those that manage purchasing relationships (versus manage the purchasing department), and can best be 144 classified as purchasing representatives. These two organizations were chosen in order to allow data to be collected from both marketers and purchasers. Some of the common titles of the respondents from both groups are listed in the Table 4.2 below: Table 4.2 - Common Respondent Position Titles NAMA NAPM National Account Manager Purchasing Manager National Sales Director Senior Purchasing Agent Director of Sales Senior Buyer Account Executive Purchasing Specialist Corporate Account Manager Supply Chain Manager 3.1.1 The NAMA group NAMA provided their mailing list of 977 US based members and $1500 to support the research. In addition, their executive director co-signed the contact letter as well as provided the envelopes for the mailing. The NAMA members received the cover letter and survey in the NAMA envelope. In addition, a follow-up postcard was sent to all members approximately 3 weeks after the initial mailing. Lastly, the NAMA staff followed up with some members to gain their support directly. A total of 77 usable responses were received from NAMA members. This translates to a response 145 rate of 7.27%. As NAMA members are generally managers who travel frequently and have direct customer relationships to manage, this low response rate was not contrary to what was expected. 3.1.2 The NAPM Group NAPM provided a mailing list representing 1400 members. NAPM provided no financial support for the project. NAPM members were sent a full mailing (cover letter & survey), followed by a postcard, and concluded with a second full mailing. The total response rate was 148 usable responses for a response rate of 10.57%. The lower than anticipated response rate may have been due to the number of surveys the NAPM members receive. Some NAPM members had expressed frustration with “survey overload”, and this attitude could have accounted for the lower response rate. 3.2 Response Rate The total number of usable surveys returned was 219, for a response rate of 9.28%. The 219 surveys are sufficient for the proposed analysis and exceed the minimum goal of 170. 146 Table 4.3 below indicates the week—by-week response tally for this project. Table 4.3 - weekly Response Rate Tally NAPM NAPM NAPM NAMA NAMA NAMA WEEK Re- Total Response IRe— Total Response turned Rate turned Rate Week 1 45 45 3.2% 16 16 1.6% Week 2* 27 72 5.1% 19 35 3.6% Week 3 14 86 6.1% 14 49 5.0% Week 4** 12 98 7.0% 12 61 6.2% Week 5 13 111 7.9% 3 64 6.6% Week 6 27 138 9.9% 5 69 7.1% Week 7 10 148 10.6% 2 71 7.3% TOTAL: 219 9.28% * The postcard reminder was mailed to members of both organizations on the Thursday of this week. ** The second full mailing was sent to NAPM members on the Friday of this week. 3.2.1 Non-Response Bias ’Traditionally, a check of non—response bias is included in sstudies similar to this one. However, the validity of such 61 check is questionable because the non—respondents cannot Ioe compared to the actual respondents because we do not have data from the non-respondents. Both Leslie (1972) and lhnnstrong & Overton (1977) suggest that comparing the late Vs. early respondents is an acceptable proxy for a true non-response bias test. Their belief is that late 147 responses are the most similar to non-responses and thus act as an acceptable proxy. In this study there is no sound basis to make such an assumption. There are numerous arguments against this assumption, which include: 1. As many of the respondents are individuals that would travel as a central part of their job, some may have only happened upon the survey after it had sat idle for quite some time. Thus, their response was late, but not because they were similar in attitude and intention to those that did not respond. 2. It can be more convincingly argued that those that responded immediately share a very similar trait to those that did not respond at all; they practice the same time management principle. “Touch it Once” is a popular adage in time management and would dictate that individuals either, respond immediately, or discard the survey. Thus, many of the intentional non-respondents could just as easily be represented in proxy by the “first” respondents. 148 What both of these examples illustrate is that there is no sound basis for determining that late responders are the best proxy for non—responders. Thus, as much as a valid non—response bias test is desired, the proposed tests by Leslie and/or Armstrong & Overton do not suffice. With that said, it is not without value to look at early ‘vs. late respondents. This was done, not as a proxy for a non—response test, but to see if there was an “order of response" effect. If an effect were detected, an attempt to isolate and explain that effect could be undertaken to ascertain if that effect might impact the validity of the jproposed model test. To that end, the sample was split into thirds (early—middle—late) and a series of ANOVA tests (mere run to see if there was any differences between these groups on the variables of interest. The results indicated that there were no significant differences between the respondents, their firms, or the relationships they selected to be evaluated. In addition, only 5 of the items (out of 90+) that make up the model constructs were found to Show significant differences in means across these three groups. With such a small number of differences it can be assumed that these represent a random, not systematic, set 149 of differences. Thus, there is no evidence that “order of response" effects pose a threat to this study’s validity. 4. THE SAMPLE DESCRIPTION This section details the actual sample obtained. It also discusses the evaluations made concerning any potential problems that having two separate sub—samples may have posed. 4.1 NAPM VS. NAMA Of obvious concern are any significant differences that the tnvo sub—samples may have. It can be expected that there 'will be some differences, in fact these are desired to increase the variance on the items measured and allow for a model test that captures relationship management from both the marketing and purchasing perspectives. A later Project may attempt to isolate the important differences between what could be considered a “marketers” model of Conflict versus a “purchasers” model. However, at this Point the dissertation aims to begin such a process by identifying a base model that is built based on the MO perceptions of both groups. Thus, to evaluate the differences between the NAPM & NAMA sub-samples, a large number of cross—sample analyses were conducted. The results of those are discussed below. A two-tailed t—test executed with SPSS was conducted on a large array of variables, both those that describe the samples, and those that are intended to be part of the dissertation model. The results relevant to the descriptive variables are presented first. This section mull not only compare the two sub—samples but also summarize the descriptive statistics related to the combined sample. 4.1.1 The Respondents The respondents were intended to be those that have direct management responsibility over a business—to—business relationship. Table 4.4 documents that the respondents can be considered qualified to share their perceptions as they have managed the specific relationship which forms the basis of their observations for an average of 4.6 years and have been in their industry an average of 16.6 years. Of the four descriptives shown in the table below only the 151 “time in current position” shows a significant difference (at . = .05) between the NAPM and NAMA sub-samples. The NAPM group had a significantly higher mean in that case. There is no logical reason why this difference should be considered significant and may simply reflect more position stability in the purchasing function. Table 4.4 - Respondent Demographics Descriptive Combined. NAPM NAMA Significant Mean Mean Mean at .05? Time 4.6 4.3 5.1 Not Managing years years years significant Relationship Time in 5.6 6.2 4.4 . . . Significant Current years years years Position Time with 11.2 10.8 11.9 Not Current Firm years years years Significant Time in 16.6 15.9 17.9 Not Industry years years years Significant 52 4.1.2 The Respondents' Firms In addition to details on the respondent, details related to the firm they represented were of interest. In this case three descriptive measures were obtained: (1) firm size as measured by respondent—reported annual sales dollars, (2) the years the firm has been in business, and (3) the number of employees. The only significant difference between the NAPM & NAMA sub—samples is that the NAPM organizations have a smaller number of employees on average. There is no a priori reason that this difference should be of concern. Table 4.5 presents these findings. Table 4.5 - Respondent Firm.Demographics Combined NAPM NAMA Significant Descriptive Mean Mean Mean at . 05 7 Organization 2.6 2.15 3.4 Not Size (annual Billion Billion Billion Significant sales) NUmber of 57 55 61 Not Years in years years years Significant Business IWumber of 11,000 8,600 15,850 Significant Employees in firm 153 Table 4.5 indicates that the firms represented are generally large, long-standing firms. This does pose a potential bias to the generalizability of the results as it could be argued that the results would only apply to similarly sized firms. However, if we look at other descriptive statistics this potential bias does not seem quite as large. First, within the combined NAPM — NAMA sample, the median firm size is $500 million in sales. The mean ($2.6 Billion in sales)is obviously driven up by the extremely large firms (the largest being a 40 billion dollar firm). In addition, the smallest firm is only $1 million in sales. Similarly, the number of employee median is 2,000. The mean is skewed by the presence of a few very large firms (the largest employing 200,000 individuals). 'Dhe smallest firm in this category employs only 15 individuals. Thus, upon further analysis the sample tends to broadly represent the population of firms and while no sample is perfectly representative there is no reason to believe that this sample does not provide an adequate theory test for the model proposed. 154 4.1.2 The Evaluated Relationship As each respondent was asked to select a relationship they managed, the next relevant evaluation is whether the relationships selected have any significant differences between the NAPM and NAMA sub—samples. Three relationship descriptive measures were employed. First, the length of the relationship was documented, second, the percent of business that this relationship represents was recorded, and third, the overall quality rating of the relationship was documented. Some differences were found. Table 4.6 documents these results. ‘Table 4.6 - Relationship Demographics 1 Combined NAPM: NAMA Significant Descriptive Mean Mean Mean at . 05? Length of 10.83 10.89 10.72 Not Relationship years years years significant Significant % of 33.1% 44.4% 9.86% business Overall 2.27 2.27 2.27 Not Quality Significant (l=best, 7=worst) 155 The percent of business is statistically different. The likely cause of this result is embedded in the question itself: “What percent of purchases/sales (with this product category) does this relationship represent to your firm?” It can be expected that the purchaser would only source from a few suppliers for each product category (and have many product categories purchased) and thus the percent of purchases represented by a single relationship would be high within each category. On the other hand, the marketer, whose entire business is often focussed on this single product, would have many customers, and thus the percent of sales that a single relationship represents within that product category is likely to be much smaller. Because this explanation is logically consistent the significant “percent of business” result was not deemed a serious threat to the validity of the overall research results. The overall quality rating of relationships, while not different between the NAPM & NAMA sub—samples, does indicate a potential generalizability threat. The mean is 2.27 where 1 = a “best” relationship and 7 = a “worst” relationship. This apparent lack of variance in this area does weaken the generalizabilty of the findings, as the 156 Inodel will be tested on relationships that are generally considered “best” or “near best.” This problem should have ibeen handled in the data collection stage where certain potential respondents could have been asked to rate the “best” relationship they manage, and others could have been asked to rate the “worst” relationship they manage. This ‘would have ensured a greater amount of variance in this area. However, this was not done and there are other circumstances that mitigate the seriousness of the :resultant problem. First, there is nothing in the theory ‘that indicates that the general model being tested should Iiot hold for a sample of generally better relationships (over one with a broader range of relationship quality. In éaddition, the vast majority reported that the relationship liad encountered conflict as recently as the past year. 'Thus, even these high quality relationships had to deal \Nith conflict and its potential effects on overall :relationship quality. And it is the phenomena of conflict .and.conflict outcomes that this dissertation addresses. 'Therefore, although a greater amount of relationship quality variance would have been preferred, the lack of such variance does not destroy the value of the model test nor the results obtained from that test. The potential to 157 retest the model on a broader range of relationship quality remains open for a future study. 4.1.2 The Focal Variables If the proposed model fully represents the true model then 90 measured variables would be included in that final Inodel. Of these variables only 23 show significant differences between the NAPM and NAMA samples. This is not (deemed a serious threat to the validity of the model test for the following reasons: 1. In no cases are all the questions for a single construct deemed to be different. There are 16 first order constructs in the full model and in 5 cases there are no differences for any of the items that make up that construct, while in the worst cases no more than 50% of the items within a single construct show differences. 158 2. Where there are differences, no specific pattern fully emerges. The NAPM sample is the sample with the higher mean in 15 of the cases, while NAMA is higher in 8 of the cases. Within specific constructs, the specific pattern of which group has the higher mean on different items is also often inconsistent. 4.1.3 NAPM vs. NAMA Conclusion IBased on these findings outlined above, the two samples xvill be combined to form a single sample that will be used ‘to test the dissertation model. While some differences do exist, there seems to be no differences that would a priori invalidate a model test with a combined sample. In eaddition, the differences that do exist can just as easily ibe deemed as helpful because they confirm the value of having both marketers and purchasers involved in the study as they both bring different perceptions to bear on the evaluation of the model. This will allow for a robust and broad theory test (a goal from the onset of the project). 159 4.2 INDUSTRIES REPRESENTED There were 23 industry categories that the respondents could choose from to identify themselves. Table 4.7 documents the industries identified. The results indicate that a primary goal of the research design was achieved; a Ibroad representation of industries. As no single industry category comprises more than 10% of the sample and a great Inumber are represented overall, the sample should allow for a: robust test of the model without any concerns of a sserious industry effect. Therefore, the results should liave generalizability over a great number of industries, aand more importantly the specific effects of any industry :should be captured in a positive way without skewing the entire set of results. up Table 4.7 - Industries Represented Industry Frequency of Percent of Response Total Appliances 4 1.8% Automotive 5 2.3% Building Materials 6 2.8% Chemicals/Plastics 20 9.2% Clothing Textiles 3 1.4% Computer 11 5.1% Construction Equipment 5 2.3% Department Store 1 0.5% Electrical Machinery 6 2.8% Electronics 8 3.7% Food & Beverage 7 3.2% Furniture 1 0.5% Hardware 0 0.0% Machine Tools 2 0.9% Metal Products 7 3.2% Mining/Minerals 4 1.8% Office Equipment 1 0.5% Paper Products 11 3.2% Petroleum Products 1 0.5% Pharmaceuticals 7 3.2% Primary Metals 1 0.5% Rubber Products 3 1.4% Other 103 47.5% 161 5. SAMPLE CONCLUSIONS To summarize the highlights of the above discussions the following conclusions are offered: 1. No serious problems should have arisen from the combination of the NAPM & NAMA sub— samples as there were few important differences between these sub-samples. 2. The respondents had considerable experience managing the relationship they used as the basis for the completion of the questionnaire, and they also had considerable time in their industry. Therefore, the tests conducted on the model using the respondent’s perceptions should be considered appropriate. 3. A broad range of industries is represented and should have provided a highly robust test of the proposed theory. 162 4. The response rate was low, however, a sample large enough to use the proposed methodology was obtained. 5. No serious effects were found based on the order in which the surveys were received. Edith these results it can be concluded that the sample (obtained should have provided a legitimate test for ‘the theory proposed. The following chapter (Chapter 5) mull review the measurement and structural testing c>f the model and whether the data supports these z>roposed models. Chapter six will document the :Eindings and offer managerial and theoretical evaluations . 163 Chapter 5 Results of the Statistical Analysis 1. INTRODUCTION This chapter outlines the results of the data analysis. At the basic level, the chapter reports the answers to two questions: 1. Were the proposed constructs supported by the data? 2. Were the hypotheses supported by the data? As will be reported in this chapter, there were mixed answers to both questions. The measurement results indicated that a sufficient number (but not all) of the key constructs gained measurement support and thus the structural tests of the hypotheses could proceed. The structural results were also mixed as some of the kunpotheses gained support while others did not. This cflnapter reviews the results of the procedures executed to NM analyze the data, beginning with the measurement model results, followed by the tests of the hypotheses. 1.1 Chapter Organization This chapter has four sections following this introduction. Section 2.0 presents details of the data, with special attention given to the normality of the data obtained. Section 3.0 documents the results of the measurement testing. This section includes a table that summarizes the support for the constructs and a pair of figures that are visual representations of the measurement models tested. The results of the hypothesis tests are reported in Section 4.0. A path model that captures the hypotheses was specified and tested and these results are reported. Lastly, Section 5.0 offers a conclusion to the analyses and acts as a bridge to Chapter Six where the implications of the findings are discussed. 2. DETAILS ABOUT THE DATA Tlris section briefly reviews the key distribution atnzributes of the variables collected in this study. 165 Specifically, the univariate and multivariate normality of those variables is assessed. 2.1 The Distributions of the Variables The key concern with the data was its conformity to a normal distribution for both the univariate and multivariate cases. This was of concern because the choice of estimation method in both the measurement tests and the path model could be influenced by these evaluations. The structural equation modeling estimation technique of first choice, Maximum Likelihood (ML), is sensitive to violations of normality and thus this evaluation was needed. The assumption used is that multivariate normality is desired and this requires univariate normality. Both of these are evaluated below. 2.1.1 Univariate Descriptions Both kurtosis and skewness were evaluated. EQS reports this for each set of variables used and this was evaluated during the execution of each CFA. Univariate normality was supported in each case with no extreme cases of kurtosis 166 (over 3) or skewness (over 7). This provides a basis for evaluating multivariate normality below. 2.1.2 Multivariate Nbrmality For each run of EQS, the Mardia coefficient is reported. While there is no exact definition of what is “too large,” this dissertation determined that any value over 40 would be considered suspicious. When 40 was exceeded, the Robust Chi—Square and parameter significance tests developed by Sattora-Bentler and recommended by West, Finch & Curran (in Hoyle, 1997) were used and reported. 3 . TESTING THE MEASUREMENT MODELS The proposed measurement model was rather complex, and thus to facilitate analysis within the sample size constraints the entire measurement model was first divided into groups to allow for some respecifications within these groups before the full measurement structure was assessed. This section presents the results of all the relevant :measurement model tests and documents the necessary measurement respecifications. 167 The criteria for evaluating the measurement models are set out in chapter three. The originally specified models, as well as the respecified models, were subjected to these criteria and judgments concerning the model’s adequacy were made. In some cases respecifications were required. In order to execute these respecifications, an iterative process was used that involved the evaluation of reliability values as well as the results of EFA and CFA procedures. However, the final respecifications were based on the following two criteria (presented in appropriate rank order): 1. Theory — since all constructs had multiple measures assigned to them, within—factor respecifications could be made that remained consistent with the originally proposed theory. If other respecifications were suggested by the statistical techniques employed, these modifications were acted upon only when the resultant measurement structure remained consistent with the proposed theory. 168 2. Statistical Evidence — a combination of reliability (Cronbach’s Alpha), EFA, and CFA procedures were reviewed to obtain the best measurement model. Statistical evidence was used only in theory consistent ways to avoid a statistically strong, but theoretically and practically weak result. To start the measurement model assessment process, two sub— groups of constructs were considered. The first sub-group of constructs evaluated were those that were conceptualized to make up Relationship Structure (as proposed by Robicheaux and Coleman, 1994). The second sub—group of constructs considered was the conflict characteristics (frequency, intensity and importance). These constructs were primarily culled from Brown & Day (1981). Both of these sub—groups were evaluated and necessary measurement respecifications were made. In addition to these two sub- groups of constructs, the proposed second—order measurement Imodel for Constructive Conflict Outcomes was assessed and inespecified. With these tasks completed, all the relevant cmnnstructs were subjected to a CFA to verify their 169 measurement properties. All of these procedures and the associated results are reported in the following sections. 3.1 Relationship Structure Relationship structure was conceptualized to consist of two components: Decision—Making Structure and Operational Integration. Both of these components were conceptualized (and operationalized) to consist of multiple sub-components or second—order constructs. The results of the tests of these original conceptualizations, and the subsequent respecified models, are presented in sections 3.1.1 and 3.1.2. 3.1.1 Decision-Making Structure A multi—factor model for Decision—Making Structure was proposed. The four first-order constructs that made up Decision-Making Structure were 1) Formalization, 2) Decentralization, 3) Participation, and 4) Shared Paradigm. The reliability of the items that made up these four constructs was assessed first. The results are presented in Table 5.1. 170 Table 5.1 - Decision—Making Structure Item.Reliabilities Construct # of Items Cronbach's Alpha Informalization 6 items .5870 Decentralization 6 items .5411 Participation 7 items .7686 Shared Paradigm 5 items .8130 The reliability scores point to potential problems with both the Informalization and the Decentralization constructs. However, for Decentralization, a four—item construction obtained a reliability score of .7439. This value, as well as the reliability scores for Participation and Shared Paradigm, exceeds the .7 cutoff adopted in this dissertation. No combination of the Informalization items could be found that exceeded this cutoff, therefore Informalization was removed from subsequent analysis. The next stage in the measurement testing was to take the reliable measures and subject them to an exploratory factor analysis. The results indicated potential problems with 171 some of the measures. After interpreting the results of the EFA and eliminating a few items from the original factor structures, a confirmatory factor analysis was conducted to test the respecified constructs. Table 5.2 (on the following page) documents the results. As indicated in Table 5.2, the items used showed acceptable reliability. The Participation items were only slightly below the .7 cutoff and this was not considered a serious concern considering the exploratory nature of these constructs (especially related to the intraorganizational to interorganizational transfer of these constructs). The respecified measurement model is supported via the CFA. Since no indication of a multivariate kurtosis problem was found, Maximum Likelihood was used as the estimation technique. The model met the criteria for preliminary fit, indicating that the results should be legitimate. Violations of these preliminary fit criteria would indicate ‘unrealistic results and would point to identification or specification problems. The model as a whole also gxarformed well against the relevant criteria. The Chi- :yquare value was acceptable considering the sample size. 131.addition, the fit indices both surpassed .85. 172 Table 5.2 - Decision-Making Structure CFA Results RELIABILITY: Centralization (4 items): Alpha = .7439 Participation (4 items): Alpha 2 .6839 Shared Paradigm (5 items): Alpha = .8130 (See Appendix A2 for final items used) CEA RESULTS CRITERIA: RESULTS: Normality: Mardia Coefficient: 28.53 - no multivariate Kurtosis problem, thus ML estimation was used. Preliminary Fit: Negative Error No Variances: Factor Loading < .5: No Factor Loadings > .95: No Model Fit: Chi-Square & p—value: 94.268 with 74df p=.05613 BBNNFI: .968 CFI: .974 Construct Validity: Convergent Support: All Loadings Significant on predicted factor. Discriminant Support: LM Test Method: Only 1 crossloading, not theory relevant. 173 Convergent validity was assessed by reviewing the factor loadings. All factor loadings were significant and each item loaded on the proposed construct. Discriminant validity was supported via an assessment of the LM test results. In this case one modification was suggested, however, if this path was allowed there would be no support for it in theory (thus it would have been data driven, not theory driven). Because the support for this measurement model was strong, the constructs will be carried into the next test as constructed in this test. 3.1.2 Operational Integration A four—factor model of Operational Integration was initially proposed. The four constructs were: 1) Assistances, 2) Monitoring, 3) Joint Actions, and 4) Information Exchange. The original items for each of these constructs are outlined in Appendix A1. The reliability test results are presented in Table 5.3. 174 Table 5.3 - Operational Integration Item Reliabilities Construct # of Items Cronbach's Alpha Assistances 5 items .7700 Monitoring 5 items .5060 Joint Actions 6 items .6884 Information Exchange 11 items .8265 (combined) Frequency 3 items .5654 Type 5 items .6998 Capability 3 items .7447 The reliability results pointed to problems with Monitoring as a construct. Using a three—item version of the construct, a reliability of .6627 was obtained, which still falls short of the .7 guideline. In addition, the poor reliability of the frequency component of Information Exchange indicated problems. With this in mind, and using the assumption that factor analysis is the more important analysis (as it relates to validity, versus simple reliability), the items were subjected to a series of EFA 175 tests. Items were deleted in an attempt to get a “clean” factor structure (.6 loading, with no cross—loading of .4). After many iterations and a great deal of respecification a “clean” two—factor structure was achieved. The resulting two—factor solution is substantially different from the original specification. The new structure involves the items grouping around two groups of functional activities related to integration. The first involves the integration of the activities that enable effective communication and coordination within the relationship (titled Communication & Coordination Integration). The second involves grouping the activities that relate to the facilitation of efficient transactions within the relationship (titled Transactional Integration). These new constructs remain logically consistent with the original conceptualization in that, between them, they incorporate questions from each of the original four Operational Integration sub-constructs. The data indicated that the grouping by activities was stronger than a conceptual grouping. Unfortunately, there is no way to tell if this was idiosyncratic to the sample or represented a proper reconceptualization without having another sample 176 to retest this configuration. However, it was decided that this conceptualization would be carried forward and subjected to a CFA. This decision was made because the new conceptualization makes logical sense given the relationship as the unit of analysis (thus activities related to integrating the relationship form a logical basis for construct development). In addition, this new conceptualization incorporates all the originally proposed components of integration (in one or the other of the two new constructs). The results of CFA are presented in Table 5.4 (on following page). The reliabilities of the items making up the two constructs meet the .7 cut-off. The model as a whole fairs well. The chi—square is significant, but this is expected given the sample size and number of variables. The Fit Indices show adequate fit. As no single measure of model fit is without problems, this combined assessment makes the most sense and this model satisfies the overall criteria set out in chapter three. 177 Table 5.4 - Operational Integration CFA Results RELIABILITIES: Communication & Coordination Integration (6 items): .8334 Transactional Integration (4 items): .7245 (see Appendix A2 for the actual measures used) CEA RESULTS CRITERIA: RESULTS: Nbrmality: Mardia Coefficient: 24.6 No serious multivariate kurtosis problem. Preliminary Fit: Negative Error Variances: No Factor Loading < .5: YES (.468, .491) Factor Loadings > .95: No Mbdel Fit: Chi-Square & p—value: 57.891 with 34df p=.00649 BBNNFI: .949 CFI: .961 Construct validity: Convergent Support: All Loadings Significant on predicted factor. Discriminant Support: LM Test: Only 1 crossloadings suggested, not theory relevant. 178 Convergent validity support is mixed. While all factor loadings are significant (based on the t—statistic test) there are two standardized factor loading values that are below .5. This is a problem as there are only four items for this construct. This result, combined with the fact that this construct was a creation of the data structures (versus a proposed structure), was considered sufficient evidence to reject the Transactional Integration construct (it did not explain sufficient variance in the items it was suppose to indicate). Therefore, only the Communication and Coordination Integration construct was carried forth into subsequent analyses. 3.1.3 Relationship Structure Measurement Conclusions In conclusion, there were four first—order constructs representing Relationship Structure that gained preliminary measurement support and thus were carried forward into the full model CFA. These are Participation, Decentralization, Shared Paradigm and Communication and Coordination Integration. These constructs showed sufficient reliability and validity. The second—order structures originally proposed for both Decision—Making Structure and 179 Operational Integration did not gain support as some of the proposed first-order constructs failed to gain reliability support, thus eliminating the chance to fully test such structures. However, the remaining first—order constructs, if supported in the full—construct CFA, are sufficient to test the proposed Relationship Structure related hypotheses with only limited modifications to those hypotheses. 3.2 Conflict Characteristics There were four original conflict characteristics proposed. However, the items intended to tap Conflict Focus failed to gain reliability support and thus this construct was removed from the analysis. The remaining proposed constructs were Conflict Frequency, Conflict Intensity, and Issue Importance. Conceptually there was a problem with the data collection related to these constructs. The respondents were asked to identify how frequent conflict was, how intense the conflict was and how important the issue was for twenty-one (21) separate issues. The potential problem arose because it was logical to assume that there were underlying groupings of issues within the list of twenty-one. To um assess this, a series of EFA procedures were conducted. The “issue importance” scores were used within these EFA runs because it was most logical to have the issues form groups based on their underlying importance. A three— factor structure emerged. Table 5.5 presents these results. The first factor (or grouping of issues) was titled “Relationship Management,” the second “Logistical,” and the third “Strategic.” The individual issues that represent each category are presented in Table 5.5. The individual scores on each group of issues for each of the three conflict characteristics were averaged and in the end nine variables were created. 181 Table 5.5 — Conflict Characteristic Issue Groupings ISSUE ROTATED FACTOR GROUP ING INDIVIDUAL ISSUES LOADING” Level of sales .711 support Level of service .783 support Relationship Management Issues Communication of .698 problems Frequency of .745 communication Ease of communication .673 Inventory policies .789 Logistical Issues Warranty policies .757 Delivery terms .671 Strategic objectives .764 Strategic Issues Control of resources .809 Sharing of market .677 information *‘Varimax rotation was used and only loadings over .4 are reported . 182 The nine variables created are listed below: (1) Relationship Management Conflict Frequency (RMCFREQ) (2) Relationship Management Conflict Intensity (RMCINT) (3) Relationship Management Issue Importance (RMCIMP) (4) Logistical Conflict Frequency (LOGCFREQ) (5) Logistical Conflict Intensity (LOGCINT) (6) Logistical Issue Importance (LOGCIMP) (7) Strategic Conflict Frequency (STRCFREQ) (8) Strategic Conflict Intensity (STRCINT) (9) Strategic Issue Importance (STRCIMP) Table 5.6 presents these variables and the associated reliabilities. These nine variables were then used as indicators of three constructs: (1) Conflict Frequency, (2) Conflict Intensity, and (3) Issue Importance (i.e. RMCFREQ, LOGCFREQ & STRCFREQ indicated Conflict Frequency). These constructs and their associated indicators were carried into the full model CFA and their measurement support was derived from that analysis. The results will be presented subsequently in this chapter. 183 Table 5.6 - The Conflict Characteristic Construct Indicators Reliabilities for Each Conflict Characteristic - Issue Group Variables* Frequency Intensity Importance Relationsmp RMCFREQ RMCINT mom Management (5 issues) a = .7961 a = .7990 a = .8133 (3 issues) a = .6913 a = .6906 a = .6619 (3 issues) a = .7897 a = .7839 a = .7056 * Variable Names: issue group + conflict characteristic (i.e. LOGCINT = logistical issue conflict intensity) 3.3 Constructive Conflict Outcomes (CCO) Constructive Conflict Outcomes was originally proposed to be a second-order construct comprised of the following first—order constructs: (1) Hostilities Release, (2) Activity Adjustment and (3) Norm Adjustment. That structure failed an EFA analysis in the pretest. However, since the sample size was small in the pretest, a second attempt was made to confirm such a structure. The first appropriate assessment was a reliability analysis of the first-order constructs. The results are presented in Table m4 5.7 and indicate that there are severe problems with the reliability of those constructs. Table 5.7 - CCO Item.Reliabilities CCO Sub-Construct Reliability Hostilities 5 items .4434 Release Activity 3 items .4174 Adjustment Norm Adjustment 4 items .4146 All three sets of items showed poor reliability. This is consistent with the pre—test results. No confirmatory factor analysis was attempted because of these extremely poor reliabilities. However, a single-factor solution consisting of five items (similar to the pre—test result) was constructed Via an EFA and was carried forth into the full—model CFA presented subsequently. 185 3.4 The Full-Model CFA As presented in the previous sections the constructs listed below would be tested in a CFA designed to simultaneously test of all the relevant model constructs. CFREQ - Conflict Frequency CINT — Conflict Intensity CIMP — Conflict Issue Importance CCINT — Communication & Coordination Integration PART - Participation DECENT — Decentralization SHARPAR — Shared Paradigm CCO - Constructive Conflict Outcomes In addition to these constructs previously discussed, there were two more relevant constructs that were in the proposed model. These two are listed below. SAT — Relationship Satisfaction TRUST — Relationship Trust Thus, in the end there are ten (10) constructs that were incorporated into the full-model CFA. The measurement 186 model tested is presented in Figure 5.1 (on following page). The exact number of items used to indicate each construct and each item's variable number is represented in that figure. In some cases the number of items used is fewer than when the CFA process began. A few minor modifications were made to improve the performance of the measurement model, however the final—model presented and defended in the following sections still involves all ten relevant constructs. 187 Hopes use—556.32 manmfihmuwlaad 0cm. .. H.m masons Acetone—...... so: >=o=m> EmEEtoeov 3:68.685 .585ch 2833.... 62 s... 32:; E85259 cocoa. oo~_mo£oo>c co omo. moist? __< 5. v o .m .585ch 9... 888.; 2:22:08 58 __< mquzmquEma mmhm2m PN> Nm> m_.> m; T“; _ a; $> «S :> To: Tie _om>_ _§>_ 550 I R> E H E E E E a _B>_ Si v> 4 m9 75> _ _mm> _ .5 I Tm; T~> _ nm> 188 3.4.1 Statistical Support for the Model The model was evaluated in a series of ways. First, the reliability was assessed and the model gained sufficient support in that area. Second, the goodness—of—fit measures were reviewed and again the model was supported. The third review was an evaluation of the construct validity of the constructs in the model. In this evaluation, problems were found with the content validity of Constructive Conflict Outcomes as well as problems with the discriminant validity of Communication & Coordination Integration and Shared Paradigm. Each of the evaluations is addressed more fully in the sections that follow. After those evaluations are reviewed, a “Conclusions” section is presented that indicates that a set of model respecifications is necessary for the measurement model. The respecified model is discussed in Section 3.5 of this Chapter. .Reliability; The reliability of the set of items making up each construct was assessed and the results are ;presented in Table 5.8. All but three of the constructs .achieved a reliability of .7 or greater, and the remaining tflrree all exceeded .64. These results indicate that the 189 items used to indicate the constructs generally hang together as is required for a suitable measurement model. Table 5.8 - Item Reliabilities Construct # of Reliability Items CFREQ 3 . 7904 CINT 3 .7862 CIMP 3 .6429 CCINT 5 . 8126 PART 4 .6839 DECENT 4 . 7439 SHARPAR 3 . 7025 CCO 3 .6590 SAT 4 .9292 TRUST 4 .8345 MOdel Fit. The confirmatory factor analysis was executed using EQS. At a technical level, the EQS program ran without error messages or conditions codes indicating that the results were the product of a properly executed program run. In addition, the results were rational (e.g. no negative error variances or parameters that were 190 unreasonably large or small) indicating that the run was legitimate. Hoyle and Panter (in Hoyle, 1995) recommend that a group of model fit evaluations be made. Specifically these authors recommend that the Chi—square statistic and its associated p—value be presented and the value of at least two incremental fit indices (that compare the model to a baseline model) be reported. In addition, Hair et. al. (1995) recommend that the evaluation of the Chi—square statistic be complimented by a second evaluation of absolute fit such as the root mean square error of approximation (RMSEA). All of these recommendations and/or suggestions are heeded and a full set of model fit measures are presented in Table 5.9. 191 Table 5.9 - Goodness-of-Fit Measures for Initial Measurement Model GFI Measure Value Chi-Square 741.87, 539 df p < .001 S-B Scaled Chi— 650.73, 539 df Square p = .00066 BBNNFI .936 CFI .945 RMSEA .045 Based on the overall evaluation of the fit measures presented in Table 5.9, the general conclusion is that the model proposed passed the goodness-of—fit evaluation. Specifically, the BBNNFI and the CFI exceed .90 and this is generally considered to represent values that indicate acceptable model fit. Additionally, the RMSEA was below .05 and thus is considered acceptable (Hair, et. al., 1995). Both of these results indicate good model fit. The Chi-Square value is significant indicating that the covariances implied by the proposed model do not match the observed covariances. This indicates a lack of model fit. 192 However, this statistic has been shown to perform poorly when sample sizes are in the range used in this study (n=200) and thus the significant result is not surprising, nor fatal to an assessment of model fit. Thus the final conclusion is that this measurement model has an acceptable level of model fit. Construct validity. There are four components of construct validity that are considered for the measurement model: (1) content validity, (2) nomological validity, (3) convergent validity, and (4) discriminant validity. Each of these is discussed in the following paragraphs. Validity is the extent to which a set of items properly represents the concept under study (Hair, et. al., 1995). The content validity (or face) validity is an assessment of whether the content of the items appropriately taps the concept under study. For the majority of the constructs there are no problems in this area. However, for the Constructive Conflict Outcomes (CCO) construct the content validity of the items is questionable. The items that represent CCO only tap the conflict outcomes from a positive or constructive viewpoint. That is, they do not capture negative or destructive outcomes. The original set 193 of items designed to tap CCO did include some questions that attempted to address destructive outcomes of conflict, however, the EFA procedures indicated that these items would not load with the items that addressed the constructive outcomes of conflict. The elimination of these questions is problematic because the intent of this dissertation was to include a full range of outcomes (destructive to constructive) and the current CCO construct only assesses the conflict from a neutral to constructive standpoint. The steps taken to rectify this problem are outlined in Section 3.5. In addition to the content validity, a second type of validity was considered. That is the nomological validity of the measurement model. Nomological validity is supported when the web of relationships suggested by theory is supported by the data. A review of the correlations between the constructs shows basic support for the nomological validity of the model. There were strong correlations between trust and satisfaction (.930), conflict frequency and conflict intensity (.856), conflict frequency and.satisfaction (—.450) and trust and conmmnication and coordination integration (.888). All of tflnese results are logically and theoretically consistent. 194 Thus, the support for the nomological validity of this model was obtained. Convergent validity assesses the degree to which the items “converge” upon the construct that was proposed to be associated with each item. Convergent validity is supported when the path coefficients (lambda) are significant and each item loads on the construct it is supposed to load on. Within the model tested, all of the path coefficients were highly significant and each variable loaded on the construct it was proposed to load on. Therefore, there is support for the convergent validity of the constructs within this measurement model. The support for this is presented in Table 5.10. 195 Table 5.10 - Convergent Validity Support for the Constructs within the Initial Measurement Model STD. CONSTRUCT VARIABLES mm T-VALUE LAMBDA v44 RMCFREQ 1.040 14.20 .899 ’1 ’ CFREQ v47 LOGCFREQ .795 10.26 .600 0‘ = -7904 V50 SRTCFREQ .792 10.59 .678 v45 RMCINT 1.037 13.86 .884 F2 ' CINT V48 LOGCINT .760 10.08 .574 a = -7862 v51 SRTCINT .821 11.72 .724 V46 RMCIMP .998 9.24 .747 ’3 ' CI"? v49 LOGCIMP .830 7.66 .494 a = ~5429 v52 SRTCIMP .949 8.43 .610 v3 ASSITl .738 7.54 .531 v19 ASSIT4 1.075 11.50 .743 F4 ' CCINT v37 ASSITS .846 11.34 .735 a = .8126 v4 INFOTYPl 1.005 9.36 .635 v34 INFOCAP3 .972 10.53 .696 v11 PARTl 1.127 9.29 .714 F5 - PART V12 PART4 1.200 8.95 .690 a = .6839 v13 PARTS .896 6.81 .538 v14 PART6 .869 6.45 .512 v9 CENT2 1.103 9.24 .666 F6 — DECENT v29 CENT4 1.149 8.96 .649 a = .7439 v33 CENTS .902 8.40 .616 v40 CENT6 1.114 10.27 .725 V6 SHARPARl .819 11.31 .748 F7 ‘ SHARPAR v21 SHARPAR3 1.020 9.56 .651 9': ~7025 v43 SHARPARS .735 8.33 .582 v28 ACTAD2 1.194 9.53 .769 ’8 ‘ °c° v36 ACTAD4 .806 6.70 .537 0‘ = -6590 v7 HOSTRELl .857 7.84 .625 v5 SATl 1.335 15.20 .882 F9 - SAT V18 SAT2 1.282 15.80 .902 (x = .9292 v32 SAT4 1.205 14.52 .858 v39 SAT5 1.313 14.13 .843 v23 TRUSTl 1.166 11.71 .750 F10 - TRUST v24 TRUST2 .979 11.88 .759 (x = .8345 v26 TRUST4 1.255 11.59 .746 v27 TRUSTS 1.167 11.54 .744 196 Discrimina: each const: assessment n0 differer. constructs initially a were twelve loading was Coordinatio Paradigm (5 Change : 7. 0f the Strc L955) sus; construCtS further ass 0r19ina1_1nc covaria,n Ce Discriminant validity is an assessment of the uniqueness of each construct in the model. A model that fails this assessment does so because the data indicate that there is no difference between a pair (or set of pairs) of constructs in the model. Discriminant validity was initially assessed by reviewing the LM test results. There were twelve significant cross—loadings suggested by this test. There was one path that was troublesome. A cross— loading was suggested between a Communication and Coordination Integration (CCINT) item and the Shared Paradigm (SHARPAR) construct (V4,F7 : predicted Chi-square change = 7.791). Considering this cross-loading in light of the strong correlation between these two constructs (.955) suspicion was raised as to whether these two constructs were truly distinct from one another. To further assess this potential discriminant validity problem a nested model review was conducted. In this test the original model is compared to a model that has the covariance between these two constructs set equal to one. If the constructs are distinct it can be expected that this constraint would cause the model to perform worse. Specifically, with the one degree of freedom change in the model, the Chi—square should change (increase) by at least 3.84 if the constructs are indeed distinct. When this test 197 was conduc‘ STRPAR se was not 51 dfl. This validity 1 Other sig determine relevant. on all pa SUggestec over .8. fit was 5 Therefore was conducted, with the covariance between CCINT and SHARPAR set equal to one, the deterioration of model fit was not significant (change in Chi-square < 3.84 with 1 df). This indicates a lack of support for the discriminant validity of these two constructs. Other significant LM Test results were reviewed and it was determined that releasing these paths would not be theory relevant. In addition, the nested model test was conducted on all pairs of constructs where a cross—loading was suggested or the correlation between the two constructs was over .8. In all of these cases the deterioration of model fit was significant (change in Chi—square > 3.84). Therefore, the discriminant validity of the other constructs was supported. 3.4.2 Conclusions In a final assessment, the measurement model does not perform adequately. While there is support for the nomological validity and convergent validity of the model, it fails to gain full support in the content and discriminant validity areas. With these failures in mind, the measurement model was respecified in an attempt to 198 overcome t“. of Constru: as a combi. the tests following 3.5 Because 0 measureme; Created a; This mode while sti basic ess overcome these problems. Specifically a new construction of Constructive Conflict Outcomes was considered, as well as a combined CCINT/SHARPAR construct. These changes and the tests of this respecified model are presented in the following section. 3.5 The Respecified Measurement Model Because of the problems with the previously presented measurement model, a revised or respecified model was created and tested. This model is presented in Figure 5.2. This model overcame the shortcomings of the previous model while still allowing for hypothesis tests that capture the basic essence of the originally proposed model (See Chapter 2). Specifically, the model used a single—item measure of Constructive Conflict Outcomes and a newly created construct that captured the combination of Communication & Coordination Integration and Shared Paradigm. This model received full support as it showed acceptable reliability, model fit and passed all four validity assessments. This section presents the conceptual support for the two new constructs (sub—sections 3.5.1 and 3.5.2) and then reviews the results of the statistical tests (e.g. CFA) designed to assess the support for the model (sub—section 3.5.3). 199 This SE that pC forward This section finishes with a conclusion (sub—section 3.5.4) that positions this model as the one that was carried forward into the path model testing. 200 aka WE . 1 A mn> E ~ mm> ~ ~ mw> \ E Hmpoz unmfimusmmwz pmemeumammm one. .. N.m Gunman 26:65 2558629 .3086me 952.850 .39: 863mm: 02 3:68.688 E922 >665 o: £885 .8. 5... Sea; 2896269 563 8585093 :0 poo. 839:; __< 5. v o 6 E8536 o8 AmooEm; 9:22:80 Ema __< ovo. mp F. mmm. new. com. Ohm. ”m 6.9 vm> 95 9S cs 85 was was Nm> 64> mm> 64> om> 3S <8. I 8> m> m> a; a; B> m; 93 61> a? mw> o> « as as vm> 201 As was p Conflict validity destructi different fOllOWS: RGflq the ] of t] DEST] (him 3.5.1 Changes to Constructive Conflict Outcomes As was presented in sub—section 3.4.1, the Constructive Conflict Outcomes construct had questionable content validity because the group of items failed to capture the destructive side of the outcomes. To overcome this a different CCO variable was considered. The item was as follows: Reflecting on the conflict that has occurred between the parties, I would characterize the overall impact of that conflict as: DESTRUCT IVE CONSTRUCTIVE (hindered/hurt relationship) (improved relationship) 1 2 3 4 5 6 7 'Phis item has the advantage of addressing conflict outcomes frrnn both the destructive and constructive sides. Its (iisadvantage is that it is a single item and thus there is rupggood.way to access its reliability. However, validity CEHI be assessed, and if validity is supported, than relijfloility can be assumed. The measurement model presented in section 3.4 was altered and the 3—item CCO construct was replaced with the single—item version presented above. The model performed equivalently to the 202 original fit, all validity presentec overcome existed t Integrati of the fu sub-secti original model on the relevant measures (equivalent model fit, all loading paths significant and discriminant validity mostly supported). The specific results are not presented because the model was further respecified to overcome the discriminant validity problem that still existed between the Communication & Coordination Integration and Shared Paradigm constructs. The results of the fully respecified model test will be presented in sub-section 3.5.3. 3.5.2 Changes to Relationship Structure The results of the original measurement model test indicated a lack of discriminant validity between Comnmnication & Coordination Integration and Shared Paradigm. These two constructs were supposed to tap separate components of the structure of the relationship. Even though these constructs failed to gain an appropriate level of measurement support, it is still desirable to capture components of the relationship structure because the research made hypotheses concerning the effects of relationship structure on Constructive Conflict Outcomes (the focal variable). Therefore, an attempt was made to cxmnbine these two constructs. Obviously, the data already 203 suppor level « would 1 follow: constrL for the To begi the two present questio Structu had the was cre supported such a merger, however, there needed to be a level of conceptual acceptability to the merger before it would be executed in this research. Therefore the following paragraphs outline the process by which the new construct was developed and provides the conceptual support for the new construct. To begin the process, the individual questions that made up the two original constructs were reviewed. These are presented in Table 5.11 (on following page). All eight questions were subjected to an EFA and a single factor structure emerged. After removing the three questions that had the lowest factor loading scores a five—item construct was created. This construct is presented in Table 5.12. 204 Table 1 {... Coo: ASSIT this I each c potent proble delays 5‘ ASSIT each 0 analys prOble: \ ASSIT ; each o proble; OChEr' \ INFOTY infOrm \ able t info:m Table 5.11 - The CCINT and SHARPAR Items Communication 5 Coordination Integration (5 items) Shared Paradigm (3 items) ASSIT 1: The parties in this relationship advise each other in advance of potential operational problems (such as delivery delays or stock outs) SHARPARl: The parties often agree on the appropriate course of action for this relationship ASSIT 4: The parties help each other in value analysis, cost reductions, problem solving, etc. SHARPAR3: The parties often disagree on what is in the best interest of the relationship (R). ASSIT 5: The parties advise each other of potential problems in meeting each other’s needs INFOTYPl: Strategic information is readily shared in this relationship INFOCAPB: The parties are able to exchange data or information easily and in usable formats. SHARPARS: The parties often agree on what constitutes appropriate relationship behavior 205 Table 5 ASSIT ASSIT INFOCA fafé/ Table 5.12 - The IPSA Construct Items Integrated Problem.Solving Approach (IPSA) a = .8303 ASSIT 4 The parties help each other in value analysis, cost reductions, problem solving, etc. ASSIT 5 The parties advise each other of potential problems in meeting each other’s needs INFOCAP3 The parties are able to exchange data or information easily and in usable formats SHARPARl The parties often agree on the appropriate course of action for this relationship SHARPARS The parties often agree on what constitutes appropriate relationship behavior rm The new Approacl problem (the pa: issues a and (2) viewpoir within t The new construct was titled Integrated Problem Solving Approach (IPSA). The IPSA construct taps an “integrated problem solving approach" at two levels; (1) Operationally (the parties interact directly to deal with important issues and share necessary information in usable formats) and (2) Philosophically (have a coordinated or integrated viewpoint as to what constitutes appropriate behavior within the relationship). The creation of this variable makes it more difficult to specifically test the “Relationship Structure” configuration proposed by Robicheaux and Coleman (1994) because it blurs the distinction between “Decision—Making Structure” and “Operational Integration,” which are separate components of that structure. Communication and Coordination Integration was a construct under the Operational Integration umbrella, while the Shared Paradigm construct belonged to the Decision—Making Structure group. 'Therefore, this combination of these two constructs is a theoretical disadvantage. The Robicheaux and Coleman (R&C) configuration offered a formal framework by which to assess and.incorporate the underlying structure of the exchange :relationships under study and with the elimination of these two constructs as separate entities it now becomes very 207 difficul the seve The firs structui Assistar already measuren purpose to incor underlyi relation fits Wit the unde the R&C difficult to test the efficacy of that structure. However, the severity of this problem is mitigated for two reasons. The first reason is that the ability to fully test the R&C structure was already hindered because the Formalization, Assistances, Monitoring, and Participation constructs had already been eliminated from the research project due to measurement problems. The second reason is that the purpose of using the Robicheaux and Coleman structure was to incorporate a formal framework that captured the underlying condition or structure of the exchange relationship. And even though this new construct no longer fits within the R&C framework, it still helps to capture the underlying condition of the relationship. Thus, on a practical level, the ability to fully test the specifics of the R&C framework had already been severely diminished and on a conceptual level the new construct (in conjunction with Decentralization and Participation) does offer a way to assess the underlying condition of the exchange relationship. Furthermore, the ability to make an assessment of the effect the underlying relationship structure has on Constructive Conflict Outcomes is more fundamental to the research than the use of any one specific framework for describing that underlying relationship structure. 208 On a statistical level this construct is sufficiently reliable (a = .8303) and has very strong factor loading values in an EFA setting (all over .7). Its measurement performance is also strong when assessed in the all— construct CFA. The results of that CFA will be reported in the sub-sections that follow. 3.5.3 Statistical Support for the Respecified Model The respecified model gained strong support on a variety of measures. The first assessment was reliability. This tests whether the items used to tap each construct have consistency in their measurement. As indicated in Table 5.13 (on following page), the groups of items showed acceptable levels of reliability. Conflict Importance and Participation were below the generally accepted .7 level, Ihowevemy their reliability values were still strong enough to be acceptable for this research. 209 Table 5.13 - Item Reliabilities for Respecified Measurement Model Construct ;:;:: Reliability CFREQ 3 .7904 CINT 3 .7862 CIMP 3 .6429 IPSA 5 .8303 PART 3 .6620 DECENT 4 . 7439 CCO 1 n/a SAT 4 .9292 TRUST 4 .8345 210 The model was evaluated via a confirmatory factor analysis that was executed using EQS. The program ran without errors or condition codes and no irrational results were found (such as a negative error variance). The model did require one nontraditional programming component. The measurement errors between each set of conflict issues were allowed to covary. Specifically, the errors between the three relationship management conflict variables (RMCFREQ, RMCINT & RMCIMP), the errors between the three logistics conflict variables (LOGCFREQ, LOGCINT & LOGCIMP) and the errors between the three strategic conflict variables (STRCFREQ, STRCINT & STRCIMP) were allowed to covary. This was done because of the obvious connection between these variables. The intent of the research was to evaluate the effect of the conflict characteristics (frequency, intensity & issue importance), however it is perfectly logical that the data would also attempt to group by the underlying issues. Thus, to account for this logical connection, and to still stay focussed on the intent of the research, these errors were allowed to covary. (This programming action was also taken when the previous measurement model had been evaluated, thus this was not a point of difference between the two CFA programs). 211 The results of the CFA supported the measurement model. The model as a whole showed adequate model fit. As Table 5.14 displays, the Chi—Square statistic was significant, however the Satorra—Bentler Scaled Chi-Square indicated a non—significant model. The Satorra—Bentler Scaled Chi- Square does a better job dealing with any normality violations that the data may exhibit (and are suggested by the Mardia Coefficient value that is over 40) and thus may be the more appropriate measure of model fit. A non- significant model indicates a match between the model- implied covariances and the actual covariances. This result, combined with the high CFI (.967) and low RMSEA (.040) indicate a model with acceptable fit. Table 5.14 - Goodness-of-Fit Measures for Respecified Measurement MOdel GFI Measure Value Chi-Square 463.25, 361 df p<.001 S-B Scaled Chi-Square 405.86, 361 df p=.0516 BBNNFI .960 CFI .967 RMSEA .040 2H The model had an acceptable level of content validity as was discussed earlier. In addition, the pattern of correlations between the constructs provides support for nomological validity. Conflict Frequency and Conflict Intensity are highly correlated (.861) as would be expected. Satisfaction and Conflict Frequency have a negative and significant correlation (—.436) as do Satisfaction and Conflict Intensity (—.437). Satisfaction and Trust also have a strong positive association (.917). All of these associations are logical and are supported by past research findings. Convergent validity of the measurement model is supported because the items “converge” on the predicted construct and have highly significant path coefficients. Table 5.15 (on following page) presents these results. Lastly, this respecified model passes tests of discriminant xnalidity (a problem with the previous measurement model). .A review of the LM test indicates that there are some significant modifications suggested, however there is no theory compelling reason to release the suggested paths. In addition, a series of nested models were considered and 213 under no circumstances did a model that had two constructs with their covariance set equal to one outperform the non- restricted model. These two results support the discriminant validity of the constructs in the model. In summary, the model shows acceptable validity in the four categories considered: (1) content validity, (2) nomological validity (3) convergent validity, and (4) discriminant validity. 214 Table 5.15 - Convergent validity Support for the Constructs within the Respecified Measurement Model T-VALUE STD. CONSTRUCT VARIABLES LAMBDA (robust ) LAMBDA v44 RMCFREQ 1.024 13.756 .905 ’1 "CFREQ ‘v47 LOGCFREQ .816 9.470 .621 0! = -7904 V50 SRTCFREQ .789 9.581 .673 v45 RMCINT 1.031 13.210 .904 F2 ' CINT V48 LOGCINT .780 10.592 .593 a = .7862 v51 SRTCINT .783 11.049 .690 F3 _ crup v46 RMCIMP 1.038 7.731 .789 _ v49 LOGCIMP .820 7.989 .486 a ‘ '6429 v52 SRTCIMP .973 7.990 .606 v19 ASSIT4 1.062 10.434 .734 F4 - IPSA V37 ASSITS .823 8.928 .729 a = .8303 v34 INFOCAP3 .966 8.939 .695 V6 SHARPARl .785 12.111 .735 v43 SHARPARS .763 8.563 .599 F5 _ PART v11 PARTl .991 7.688 .623 _ v12 PART4 1.285 9.572 .738 a ’ '6620 v13 PARTS .950 6.581 .577 v9 CENT 2 1.105 8.919 .668 F6 - DECENT v29 CENT4 1.056 8.865 .606 a = .7439 v33 CENTS .878 7.127 .613 v40 CENT6 1.101 10.463 .713 F8 - CCO V22 CCOl n/a n/a 1.000 v5 SATl 1.308 14.354 .877 F9 - SAT V18 SAT2 1.268 13.035 .896 a = .9292 v32 SAT4 1.189 13.149 .853 v39 SAT5 1.303 11.919 .839 v23 TRUSTl 1.171 11.903 .752 F10 - TRUST V24 TRUST2 .943 9.268 .742 a = .8345 v26 TRUST4 1.218 10.180 .733 v27 TRUSTS 1.153 10.653 .744 215 3.5.4 Conclusions Concerning the Respecified Measurement Model The respecified model performs well and was the model carried forward into the path model evaluations. The model had a more appropriate Constructive Conflict Outcomes variable (based on face validity and research intent grounds), had a new structure construct that did not suffer from discriminant validity problems, and the groups of items used for each construct showed acceptable reliability. In addition, the model showed acceptable fit and passed all the validity assessment tests. Thus, the constructs in this measurement model possess acceptable construct validity support. Table 5.16 (on following page) summarizes the measurement model and its statistical support. 2w Table 5.16 - Summary of the Measurement Results for the Supported Constructs # OF CONSTRUCT ITEMS .ALPHA VALIDITY SUPPORT CONFLICT 3 7904 Convergent & FREQUENCY ' discriminant validity (CFREQ) supported via CFA. CONFLICT 3 7862 Convergent & INTENSITY ' discriminant validity (CINT) supported via CFA. CONFLICT ISSUE 3 6429 Convergent & IMPORTANCE ' discriminant validity (CIMP) supported via CFA. INTEGRATED Convergent & PROBLEM SOLVING 5 6620 discriminant validity APPROACH (IPSA) ' supported via CFA. PARTICIPATION 3 .6839 C9nveig§nt & . . (PART) discriminant validity supported via CFA. DECENTRALIZATION 4 .7439 Cénveigént & . . (DECENT) discriminant validity supported via CFA. CONSTRUCTIVE gintjgt Vgiggfizinant CONFLICT OUTCOMES 1 N/A p? . ' . (CCO) validity support Via the CFA. RELATIONSHIP 4 9292 Convergent & SATISFACTION ' discriminant validity (SAT) supported via CFA. RELATIONSHIP 4 8345 Convergent & TRUST ' discriminant validity (TRUST) supported via CFA. 2U 4.0 TESTING THE STRUCTURAL RELATIONSHIPS In order to test the hypotheses a path model was created and tested. This section outlines that process. Section 4.1 discusses the proposed path model and illustrates how that model represents the hypotheses developed and documented in Chapters Two and Three. Section 4.2 reviews the results of the testing of the path model and includes an assessment of the path coefficients and the overall model fit. The implications of the statistical findings will be discussed in Chapter Six. 4.1 The Path Model In order to test the hypotheses set forth in Chapter Three, a path model was created. The path model is illustrated in Figure 5.3 (on following page). Each path in the model represents one of the hypotheses this dissertation is designed to test. Table 5.17 lists the hypothesis that each path represents. The variables in the model are the same nine as were tested in the repecified model CFA. The measurement support for those constructs is highlighted in Table 5.16. 2m To test the path model the covariances between the factors from the final measurement model were used as the input covariance matrix for the EQS program. The matrix is presented in Table 5.18. This method allowed the path model to incorporate the results of the measurement model tests and thus the path model accounts for the measurement error associated with the constructs (variables) used. 219 Full-variable Path Medel with Standardized Coefficients * Significant at p < .01 CFREQ CINT CIMP V1 V2 V3 a328* IPSA 238* .049 V4 SAT .747 'V8 PART CCO r”’7%fi:" V5 ” V7 2026 .726* DECENT -008 TRUST V6 V9 MODEL FIT MEASURES Chi-Square: 1096.5, p < CFI: RMSEA: .001 28df .324 .417 TOTAL EFFECTS V1 - V8: —.224 V2 - V8: .162 V4 - V8: .510 V7 - V8: .720 V1 - V9: -.232 V2 — V9: .168 V4 — V9: .527 V7 - V9: .744 Figure 5.3 - The Full-variable Path Model rm Table 5.17 — The Paths and the Hypothesis Each Represents Path Hypothesis Represented by the Path V1 V7 Conflict Frequency (CFREQ) will have a significant and positive effect on Constructive Conflict Outcomes (CCO) V2 V7 Conflict Intensity (CINT) will have a significant and negative effect on Constructive Conflict Outcomes (CCO) V3 V7 Issue Importance (CIMP) will have a significant and negative effect on Constructive Conflict Outcomes (CCO) V4 V7 Integrated Problem Solving Approach (IPSA) will have a significant and positive effect on Constructive Conflict Outcomes (CCO) V5 V7 Participation (PART) will have a significant and positive effect on Constructive Conflict Outcomes (CCO) V6 V7 Decentralization (DECENT) will have a significant and positive effect on Constructive Conflict Outcomes (CCO) V7 V8 Constructive Conflict Outcomes (CCO) will have a significant and positive effect on Relationship Satisfaction (SAT) V7 V9 Constructive Conflict Outcomes (CCO) will have a significant and positive effect on Relationship Trust (TRUST) 221 Table 5.18 - The Input Matrix for the Path Model V1 V2 V3 V4 V5 V6 v7 V8 V9 v1 1 CFREQ V2 CINT .861 1 V3 CIMP .067 .129 1 V4 IPSA .343 .439 .280 1 V5 076 094 156 378 1 PART . . . . V6 331 352 307 657 356 1 DECENT ' ' ' ‘ ‘ V7 378 360 263 758 264 516 1 CCO . . . . . . V8 436 437 192 889 277 576 701 1 SAT . . . . . . . (V9 496 479 173 856 295 615 726 917 1 TRUST . . . . . . . . * The matrix is equivalent to a correlation matrix because the variances of the constructs were set equal to one when conducting the CFA (to allow the model to be identified). This had the effect of standardizing the covariance values that were reported in the EQS output. 222 4.2 The Path Model Analysis Results The path model analysis was executed using EQS. The program ran without errors and no condition codes were identified. The presence of either of these would indicate problems with the legitimacy of the EQS output reported. Since the program run was deemed valid, the results are reported in the following sections. The results were highlighted within the previously presented Figure 5.3. Sub-sections 4.2.1 and 4.2.2 discuss the model fit evaluations and the results of the path significance tests respectively. This section includes with Table 5.21 that presents a summary of the results of the hypothesis tests. 4.2.1 Model Fit The path model tested does not fit the data according to the EQS output. As documented in Table 5.19 the fit measures are in agreement: the model lacks fit. Because of this lack of fit, any path tests that arise from the progranlrun have to be interpreted with caution. 2B Table 5.19 — Path Model Goodness-of-Fit Measures GFI Measure value 1096.48, 28 df Chi—Square p<.001 BBNNFI .130 CFI .324 RMSEA .417 There are two logical reasons why the model fit so poorly: 1. There are missing variables in the model. This is highly likely for two reasons. Statistically, the LM test results imply that there are missing variables and the highly significant Disturbance terms also point to this situation. This result is predictable for a second reason; theory. For instance, the dependent variables are concepts that theory links to many other concepts such as dependence, commitment, communication frequency and others. These concepts were not modeled to keep the model simple. Therefore, both the statistics and theory suggest that there are missing variables and thus, it is highly likely that this contributes to the poor model fit. 224 2. There are paths not modeled or existing paths are modeled incorrectly. This is supported by the LM test results. In addition, theory and past research results would predict other paths such as a direct path between conflict frequency and satisfaction (also suggested by the LM test). With these reasons in mind, a great deal of caution is necessary when considering the structural path coefficients. 4.2.2 Structural Path Testing As indicated previously, due to the lack of model fit the structural tests need to be viewed with caution. For instance, the model could be misspecified in a way that causes a path to be significant when it is not (or vice versa). However, the results can provide evidence as to whether a particular path (and thus hypothesis) is supported by the data. The results are presented in Table 5.20. As can be seen from the results highlighted in Table 5.20, some of the paths are significant, while others are not. 225 Section 4.3 connects these statistical results with the appropriate hypotheses, while Chapter Six reviews the implications of these results. 4.3 The Support for the Hypotheses Table 5.21 reviews the evidence of support for each hypothesis made. As documented, there is support for three of the eight hypotheses. Those not supported include some that are deemed to have no significant effect, and some that the data indicates have a reversed effect versus the effect set forth in the hypothesis. The implications of each result are discussed in Chapter Six. 226 Table 5.20 - Path Test Results Path Unstandard- ized Coefficient T-Test value Standard- ized Coefficient V1 - V7 Conflict Frequency - Constructive Conflict Outcomes —.328 -7.800 —.311 V2 - V7 Conflict Intensity - Constructive Conflict Outcomes .238 5.655 .226 v3 - V7 Issue Importance — Constructive Conflict Outcomes .052 1.226 .049 V4 - v7 Integrated Problem Solving Approach - Constructive Conflict Outcomes .747 17.747 .708 V5 - v7 Participation — Constructive Conflict Outcomes —.028 -.661 —.026 'V6 - v7 Decentralization — Constructive Conflict Outcomes .009 -.211 -.008 ‘v7 - V8 Constructive Conflict Outcomes — Relationship Satisfaction .701 15.351 .720 ‘V7 — V9 Constructive Conflict Outcomes — Relationship Trust .726 16.488 .744 ZN Table 5.21 - The Results of the Hypothesis Tests Path Hypothesis Represented by Overall significant and positive effect on Relationship Trust (TRUST) the Path Statistical Finding (3931:“ Frquengf’. (CFREQ) d NOT SUPPORTED V1 V7 ggsitiizeeffzégnclmlcant an The effect found is Constructive Conflict iéggtfizant and Outcomes (CCO) C9311?” Intel‘s 19f”. (CINT) d NOT SUPPORTED V2 V7 ::gati::ee:f::gnclmlcant an The effect found is Constructive Conflict Significant and Outcomes (CCO) pOSltlve' Issue Importance (CIMP) NOT SUPPORTED V3 V7 mull have a Significant and The effect is negative effect on . . . . Constructive Conflict in51gnificant. Outcomes (CCO) Integrative Problem Solving V4 V7 Approach (IPSA) will have a SUPPORTED significant and positive Strong positive effect on Constructive effect found. Conflict Outcomes (CCO) PartiCipation (PART) Will NOT SUPPORTED ‘V5 V7 have a Significant and. The effect is positive effect on . . . . Constructive Conflict inSignificant. Outcomes (CCO) Decentralization (DECENT) NOT SUPPORTED V6 V7 Will have a Significant and The effect is positive effect on . . . . Constructive Conflict inSignificant. Outcomes (CCO) Constructive Conflict SUPPORTED V7 \NB Outcomes (CCO) will have a Strong positive significant and positive effect found effect on Relationship ' Satisfaction (SAT) Constructive Conflict V7 \M) Outcomes (CCO) will have a :EZZSSTSZSitive effect found. ZR 5. STATISTICAL ANALYSIS CONCLUSION In the final conclusion, there was sufficient measurement support for the constructs used, thus allowing for a path model to be constructed that tested the majority of the hypotheses originally proposed. From a measurement standpoint, there was strong support for the constructs used in the final path model, however, further measurement refinement is suggested for the group of constructs that were supposed to make up Relationship Structure and the construct designated to represent Constructive Conflict Outcomes. From a structural standpoint the results indicate only limited support for the model as constructed. The data thus suggest that a different model is needed to capture the correct set of relationships that comprise the antecedents and consequences of Constructive Conflict Outcomes. However, the results from the path model do jprovide a good starting point for building such a model. The inmflications of these statistical results will be discussed in Chapter Six. ZN Chapter 6 Dissertation Conclusions 1 . INTRODUCTION This final chapter reviews the implications of the data analysis and offers conclusions related to the measurement and structural findings. This chapter also reports a set of managerial conclusions. In addition, a set of limitations of the research, as well as a set of future research directions are reviewed. The chapter is organized so each of the five areas, (1) measurement conclusions, (2) structural conclusions, (3) managerial conclusions, (4) limitations, and (5) future research are contained in their own section. Lastly, a final section acts as a conclusion to the chapter and the entire dissertation. 2 . MEASUREMENT CONCLUSIONS As tflnis research was exploratory in nature, and many of the cxnmcepts that were addressed lacked a measurement pedigree, tlmaxneasurement conclusions reviewed in this section are 230 very important. The specific areas reviewed include the measurement properties of the focal variable, Constructive Conflict Outcomes, the variables that were designed to capture relationship structure and the variables used to represent the perceived relationship conflict. Each is discussed in the sub—sections that follow. 2.1 Relationship Structure Conclusions The “relationship structure” configuration proposed by Robicheaux and Coleman (1994) was not supported. Some of the constructs lacked reliability (e.g. formalization, monitoring) while others did not distinguish themselves as distinct concepts (the CCINT — SHARPAR discriminant validity problem). Whatever the reason, the theoretical impact is that there is still no supported group of measures that fully taps the underlying structure of an exchange relationship. Why did these concepts perform poorly? The interorganizational setting. The constructs mainly come from intraorganizational settings where their construct validity has been supported via past research. .However, in order for these constructs to perform as well 231 in the interorganizational setting the respondents would have needed to perceive the exchange relationships in the same way as they might perceive an intra—department relationship. In other words, the exchange relationship would have needed to be an “organization” in its own right. This is probably not the way the average exchange relationship manager perceives the relationships they manage. More likely, there is still a strong perception that the relationship is the meeting of two points (each organization representing one of the points) and not the overlap of two organizations to the point where the relationship has an underlying structure in its own right. Thus, in this research, some level of the relationship as actual organization was captured (measurement support for Participation and Decentralization for instance), but not to the level that would have provided measurement support to the full set of constructs that have proven useful in an intraorganizational setting. The wrong configuration was proposed. A second explanation is that the configuration proposed by Robicheaux and Coleman was not correct. This is supported by the problems with the “operational integration” constructs as a whole. Perhaps a better conceptualization of the level of 232 integration is required before such a concept can be successfully tapped from a measurement standpoint. Even given the poor overall performance, the research still supports the use of relationship structure (or a related concept) to capture the underlying situation in the relationship when building models that attempt to explain how those relationships arrive at positive outcomes such as constructive conflict. The significant effect that Integrative Problem Solving Approach was shown to have on Constructive Conflict Outcomes supports the further refinement of this and other related constructs so that a model that fully specifies the effect of structure on outcomes can be created and tested. Until the measurement performance is improved, the concept of structure will remain a logical antecedent to outcomes, but not a concept with strong empirical support. 2.2 Integrative Problem.Solving Approach The construct titled Integrative Problem Solving Approach shows promise, but at this point lacks full construct validity support. Because of this the results associated with this construct need to be viewed with caution. There 233 are two inherent weaknesses to this construct and its current level of measurement support. The construct is a creation of the data set that then provided the measurement-support test. This construct was not proposed prior to the analysis of the data. Therefore, it may exist only in relation to the particulars of the data collected. Before this construct can be considered confirmed it needs to perform adequately in other relevant data sets and other relevant contexts. This construct’s actual items came from what were originally three separate constructs. The original conceptualization proposed that these questions would not tap the same concept, the data indicated otherwise. Therefore, further research is essential to provide confirmation as to the adequacy of this construct. It performed too well within the path models. This construct had the strongest effect on the outcome variables. There are two explanations for its performance: (1) it is a great construct that really captured an important concept for exchange relationship management, and (2) any first—time construct that performs that well must be tapping other concepts and probably lacks true 234 conceptual distinctness from those. Obviously the first explanation is an optimistic account, while the second is highly pessimistic. Since the construct gained measurement support in the CFA executed, this dissertation takes a position that this construct shows promise, but that further conceptual refinement is warranted and desired before IPSA is allowed to be considered “an important construct.” In other words its current status would be that of “a potentially important construct.” Further research will be required to adequately consider that potential. 2.3 CCO conclusions The original conceptualization of Constructive Conflict Outcomes (CCO) was that of a multi—layered construct that had three valid sub—components. This was not supported. There was basic measurement support for a single, multi— item construct that captured CCO. However, this version of the construct failed to pass a face validity test in relation to it capturing the full range of conflict outcomes (negative to positive). In addition, its nomological validity support was weak as it tended to have weak connections to the other constructs it was supposed to 235 have strong connections to. This normally indicates measurement problems. What then went wrong? The first explanation for these results is simply that the construct is too new and thus still requires further refinement. Only one study documented in the literature review actually had an Operationalized Constructive Conflict Outcomes variable. Most other studies just assumed the presence of the concept without attempting to capture it in the data collected. Thus, the simplest explanation is that this author did a poor job of employing a set of items that would capture this concept. While the total set of items passed a face validity review originally, after the EFA / CFA procedures reduced the number of items, the remaining items in the new statistically valid construct became troublesome from a face validity standpoint. The new collection of items included none of the originally proposed items that were designed to capture the destructive side of conflict outcomes. This was a serious problem as the intended construct was supposed to represent the full range of conflict outcomes, not only those outcomes that were positive. 236 A second possible explanation for these results is that the endpoints of the continuum (highly destructive and highly constructive) actually represent separate variables that can exist simultaneously in the relationship. In other words, the respondents saw nothing inconsistent about perceiving a cluster of questions that captured the destructive endpoint, and a set of questions that captured the constructive endpoint. This scenario is made more likely by the fact that the respondents were asked to assess the conflict that had occurred over the previous year in the relationship (versus a single episode). Thus, that year could have had individual conflict episodes that resulted in both destructive outcomes, and constructive outcomes. Working against this explanation is the fact that while a cluster of items emerged that could be labeled “constructive conflict” (and passed the majority of the standard validity tests), a second grouping did not cleanly emerge to represent a “destructive conflict” construct. There was some measurement support for such a cluster, but not enough to declare the construct valid. This may be because the explanation is off base, or may be because the design of the study did not preconceive of such a result and thus did not create an appropriate scenario that would 237 have allowed that result to emerge. Future study will be required to address this issue. Even though the Constructive Conflict Outcomes variable that ended up being used for the structural tests was the single item measure, the near success of the multi-item Constructive Conflict Outcomes construct was encouraging. This construct had no real measurement history, and this dissertation came close to creating an operationalizable version of it. Further effort will be required, but a good starting point has been set. This construct (based on the use of the single item version in the statistical analysis) does seem to be theory significant, that is, it existed and had relatively strong relationships with other constructs relevant to a business-to-business relationship manager. The further refinement of a multi—item version of this construct will make for stronger tests of the relationships surrounding this concept. 2.4 Conflict Characteristic Conclusions Conflict Focus. The first measurement conclusion is that Conflict Focus, as a construct, shows no promise as operationalized in this study. The concept was derived 238 from Social Conflict Theory, but it completely failed the relevant measurement tests. Therefore, if a future study wishes to incorporate a similar construct it will be forced to create a suitable set of items to capture the concept. Conflict Intensity, Frequency and Issue Importance. Measuring conflict was a popular topic for the marketing literature during the 1970’s and up into the early 1980's. The seminal piece on the topic was Brown & Day (1981). Its conclusion was that a multiplicative combination of conflict frequency, conflict intensity and conflict importance was the best method to capture conflict as a construct. The results of this diSsertation casts doubt on that recommendation. As was shown in the path model, conflict frequency and conflict intensity have opposite effects. This means that a multiplicative combination of these two constructs is not logical because it would only hide and/or confuse the effects of the individual conflict characteristic. The statistical results related to these three characteristics (Intensity, Frequency & Issue Importance) and their relationship with CCO were not fully conclusive. Thus, the exact relationship in and between the three conflict characteristics is not yet confirmed. However, the results do indicate that the effects may be in 239 opposition to each other and thus make a strong case for incorporating these characteristics as separate elements of a theoretical model. The Brown and Day (1981) configuration, where the three characteristics are combined through multiplying their scores, would thus be inapprOpriate. Additionally, if the distinction between the three characteristics is relevant in this study, then there is reason to believe that it would also be relevant in other studies where these three conflict characteristics were used. If this is so, then past studies that used the multiplicative conflict variable took an incomplete and possibly misleading look at the effects of conflict. 3. STRUCTURAL CONCLUSIONS This section reviews the conclusions related to the structural hypothesis tests that were conducted. There is a sub-section that reviews the conclusions related to each hypothesis. The purpose of each sub—section is to attempt to explain why the results occurred and what those results mean for further theory development in this area. %m 3.1 The Overall Effect of Relationship Structure The findings in this area were highly mixed. There is some support for the general contention that relationship structure variables do have an effect on relationship outcomes because of the three structure variables that survived the measurement testing, one had a strong effect on CCO. This finding indicates that before the presence or degree of presence of relationship outcome variables is assessed and interpreted, some consideration needs to be given to the structure of the relationships under consideration. What emerges from integrating these results with those of the numerous relationship studies that preceded this research is that outcome variables are indeed impacted by the structure of the given relationships. Morgan & Hunt (1994), Anderson & Narus (1990), Crosby, Evans & Cowles (1990), Anderson & Weitz, (1989) and others all found some connection between a construct (or set of constructs) that can be considered a “relationship structure” construct and a construct (or set of constructs) considered to be “relationship outcomes”. This dissertation adds to this general finding. 241 However, two problems emerge from this. First, there is disagreement on the exact variables that should be used to capture relationship structure as a concept. The contribution of the results of this dissertation in that area will be discussed a bit later. The second problem is that many different outcome variables have been used and their exact order and position within the models proposed +' also varies. The literature and the specific results of this research indicate that there is a group of interconnected outcomes such as Constructive Conflict Outcomes, Satisfaction, Trust, Commitment and others. In addition, it has been shown that these outcomes are influenced by relationship structure variables. However, the past research has failed to identify the exact structure of the relationships amongst and between the outcome variables themselves. Thus, the question remains, what is their correct order or structure? The studies that have addressed such issues (including this one) are all cross—sectional in design and thus capture a “snap shot” of t:he phenomena, leaving the exact temporal nature of the r‘elationships as an open question. This dissertation pmositioned Constructive Conflict Outcomes as an antecedent of both Trust and Satisfaction (and support for those relationships was found), however, Hunt & Morgan in their 242 1994 piece had Trust as an antecedent of the “functionality of conflict”, a similar set a constructs, but in reverse order. The question then boils down to the chicken & egg question. Unfortunately, unless a set of relationships can be tracked through their development and maintenance, it will be impossible to answer that question definitively. While this problem does limit the full usefulness of the results for one purpose (correctly specifying the exact nature of the connections among the various relationship outcomes) it is not an overly severe theoretical problem for the main purpose of this area of the research. The findings still allow a look at the effect of relationship structure on the constructiveness of the conflict within the relationship. The interconnectedness of Constructive Conflict Outcomes with other relationship outcomes does not preclude making conclusions in this area. 3.2 Decentralization and Constructive Conflict Outcomes The results indicate that Decentralization does not have a direct effect on Constructive Conflict Outcomes. This in turn indicates that the level of decentralization has no 243 effect on the constructiveness of the conflict outcomes. This contradicts the hypothesis that indicated that there would be a significant and positive effect between Decentralization and Constructive Conflict Outcomes. A possible explanation for this is that Decentralization belongs elsewhere in a fully specified model. Given the very poor overall model fit, this explanation cannot be ignored. Further review of the data could help to determine if there is indeed a different path structure that would connect Decentralization with CCO. However, the data does not support the direct effect that was proposed. 3.3 Participation and Constructive Conflict Outcomes The results indicate that Participation does not have a direct effect on Constructive Conflict Outcomes. This in turn indicates that the level of participation has no effect on the constructiveness of the conflict outcomes. This contradicts the hypothesis that proposed that there would be a significant and positive effect between Participation and Constructive Conflict Outcomes. A possible explanation for this is that Participation belongs elsewhere in a fully specified model. Given the very poor overall model fit, this explanation cannot be ignored. 244 Further review of the data could help to determine if there is indeed a different path structure that would connect Participation with CCO. However, the data does not support the direct effect that was proposed. 3.4 Integrated Problem Solving Approach and Constructive Conflict Outcomes The results showed a strong relationship between these two constructs. Theoretically this implies that the approach firms have for handling problems in general is strongly connected to the effectiveness of their conflict resolution. This implies that there is a carry—over effect. Firms that communicate problems, jointly attempt to solve problems, share relevant information in usable ways, and generally share a problem-solving paradigm are also likely to consider the outcomes of conflict to be constructive. Conceptually this is logical because the firms would use the problem solving approach to create Opportunity out of conflict. Whereas in a relationship not characterized by a problem solving approach, it is very likely that conflict would lead to greater hostility and not lead to productive solutions. This would be so because the two firms would lack the environment and past history 245 that would allow them to use the conflict as a way to improve the relationship. 3.5 Issue Importance and Constructive Conflict Outcomes Social Conflict Theory (SCT) proposed a negative relationship between these two concepts. However, the data indicated a non-significant effect between these two constructs. This result could have arisen because the relationships under consideration would disband in the face of severe conflict over a highly important issue. Since these are commercial relationships, there is no requirement that these relationships must continue. Social Conflict Theory is a large group sociology theory that normally is applied to nations, ethnic groups and such. Therefore, it deals with groups that are much less likely to disband in the face of conflict. Thus, while SCT predicts that issue importance will have a negative effect on CCO, in the context of commercial relationships, this relationship may not exist. The non-significant effect could exist because if the parties disagree on a fundamentally important issue, it is likely that the relationship would disband and the parties would find other firms to do business with. In 246 this case, the negative effect would not exist in the data set because there was no data collected from disbanded relationships. Thus within the context of business-to- business exchange relationships Social Conflict Theory may not be an appropriate theory. There is a second possible explanation for the result. The insignificant relationship may be the result of the unit of measurement. That unit was the relationship, whereas it may require a conflict episode unit of measurement to truly capture the relationship between these two constructs. With the unit of measurement used, the actual relationship between Issue Importance and CCO may be obscured because some issues may have the negative relationship predicted, some may have the insignificant effect found, and others could even have a positive effect on CCO. If this indeed the case, the relationship level would not capture these individual effects and thus could lead to confused results. And these confused results could be manifested in the non— significant result that was obtained. A future study would be required to see if this explanation is indeed the correct one. ZN A third explanation for this result has to do with the “professionalism” of the relationship. Because these are commercial relationships with paid relationship managers, the importance of the issue may not have an effect on the constructiveness of the conflict outcomes. The logic behind this explanation is that the manager is paid to handle all conflict, whether over important issues or not, and thus no effect based on this attribute would be expected. There is no way to verify which of these explanations (if any) is correct without future research. However, the data in this project does indicate that a significant, negative relationship does not exist. 3.6 Conflict Frequency and Constructive Conflict Outcomes The hypothesis that proposed a positive effect between Conflict Frequency and Constructive Conflict Outcomes was not supported. The data indicated a significant, negative relationship. The original hypothesis was made because Social Conflict Theory reasons that an organization requires a certain level of conflict to be healthy and thus 248 a relationship that does not exhibit “enough” conflict would actually under—perform a relationship that had greater levels of conflict frequency. Why did the data not support this contention? There are a couple explanations: Social Conflict Theory does not accurately describe the situation when interorganizational exchange relationships are considered. SCT is a large group sociology theory and perhaps the fact that it deals in the aggregate (large groups interacting over periods of time) and deals with aggregate outcomes makes it inappropriate for use in the lesser scope context of an individual exchange relationship. Additional tests in this area would be required to firmly make a conclusion as to SCT’s application to the study of exchange relationships. Another explanation is that the relationships in question exhibited “enough” conflict to be healthy, and the additional conflict found in some of the relationships began to have a negative effect. Thus, the explanation is based on the potential existence of a non—linear effect between these two concepts. Further research could also look into this area. 249 A third explanation is that the respondents don’t perceive “more to be better” when it comes to conflict in the relationships they manage. Most people are conditioned to view conflict to be a negative phenomenon, and thus when perceptions form the basis of measurement it may be difficult to accurately uncover a positive relationship between conflict frequency and constructive outcomes, even if such a relationship does in fact exist. Additionally, since the relationship managers would be intimately involved in the management of conflict, and that conflict can often create tension and stress for those managers, their perceptions of its outcome may be biased. In effect, they may be too close to the conflict and the conflict management, to accurately perceive its positive outcomes. 3.7 Conflict Intensity and Constructive Conflict Outcomes Social Conflict Theory predicted that conflict intensity would have a significant and negative effect on CCO. However, the data indicated a significant, positive effect. There are three possible explanations for this result. First, the result may be a false statistical result caused by the strong correlation between conflict frequency and 250 conflict intensity. A further review of these variables and their effect on CCO is warranted. The simple correlation between Conflict Intensity and CCO is strongly negative, and then in the context of the model the relationship reverses. Thus, there may be some sort of interaction or moderated effect in place that is causing this. Further review of the data, especially in the context of a better fitting model, should help to shed light on this peculiar result. If the result is a legitimate statistical outcome there exists a couple of plausible explanations. First, the parties to the relationship may find constructive outcomes are the result of a “good fight”. It may be that in a highly intense dispute there is the greatest opportunity for positive outcomes. This goes against the reasoning of Social Conflict Theory (SCT) which reasons that a “violent” or intense dispute will be unable to create positive outcomes because it will be too focused on the violence or intensity to deal with the issues that warrant attention. So if the statistical result is legitimate, SCT is not supported. 251 TI A second explanation is that the commercial relationship setting is one where systems have been put in place to handle conflict. It is what the relationship managers are paid to do. Thus, even in the face of intense conflict, the relationship’s conflict resolution system kicks in and produces constructive outcomes. This explanation positions commercial relationships as an improper context for Social Conflict Theory. Before any of these explanations is consider best, further review of the current data will be required and/or additional data will need to be collected and analyzed. Firm conclusions need to be made concerning whether the result is legitimate or a product of the statistical peculiarities of the data. Once that is done, the appropriate conceptual explanation can be determined. Because this is beyond the scope of this dissertation, the significant and positive result between Conflict Intensity and CCO should be viewed with the greatest amount of caution. 252 3.8 Relationship Satisfaction / Trust and Constructive Conflict Outcomes The data indicated a strong relationship between Constructive Conflict Outcomes and both Relationship Trust and Relationship Satisfaction. These were the predicted relationships. The purpose of including these two hypotheses in the overall model was to add a level of credibility to the concept of Constructive Conflict Outcomes. Because CCO was found to be a significant antecedent of these two constructs, it adds to the importance of CCO. With these results, CCO is firmly embedded as an important construct that should be considered when building a model of how exchange relationships handle conflict. The results of this dissertation, while positioning CCO as an important construct, fail to confirm the exact structure of the relationship between CCO and Satisfaction and Trust. The research is a cross—sectional methodology, and thus there is no way to assess if there is a building process between these constructs. It is conceptually logical to assume that these constructs would grow in strength as time passed in a highly productive exchange relationship. Thus, 253 while the results indicate a strong effect by CCO on Satisfaction and Trust, the actual structure could involve a circular building of these closely related outcomes. Only future research can verify this potential theoretical structure. 4. MANAGERIAL CONCLUSIONS Because the overall model fit so poorly, and many of the hypotheses were not supported, this section should be viewed with great caution. There needs to be stronger statistical and theoretical support for the relationships between the focal constructs before managers should use the results to guide their behavior. That said, there is a logical set of managerially relevant findings that should be reviewed. 4.1 Conflict Can be Constructive The results indicate that conflict can be constructive. Support was found for the existence of constructive conflict outcomes. This is important for managers to realize so that they do not adopt strategies or tactics simply designed to avoid conflict, assuming it can only rm -l' "II—m1 hurt the relationship. Hopefully, the extension of this research will help to more firmly identify the situations that are most likely to lead to constructive outcomes, thus giving managers a road map by which to navigate through the relationship conflict. However, the current research results are individually strong enough to suggest that the wise manager should recognize the existence of constructive outcomes and incorporate that recognition into the relationship management strategies he or she employs. 4.2 An Integrated Approach Leads to Greater Constructiveness One part of the roadmap for managers that did emerge from this dissertation is that an integrated approach is strongly connected to constructive conflict outcomes. Thus, managers who wish to produce constructive outcomes should strive to create an integrated relationship management approach. This would entail creating open and effective communication between the firms while also sharing a similar outlook as to what constitutes appropriate relationship behavior. This boils down to firms having the same view of the relationship on two key levels. First, the firms need to be able to view the 255 relationship similarly on a practical information exchange level. In addition, they need to see the relationship similarly from a philosophical level. This two-level, integrated approach was shown to lead to greater levels of Constructive Conflict Outcomes based on the data collected for this dissertation. 4.3 Greater Constructiveness is connected to Greater Satisfaction There was support for the relationship between Constructive Conflict Outcomes and Relationship Satisfaction. Therefore, it makes sense for a relationship manager to use the handling of conflict as a way to improve the other relationship outcomes as well. Thus, from a “why bother” standpoint, the management of conflict becomes relevant because handling it in a way that produces constructive conflict outcomes should also create a situation where overall relationship satisfaction is maximized. Since conflict episodes are real, tangible events, the direct management of them is a controllable factor that warrants management attention. And this management attention should have direct benefits (increased levels of relationship satisfaction) according to the results of this research. 256 5 . LIMITATIONS This section deals with five limitations of the research. These limitations are factors that can limit the usefulness of the findings from either a theoretical or managerial standpoint. In most cases the limitations were the result of explicit choices made by the researcher in relation to data collection, construct measurement and statistical analysis. Had different choices been made, a different set of limitations would have arisen. Thus, the reason these limitations are reviewed is not to invalidate what was done but to highlight how some of the practical research execution decisions that were made create limitations in what was accomplished and create cause for caution when interpreting and applying the results. 5.1 Measurement of Conflict The measurement of conflict at the relationship level is problematic. As was seen in the research results, the data wants to group both by the underlying issue (e.g. Logistics issues) and the underlying characteristic of the conflict (e.g. the frequency). It is difficult to cleanly account for the groups and also cleanly account for the 257 characteristics. This problem created a fairly large amount of measurement error and required that the errors of the conflict characteristic variables be correlated to allow the underlying issues to remain linked. It would have been preferable to have less measurement error associated with these key constructs. Because of this, the results need to be viewed with caution in case the findings are flawed due to the higher than desired measurement error in this area. To overcome this problem, a future study would most likely have to collect data at the conflict episode level so that each data point is only associated with a single conflict issue. This would in essence eliminate the variance for the groupings within a single respondent’s data points, thus allowing the research to focus on the conflict characteristic — conflict outcome relationship. 5.2 Non-dyadic Approach A limitation of this study is that the respondents answered questions concerning the relationship without the help of their counterpart within the other party. Additionally, there was no attempt made to collect data from the other party. Thus, the fact that the relationships are actually 258 dyadic was not captured within the data. (However, the questions were posed with wording that encouraged the respondent to answer on behalf of the “relationship” not just the firm he or she represented.) This means that a relationship that was classified as “decentralized” by the respondent may not have been classified similarly by the other party’s relationship manager. If there would have been many divergent responses in a dyadic approach then different results may have been achieved. Only a future study could test for such problems. The future study though would be able to use the research findings from this study to achieve stronger measurement and to begin with a more reality reflective model. That, in essence, is why this study did not attempt a dyadic approach. The non— dyadic data collection both simplified the research and also allowed for an initial model to be tested. The results from this research give a starting point to future studies, including a dyadic approach study. 5.3 Reliance on Perceptions This research relied on the respondent’s perceptions. Whenever this is the case there is a risk that one person’s perception is vastly different from the next person’s, even 259 though the situation under consideration may truly be similar. Without direct observation, there is not much that can be done about this problem. However, the problem is mitigated by two factors. First, it is likely that a person’s perceptions may be equally biased, thus the association between two factors for two individuals would have the same spread, even if the exact positions the variables occupy on the scale may not be identical. Since this study mainly considers factors based on their underlying correlation, it is the simultaneous movement of (or spread between) factors that is most relevant, versus the exact location on the measurement scales used. The second factor that should act to reduce the negative impact of relying on perceptions is that there is no reason to believe that the biases in the data aren't randomly distributed. In other word, for every respondent that “inflates” their scores, there is likely one that “deflates” their scores. In the end it is likely that these cases cancel each other out and do not fatally bias the entire data set. Therefore, the reliance on perceptions, while creating some risks, does not invalidate the results and did provide a workable and pragmatic way to collect data. 260 5.4 The Cross-Sectional Methodology The data collection is this study is cross-sectional and it is likely that some of the relationships captured in this “snap-shot” arrived at their position through iterations or feedback loops. Such development is not captured in the data. Thus, questions that address such issues as how the relationship arrives at a decentralized structure, or whether there is a point at which a certain amount of past conflict creates a conflict intolerant situation cannot be answered. A related limitation to the research is that the path model methodology, which cannot allow dual direction paths, is inappropriate to capture feedback situations. Thus, it is likely that the style of data collection, combined with the statistical method used fails to capture the presence of feedback loops and/or the effect that time has on the development of the constructs in the model. This limitation is deemed acceptable at this stage in the research program. The current methodology was efficient and allowed for an initial model to be crafted and tested. This model is both useful in the short term (provides usable insights for theory and practice) and in the long term (as a starting point for future studies, including a time sensitive study). 261 5.5 The Skewed Level of Relationship Quality A limitation of the research is that the respondents generally picked very high quality relationships to use as the basis for their observations. This limits the generalization capabilities of this research. However, from a theoretical standpoint, this limitation is not overly damaging because the research was designed to test a theory, and the sample obtained can be considered to provide a valid test of that theory. The design used rests on the philosophy that this research represents a theory test. Therefore, if the theory is supported by a valid sample, then there is support for the theory and thus the theory can be generalized to other relevant or valid settings, whether those were explicitly represented in the sample, or not. While it would be ideal to use a sample that fully represents the population of interest, such samples are most often unobtainable. In addition, most published marketing studies that are intended to test theory fail a strict generalizability test and rely on the "test the theory” method utilized in this research. Therefore, while recognizing this limitation is important, it does not diminish the theory test that occurred, nor does it invalidate the results. 262 From a managerial standpoint, this limitation means that the conclusions are best applied to higher quality relationships. This is where the limitation is of more immediate concern. While the testing of theory in this area is an on—going process involving the results of this research and those of future studies (whereby a body of evidence is created), the managerial usefulness of the study is more time sensitive. Managers would like to be able to use any relevant findings as quickly as possible, and thus in the managerial realm the results should be viewed as representing the high quality relationship condition. Until other studies are done that create a preponderance of evidence in favor of the theory (thus allowing the theory to earn “strongly supported” status and be applied to any relevant context), or other studies are done using other relationship quality contexts (thus expanding the direct generalizability of the results to the populations represented by those samples), the results of this study, from a managerial standpoint, should be considered to directly apply only to higher quality business-to-business relationships. While this limitation is significant, its impact is mitigated by the amount of attention such relationships have received and the reported need to understand and effectively manage such 263 relationships. According to most accounts such relationships have become more desired and more numerous in numbers. Therefore, there are still many relationship managers who can benefit from the managerial conclusions / prescriptions made as a result of this research because it is their responsibility to effectively manage the high quality relationships that their firm has become involved in. 6. FUTURE RESEARCH This section reviews four directions for future research that are suggested by the results of this project. In the first three cases the current data set can offer at least preliminary results on the point. The final direction for future research that is identified would require a new data set. These points are identified to offer direction for additional research that would have the potential to add to the insight gained directly from the research reported in this dissertation. X“ 6.1 An Interaction Effect? The characteristics of conflict and their relationship to the conflict outcomes are an area ripe for research. In the data collected there were few cases that did not have a closely coordinated Conflict Frequency and Conflict Intensity score. A future study could aim to capture situations where there is high frequency and low intensity (or vice versa) and see if such conditions have similar effects. The basic question to answer would be: Do the characteristics interact and if so, is there an interaction effect when considering the level of Constructive Conflict Outcomes? 6.2 An Issue Group Effect? An additional consideration would be looking at this area of research on an issue grouping level. Thus, instead of looking at the cumulative conflict and the cumulative impact of the conflict outcomes, look at the specific groups of issues as separate models. A multiple group path model could be tested first to see if the underlying issue groupings have an impact and then if different models are found to have different results, the research could flfi strive to answer why. What is it about one set of issues over another that causes the different outcomes? And how can managers use such information to improve their chances of managing the conflict successfully? 6.3 Is Relationship Structure a Moderator? A third way to consider the constructs used in the model is to manipulate relationship structure so to be able to test its role as a moderator, not an antecedent. This research considered the structure of the relationship to have a direct impact on the outcomes of conflict. A second way to consider the impact of structure is as a moderator. Does the model that connects the conflict characteristics to the conflict outcomes perform differently given different conditions of relationship structure? This would Offer a competing model to the direct effect model tested in this dissertation. 6.4 Are the Results the same at the Conflict Episode Level? A final direction for future research is to tie the constructiveness of the conflict outcome to a specific episode of conflict. In other words change the unit of 266 analysis to the conflict episode, versus the relationship as a whole (the unit of analysis in this study). This would enable the research results to point to more specific situations that result in constructive outcomes. This episode—level research would be a strong compliment to the relationship- level results obtained in this dissertation. 7. CONCLUSIONS This dissertation attempted to create and test a theoretical model that described how conflict could lead to constructive outcomes. Social Conflict Theory and marketing channel theory were combined to create a model that identified a set of antecedents and consequences of Constructive Conflict Outcomes. A group of relationship managers were questioned and a theory test was executed. This theory test occurred at two levels: the measurement level and the structural level. The measurement tests revealed that the concept of constructive conflict was a relevant concept for the respondents. A basic level of measurement support was found for a Constructive Conflict Outcomes (CCO) construct. However, the data also indicated that further measurement fifl refinement is necessary before there is full support for a multi-item CCO construct. Such a construct is desirable and the results of this dissertation provide a starting point for the creation of that construct. The measurement tests also revealed that the relationship structure configuration proposed by Robicheaux and Coleman (1994) did not successfully describe the structure of the interorganizational exchange relationships that were the units of measurement in this dissertation. This result means that the correct way to describe the underlying structure of such relationships remains an open question. The structural testing indicated that the model proposed was highly deficient. The model fit statistics indicated that it was highly likely that there were missing variables and mis-specified paths in the model. While this result was disappointing, the dissertation research did provide an interesting set of conclusions. First, the hypotheses made based on the tenets of Social Conflict Theory were not supported. This indicates that this theory does not hold great potential for explaining the role of conflict in exchange relationships. In addition, although the structure configuration of Robicheaux and Coleman was not supported by the measurement tests, the structural tests 268 indicated that there is value in assessing the structure when considering the constructiveness of conflict outcomes. An integrated approach was shown to have a strong positive effect on the constructiveness of conflict outcomes. This result warrants further attention with special attention being given to identifying other structure variables that may have a direct, or indirect effect on CCO. Again, the dissertation does not provide firm answers, but does offer a valuable starting point to that quest. In the final conclusion, the dissertation made a contribution to the building of a productive, theoretical model that should eventually describe when and how conflict leads to constructive outcomes, and how those outcomes can also lead to other, more general relationship outcomes. The dissertation made the majority of its contributions to this task by showing what does not work. However, these results are still highly valuable and provide an excellent spring—board to further research that will hopefully create and confirm such a model in a more definitive way. 269 APPENDICES 270 APPENDIX A1 Construct Conceptualizations and the Items Used in Questionnaire 271 CLAN-LIKE STRUCTURE (CEJU!) Robicheaux.& Coleman (1994) conceptualize this construct as ranging from “Clan-like” to “Bureaucratic” Clan-like: Common sense of well—being among system members and shared values. Decision making is decentralized and informal communications are common. Members are more likely to subordinate their short— term individual self-interests to those of the system and, through participatory governance, compromise is common (p. 47). Bureaucratic: Relationship is governed by rules and bureaucratic machinations. Power and influence strategies are key determinants of outcomes. There is an absence of shared values and there is little if any community of interests (p. 47). Conceptualized by Robicheaux & Coleman (1994) to be indicated by four dimensions: (1) formalization (2) centralization (3) participation (4) shared paradigm 272 1. . FORMALIZATION 8 The extent to which explicit rules & procedures regulate decisions made concerning the relationship. Considering the relationship you identified, please respond to the following statements: FORMl . FORMZ . FORM3. FORM4 . FORMS. FORM6 . Strongly'agree 1 2 Based on: When decisions are made concerning this relationship, the parties rely on clearly specified operating procedures. Existing rules and procedures are often ignored when relationship decisions are made (R). The parties refer to written and/or verbal instructions to guide decision making. Formal, written contracts guide relationship decision making. When making a relationship decision, the parties fall back on standard courses of action. When faced with a relationship decision, the parties rely on behavior dictated by formalized roles and responsibilities. Strongly'disagree 3 4 5 6 7 Dwyer & Oh 1987 Dwyer & Welsh 1985 John 1984 Reve & Stern 1986 273 2 . CENTRALIZATION : The extent that decision.making authority is concentrated. Considering the relationship you identified, please respond to the following statements: CENTl . CENT2 . CENT3 . CENT4 . CENTS . CENT6 . Strongly'agree 1 2 Based on: Within this relationship, decision—making authority is concentrated in a few individuals. If a decision needs to be made, the decision is often made by those directly involved in the situation, and does not need the approval of superiors (R). A small percentage of the individuals involved in this relationship set the relationship policies. Superiors must review all important decisions before the decisions take effect. The frontline employees involved in managing the relationship are free to make decisions involving this relationship (R). There can be little action taken until relationship decisions have been approved through proper channels. Strongly'disagree 3 4 5 6 7 Hage & Aiken 1967 Dwyer & Welsh 1985 Dwyer & Oh 1987 Reve & Stern 1986 John 1984 274 3. PARTICIPATION: The degree of actual participation in decision making Consider the following decision making areas, and indicate the amount of mutual participation that is present for each decision in the relationship. PARTl. Delivery arrangements and terms PART2. Product/service design changes PART3. Length of contract/agreement PART4. Ordering procedures PARTS. Product/service prices and/or margins PART6. Scheduling of operational activities PART7. Demand and/or supply requirement forecasting The decision is made...... Entirely by Equally by One party both parties 1 2 3 4 5 6 7 Based on: Hage & Aiken 1967 Dwyer & Welsh 1985 Dwyer & Oh 1987 25 4 . SHARED PARADIGM 8 The extent to which the exchange partners share general assumptions and values related to conduct in the relationship. Considering the relationship you identified, please respond to the following statements: SHARPARl. The parties often agree on the appropriate course of action for this relationship. SHARPAR 2. Both firms share the same values. SHARPAR 3. The parties often disagree on what is in the best interest of the relationship (R). SHARPAR 4. The parties have different expectations for this relationship (R). SHARPAR 5. The parties often agree on what constitutes appropriate relationship behavior. Strongly'agree Strongly disagree 1 2 3 4 5 6 7 Based on: Measures are new but based on the conceptualizations of: Wilkins & Ouchi, 1983 Robicheaux & Coleman, 1994 276 OPERATIONAL INTEGRATION (OPINT) Robicheaux & Coleman (1994) conceptualize this construct as ranging from.“Discrete” to “Integrated” The authors do not give elaborate descriptions of these end points, but instead refer to the components of integration in order to describe these states. Thus, a “integrated” relationship would exhibit a high level of joint action, assistance, monitoring, and information sharing. There are four components to operational integration: (1) joint actions (2) assistances (3) monitoring (4) information sharing 277 .... 1. JOINT ACTIONS: The degree of interpenetration of organizational boundaries. Consider the following relationship activities and indicate the degree of joint action that characterizes the execution of these activities n this relationship. JACTl. forecasting JACT2. component / product testing JACT3. personnel training JACT4. product design and development JACTS. promotional activity JACT6. inventory control This activity involves.... No Joint Action Extensive Joint Action 1 2 3 4 5 6 7 Based on: Heide & John, 1990 TH 2. ASSISTANCES: The extent to which help is given in the relationship even when such help is not specifically called for by the preset standards of the relationship, and/or, when there is no immediate compensation given in return for such help. Considering the relationship you identified, please respond to the following statements: ASSITl. ASSITZ. ASSIT3. ASSIT4. ASSITS. Strongly'agree 1 2 Based on: The parties in this relationship advise each other in advance of potential operational problems (such as delivery delays or stock outs). The parties help each other in emergencies. Unless payback is immediate, help is not offered to the other party (R). The parties help each other in value analysis, cost reductions, problem solving, etc. The parties advise each other of potential problems in meeting each other's needs. Strongly'disagree 3 4 S 6 7 Noordewier, John and Nevin, 1990 In 3. MONITORING: The amount of monitoring or supervisory actions that the exchange partners undertake to ensure partner performance during the execution of the exchange relationship. Based on the relationship you have with this party please respond to the questions below: MONITORl. The parties spend little time checking on each other's performance (R). MONITORZ. The parties monitor each other’s inventory levels. MONITORB. The parties assess relationship performance through a formal evaluation program. MONITOR4. The parties share performance reports regularly. MONITORS. The relationship makes use of many controls. Strongly'agree Strangly'disagree l 2 3 4 5 6 7 Based on: Noordewier, John and Nevin, 1990 am 4. INFORMATION SHARING The extent of information sharing in the relationship. Characterized by the frequency, type and ease of information sharing. Frequency of Information Exchanged. INFOFREl. INFOFREZ. INFOFRE3. Communication between the parties happens often. The parties frequently share information. Information/data is exchanged Infrequently (R). Type of Information Exchanged. INFOTYPl. INFOTYPZ. INFOTYP3. INFOTYP4. INFOTYPS. Strategic information is readily shared in this relationship. Production plans are readily shared. Long-range forecasts are readily shared. Impending design changes are readily shared. Financial planning information is rarely shared in this relationship (R). Exchange Capability (Ease & USefulness). INFOCAPl. INFOCAPZ. INFOCAP3. Strongly'agree l 2 Based on: Information shared between the parties is easily understood and used by both parties. Although information is shared between the parties it is often unusable because it is in a format that cannot be understood by the other party (R). The parties are able to exchange data or information easily and in usable formats. Strongly'disagree 3 4 5 6 7 Noordewier, John and Nevin 1990 Robicheaux & Coleman 1994 Bowersox , et al 1995 281 CONSTRUCTIVE CONFLICT OUTCOMES (CCO) Conflict has the potential to create positive outcomes. Cosar (1956) considers conflict to be a necessary component in all healthy social systems. Business relationships form a social system, and thus, if Cosar’s predictions are correct, the business system should be characterized by “healthy” outcomes that were the result of system conflict. That is what constructive conflict is: the extent to which the relationship has exhibited healthy outcomes in response to conflict. Cosar (1956) predicts three specific categories of constructive outcomes; (1) the release of system hostilities before they grow unmanageable and destructive, (2) the adjustment of relationship norms, and (3) the beneficial adjustment of the activities that characterize the relationship. CCO is conceptualized to be characterized by three components : (1) Hostilities Release (2) Norm Adjustment (3) Activity Adjustment 282 ‘i‘tflJ-‘L‘: -.. u.‘ 1. HOSTILITIES RELEASE: The extent to which conflict has allowed the release of anger and hostility between the parties before such anger has boiled over and caused destructive outcomes. Based on the relationship you have with this party please respond to the questions below: HOSTRELl. The relationship conflict has often cleared the air between the parties. HOSTRELZ. The conflict that occurs helps to defuse the more serious and destructive conflict. HOSTRELB. Conflict has allowed the parties to vent frustration, but hasn’t really hurt the relationship. HOSTREL4. When meetings occur to discuss conflict, the parties often leave feeling even more upset. (R) HOSTRELS. If the past had more open conflict, the parties would be less frustrated with each other now. Strongly agree Strongly disagree 1 2 3 4 5 6 7 Based on: No literature history for this construct, however the conceptualization comes from: Cosar, 1956, 1967 Turner, 1986 NB 2. NORM ADJUSTMENT: The extent to which conflict has initiated beneficial change in the norms that govern the relationship Based on the relationship you have with this party please respond to the questions below: NORMADl. Conflict has caused the parties to have a diminished working relationship (R). NORMADZ. The roles each party is to suppose to play in the relationship have been more accurately defined in response to conflict. NORMADB. Because of the relationship conflict, the two parties have come to a better understanding of what is, and what isn’t, acceptable relationship behavior. NORMAD4. When changing conditions have caused friction, the parties have been able to make the necessary adjustments to restore balance to the relationship. Strongly'agree Strongly'disagree 1 2 3 4 5 6 7 Based on: No literature history for this construct, however the conceptualization comes from: Cosar, 1956, 1967 Turner, 1986 MM 3. ACTIVITY ADJUSTMENT: The extent to which conflict has initiated beneficial change in the relationship activities. Based on the relationship you have with this party please respond to the questions below: ACTADl. Conflict has tied up many resources that could have been used more productively (R). ACTAD2. Conflict has pushed this relationship to higher performance levels. ACTAD3. Conflict has made us review past actions and make improvements. ACTAD4. Conflict has forced the parties to come up with innovative solutions to relationship problems. ACTADS. After conflict has been resolved, the relationship returns to the status quo without making the changes required to avoid the reoccurrencb'of the conflict (R). Strongly'agree Strongly'disagree 1 2 3 4 5 6 7 Based on: No literature history for this construct, however the conceptualization comes from: Cosar, 1956, 1967 Turner, 1986 $5 CONFLICT FREQUENCY, INTENSITY E IMPORTANCE In order to operationalize these characteristics of conflict, the following list of twenty issues is offered. Respondents will be asked to indicate how frequently there is conflict over each issue, how intense the conflict is over each issue, and how important each issue is to the relationship. Minimum order quantities Payment terms Product quality Level of sales support Level of service support Advertising support Performance objectives Terms of the contract Inventory policies 10. Warranty policies 11. Delivery terms \OCDQCLUTID-UUNH 12. Delivery performance 13. Communication of problems 14. Performance expectations 15. Frequency of communication 16. Strategic objectives 1?. Control of resources 18. Sharing of market information 19. Design issues 20. Price and/or margins 21. Ease of communication 286 CONFLICT FREQUENCY, INTENSITY & IMPORTANCE 1. FREQUENCY: For each issue, respondents will consider the following statement: Conflict over this issue occurs: never Very Frequently l 2 3 4 5 6 7 2. INTENSITY: Respondents will be asked to respond to the following statement: Conflict over this issue is: net Very Intense Very Intense 1 2 3 4 5 6 7 3. IMPORTANCE: Respondents will be asked to respond to the following statement: In this relationship, this issue is: Not very Important very Important 1 2 3 4 5 6 7 Based on: Brown and Day, 1981 NW CONFLICT FOCUS: The extent to which the conflict in the relationship focuses around operational issues versus strategic or fundamental issues. Based on the relationship you have with this party please respond to the questions below: CFOCUSl. The conflict in this relationship has focused mostly on operational issues. CFOCUSZ. The parties have disagreements over fundamental issues (R). CFOCUS3. Conflict in this relationship centers on how to achieve goals, not the goals themselves. CFOCUS4. Much of the conflict centers on disagreements over the basic goals of the relationship (R). CFOCUSS. The relationship is characterized by strong disagreements over strategic issues (R). Strongly agree Strongly'disagree 1 2 3 4 5 6 7 Based on: No literature history for this construct, however the conceptualization comes from: Cosar, 1956, 1967 mm RELATIONSHIP SATISFACTION: The extent to which the party is satisfied with the performance of the relationship (relationship has lived up to the performance expectations set for it). Based on the relationship you have with this party please respond to the questions below: SATl. SATZ. SAT3. SAT4. SATS. Strongly'agree 1 2 Based on: We are very satisfied with this relationship. This relationship meets or exceeds our expectations. This relationship has performed poorly (R). Based on the objectives we hold for this relationship, we are very satisfied with the performance of this relationship. This relationship has not performed up to our expectations. (R) Strongly'disagree 3 4 5 6 7 Brown, Lusch & Smith, 1990 NE RELATIONSHIP TRUST: Confidence in the other party's reliability and integrity. Considering the relationship you identified please respond to the following statements: TRUSTl. TRUSTZ. TRUSTB. TRUST4. TRUSTS. Strongly'agree 1 2 Based on: This firm does not keep our interests in mind when making decisions or taking actions (R). This firm keeps the promises it makes to our firm. We believe the information this party provides us. We find it necessary to be cautious with this firm (R). The other party is not always honest with us (R). Strongly’disagree 3 4 5 6 7 Morgan & Hunt, 1994 Doney & Cannon, 1997 rm Appendix A2 Construct Conceptualizations and Final Items Used in the Analysis 291 CLAN-LIKE STRUCTURE (CHEM!) Robicheaux 5 Coleman (1994) conceptualize this construct as ranging from “Clan-like" to “Bureaucratic" Clan-like: Common sense of well-being among system members and shared values. Decision making is decentralized and informal communications are common. Members are more likely to subordinate their short-term individual self-interests to those of the system and, through participatory governance, compromise is common (p. 47). Bureaucratic: Relationship is governed by rules and bureaucratic machinations. Power and influence strategies are key determinants of outcomes. There is an absence of shared values and there is little if any community of interests (p. 47). Conceptualized by Robicheaux & Coleman (1994) to be indicated by four dimensions; (1) formalization (2) centralization (3) participation (4) shared paradigm 292 1 . FORMALIZATION : The extent to which explicit rules & procedures regulate decisions made concerning the relationship. This construct was dropped from the final model. Although the questions were nearly verbatim from previous literature (where their construct validity was supported) they showed very poor reliability and were troublesome in the factor analysis procedures (both EFA & CFA). The question of the theoretical significance of this construct will remain unanswered by this dissertation. The “decision making structure” construct will be respecified without a “informalization” sub-component. RELIABILTY: Alpha = .5870 (with just items 3-4-5-6 alpha = .5983 and this still fails to meet the .7 cutoff point suggested by Nunnally 1978) 293 2. CENTRALIZATION: The extent that decision making authority is concentrated. Considering the relationship you identified, please respond to the following statements: Final Questions Used (numbers from original system) CENTZ. If a decision needs to be made, the decision is often made by those directly involved in the situation, and does not need the approval of superiors (R). CENT4. Superiors must review all important decisions before the decisions take effect. CENTS. The frontline employees involved in managing the relationship are free to make decisions involving this relationship (R). CENT6. There can be little action taken until relationship decisions have been approved through proper channels. Reliability: .7439 Based on: Hage & Aiken 1967 Dwyer & Welsh 1985 Dwyer & Oh 1987 Reve & Stern 1986 John 1984 NM 3. PARTICIPATION: The degree of actual participation in decision making Consider the following decision making areas, and indicate the amount of mutual participation that is present for each decision in the relationship. Final Questions Used (numbers from original system) PARTl. Delivery arrangements and terms PART4. Ordering procedures PARTS. Product/service prices and/or margins PART6. Scheduling of operational activities Reliability: .6839 Based on: Hage & Aiken 1967 Dwyer & Welsh 1985 Dwyer & Oh 1987 $5 4. SHARED PARADIGM: The extent to which the exchange partners share general assumptions and values related to conduct in the relationship. Considering the relationship you identified, please respond to the following statements: Final Questions Used (numbers from original system) SHARPARl. The parties often agree on the appropriate course of action for this relationship. SHARPAR 2. Both firms share the same values. SHARPAR 3. The parties often disagree on what is in the best interest of the relationship (R). SHARPAR 4. The parties have different expectations for this relationship (R). SHARPAR 5. The parties often agree on what constitutes appropriate relationship behavior. Reliability: . 8130 Based on: Dfleasures are new but based on the conceptualizations of: Wilkins & Ouchi 1983 Robicheaux & Coleman 1984 NM I OPERATIONAL INTEGRATION (OPINT) Robicheaux & Coleman (1994) conceptualize this construct as ranging from.“Discrete" to “Integrated" The authors do not give elaborate descriptions of these end points, but instead refer to the components of integration in order to describe these states. Thus, a “integrated” relationship would exhibit a high level of joint action, assistance, monitoring, and information sharing. There were four components originally specified to make up operational integration: (1) joint actions (2) assistances (3) monitoring (4) information sharing This construct was respecified to have the following two components: (1) Relationship Management Integration (2) Logistical Integration A question from each Of the original 4 factors remains in the new structure and thus preserves the original concept to a certain extent. 297 1. COMMUNICATION E COORDINATION INTEGRATION: The level of integration of the activites that enable effective communication and coordination within the relationship. Final Questions Used (numbers from original system) ASSITl. The parties in this relationship advise each other in advance of potential operational problems (such as delivery delays or stock outs). ASSIT4. The parties help each other in value analysis, cost reductions, problem solving, etc. ASSITS. The parties advise each other of potential problems in meeting each other’s needs. INFOTYPl. Strategic information is readily shared in this relationship. INFOCAPl. Information shared between the parties is easily understood and used by both parties. INFOCAP3. The parties are able to exchange data or information easily and in usable formats. Reliability: .8334 Based on: New construct creation within dissertation mm 2. TRANSACTIONAL INTEGRATION: The level of the integration of the activities that facilitate efficient transactions within the relationship Final Questions Used (numbers from original system) JACTl. There is a great degree of joint action involved in forecasting MONITORZ. The parties monitor each other’s inventory levels. INFOTYPZ. Production plans are readily shared. INFOTYP3. Long-range forecasts are readily shared. Reliability: .7245 Based on: lNew construct creation within dissertation fl” CONSTRUCTIVE CONFLICT OUTCOMES (CCO) Conflict has the potential to create positive outcomes. Cosar (1956) considers conflict to be a necessary component in all healthy social systems. Business relationships form a social system, and thus, if Cosar’s predictions are correct, the business system should be characterized by “healthy” outcomes that were the result of system conflict. That is what constructive conflict is: the extent to which the relationship has exhibited healthy outcomes in response to conflict. Cosar (1956) predicts three specific categories of constructive outcomes; (1) the release of system hostilities before they grow unmanageable and destructive, (2) the adjustment of relationship norms, and (3) the beneficial adjustment of the activities that characterize the relationship. Constructive Conflict was originally conceptualized to be characterized by three components: (1) Hostilities Release (2) Norm Adjustment (3) Activity Adjustment This construct was respecified in two ways. First, a five item.first-order construct was created via questions from the original three sub-constructs. Second, single-item, continuumrbased question capturing destructive to constructive conflict outcomes was called into action. 300 1. CONSTRUCTIVE CONFLICT OUTCOMES: The extent to which the relationship has exhibited healthy outcomes in response to conflict. Based on the relationship you have with this party please respond to the questions below: The final questions used for the CCO Construct: HOSTRELl. The relationship conflict has often cleared the air between the parties. NORMADZ. The roles each party is to suppose to play in the relationship have been more accurately defined in response to conflict. NORMAD3. Because of the relationship conflict, the two parties have come to a better understanding of what is, and what isn’t, acceptable relationship behavior. ACTAD2. Conflict has pushed this relationship to higher performance levels. ACTAD4. Conflict has forced the parties to come up with innovative solutions to relationship problems. The Single Global Measure of Conflict Outcomes: Reflecting on the conflict that has occurred between the two parties, I would characterize the overall impact of that conflict as: Destructive Constructive (hindered & hurt relationship) (improved relationship) 1 2 3 4 5 6 7 301 CONFLICT FREQUENCY, INTENSITY E IMPORTANCE In order to operationalize these characteristics of conflict, the following list of twenty issues is offered. Respondents will be asked to indicate how frequently there is conflict over each issue, how intense the conflict is over each issue, and how important each issue is to the relationship. KDCDQONU'lIb-UJNH Minimum order quantities Payment terms Product quality Level of sales support Level of service support Advertising support Performance objectives Terms of the contract Inventory policies Warranty policies Delivery terms Delivery performance Communication of problems Performance expectations Frequency of communication Strategic objectives Control of resources Sharing of market information Design issues Price and/or margins Ease of communication MD CONFLICT FREQUENCY, INTENSITY & IMPORTANCE 1. FREQUENCY: The average frequency score was taken for the 21 issues. This score will act as a single question factor in the model. INTENSITY: The average intensity score was taken for the 21 issues. This score will act as a single question factor in the model. IMPORTANCE: The average issue importance score was taken for the 21 issues. This score will act as a single question factor in the model. CONFLICT FOCUS: All the original questions will be used and the average score will act as a single question factor in the model. The questions are below: CFOCUSl. The conflict in this relationship has focused mostly on operational issues. CFOCUSZ. The parties have disagreements over fundamental issues (R). CFOCUS3. Conflict in this relationship centers on how to achieve goals, not the goals themselves. CFOCUS4. Much of the conflict centers on disagreements over the basic goals of the relationship (R). CFOCUSS. The relationship is characterized by strong disagreements over strategic issues (R). MD RELATIONSHIP SATISFACTION: The extent to which the party is satisfied with the performance of the relationship (relationship has lived up to the performance expectations set for it). Final Questions Used (numbers from original system) SATl. We are very satisfied with this relationship. SAT2. This relationship meets or exceeds our expectations. SAT3. This relationship has performed poorly (R). SAT4. Based on the objectives we hold for this relationship, we are very satisfied with the performance of this relationship. SATS. This relationship has not performed up to our expectations. (R) Reliability: .9304 Based on: Brown, Lusch & Smith 1990 MM I RELATIONSHIP TRUST : Confidence in the other party's reliability and integrity. Final Questions Used (numbers from original system) TRUSTl. This firm does not keep our interests in mind when making decisions or taking actions (R). TRUST2. This firm keeps the promises it makes to our firm. TRUST3. We believe the information this party provides us. TRUST4. We find it necessary to be cautious with this firm (R). TRUSTS. The other party is not always honest with us (R). Reliability: .8581 Based on: Morgan & Hunt 1994 Doney & Cannon, 1997 305 APPENDIX B The Pretest Questionnaire 306 The pretest questionnaire below has been reduced in size to fit the requirement of the dissertation. It originally was sent to the respondents in its full 8.5 x 11 inch format. Michigan State University - Relationship Management Survey Thank you for your cooperation. Please answer all questions and retum by fax to (517-432-1112) Please select a single exchange relationship (external to your firm) that your firm has with a supplier or customer where you have been significantly involved in the management of that relationship for at least a year. The relationship can involve either a good or service. Respond based on this relationship. panics necessary relums [0 status quo Thank you for the time, please fax the completed questionnaire to 517-432-1112 307 Appendix C The Full-Survey Questionnaire 308 Each page of the survey has been reduced in size in order to conform to the margin requirements of this document. The survey, when used, was an eight-page (8.5x11) document. Each of those pages is included in this appendix. Michigan State University Relationship Management Survey Thank you for your cooperation. Please note the following instructions: Please complete the entire questionnaire Please mark only one answer per question When completed. fold this survey so that the back cover forms the outside. staple or tape shut. and drop in the mail. Postage is prepaid. mid—- If you would like access to a detailed summary of the research findings please include your e-mail address here: If you do not have an e-mail address, please include an address or fax number Address: Fax #: Attention: Only the primary researchers will have access to this information, and your responses will only be reported as part of the aggregate data (you will not be individually identified in any research report). BEGIN HERE: Please select an exchange relationshifi that your firm has with a supplier or customer that meets the following criteria: 1. You have had significant management responsibility for this relationship for a period of at least one year, and 2. A good or service is exchanged, and 3. The relationship is external to your company, and 4. The relationship is a United States based relationship. Please complete this questionnaire based on the characteristics of the single relationship you have just selected. ‘ An “exchange relationship“ is defined as the relationship between two firms de5igncd to facilitate the exchange of a product or scrvxce (c. g. Buyer-Seller. Supplier-Manufacturer. Manufacturer-Retailer. etc.) 309 Please complete this questionnaire based on the characteristics of the relationship you selected (see cover for selection criteria). The other party would be considered our: Customer Supplier How long has your firm had a relationship With this other party? Years and Months What percent of purchases / sales (within this product category) does this relationship represent to your firm? How many face to face meetings occur in this relationship? How often is product exchanged in this relationship”? times per month/week (circle one) _ Daily __ Weekly _Monthly How many years do you perceive this relationship will continue? Are you actively looking to replace this relationship with a relationship with another firm 1’ __ YES _ NO In terms ofoverall quality.lwould rate this relationshipasone of my firms... Best 1 2 3 4 5 6 7 Worst Has there been any conflict between the parties of this relationship over the past year YES NO The arrivity involves. .. Consider the followmg relations/tip activities and indicate the degree of joint action No Join! Evenswr that characterizes the execution of these activities in this relationship. ACHO'I‘ 10"" damn“ Forecasting l 2 3 4 5 6 7 Component / product testing I 2 3 4 5 6 7 Personnel training 1 2 3 4 5 6 7 Product desngn and development 1 2 3 4 5 6 7 Promotional Activates l 2 3 4 5 6 7 Inventory Control l 2 3 4 5 6 7 ‘ one firm executes activuy independent of other firm “ Both firms contribute Significantly to actiVity execution Considering the relationship you identified above, lease res nd to the followin statements' smug." Strongly p p0 g ' Disagree Agree When decmons are made concerning this relationship. the parties rely on clearly speCified l 2 3 4 5 6 7 operating procedures. When conflict arises in this relationship the parties keep each other‘s interest in mind 1 2 3 4 5 6 7 when looking for solutions to that conflict. Conflict has caused the parties have a diminished working relationship. 1 2 3 4 5 6 7 The conflict in this relationship has focused mostly on operational issues. 1 2 3 4 5 6 7 Within this relationship. decision-making authority is concentrated in a few individuals. 1 2 3 4 S 6 7 The parties in this relationship adVise each other in advance of potential operational l 2 3 4 5 6 7 problems (such as delivery delays or stock outs). Strategic information is readily shared in this relationship. 1 2 3 4 5 6 7 In the process of conflict resolution. one party normally seeks to become better off at the l 2 3 4 5 6 7 expense of the other party. We are very satisfied with this relationship. 1 2 3 4 S 6 7 310 ‘. Based on the relationship you identified, please respond to the following statements: 3:225: 532:: The parties often agree on the appropriate course of action for this relationship. 1 2 3 4 5 6 7 Communication between the parties happens often. I 2 3 4 S 6 7 The parties generally approach conflict resolution with the same goals. 1 2 3 4 5 6 7 The roles each party is suppose to play in this relationship have been more accurately l 2 3 4 S 6 7 defined in response to conflict. The parties spend little time checking on each other's performance. 1 2 3 4 5 6 7 The relationship conflict has often cleared the air between the parties. 1 2 3 4 5 6 7 Existing rules and procedures are often ignored when relationship decisions are made. 1 2 3 4 5 6 7 Information shared between the parties is ea5ily understood and used by both parties. 1 2 3 4 5 6 7 If a decision needs to be made. the decisron is often made by those directly involved in the l 2 3 4 5 6 7 situation. and does not need the approval of superiors. Both firms share the same values. I 2 3 4 5 6 7 Consider the following decision making areas. and indicate the amount of . The decrston ‘5 made... . . . . _ . . . Entirely by Equally by mutual pamCipation that is present for each type of relationship decmon. . one parrv both parties Delivery arrangements and terms 1 2 3 4 5 6 7 Product/service deSign changes I 2 3 4 5 6 7 Length of contract/agreement l 2 3 4 5 6 7 Ordering procedures 1 2 3 4 5 6 7 Product/service prices or margins l 2 3 4 5 6 7 Scheduling of operational activnies l 2 3 4 5 6 7 Demand and/or supply requirement forecasting l 2 3 4 5 6 7 Please circle the one industry category that best describes your firm. 1. Appliances 6. Computer 11. Food & Beverage l6. Mining/Minerals 21. Primary Metals 22. Automotive 7. Construction equip. 12. Furniture 17. Office Equipment 22. Rubber Products 3. Building Materials 8. Department Store 13. Hardware 18. Paper Products 23. OTHER. please 4. Chemicals/plastics 9. Electrical Machinery [4. Machine Tools 19. Petroleum Products specify 5. Clothing/textiles 10. Electronics 15. Metal Products 20. Pharmaceuticals Please respond to the following demographic items: Your Firm Other Party Size. in annual sales dollars. Years in business Number of employees 311 Overall, I would describe this relationship as one characterized by: Little Tnist l 2 3 4 5 6 7 Little Commrtment l 2 3 4 5 6 Little Interdependence l 2 3 4 5 6 Great Trust Great Commitment Great Interdependence Based on the relationship you identified, please consider the following statements: Strongly Strongly Disaster Asa. Production plans are readily shared. 1 2 3 5 6 7 Conflict has tied up many resources that could have been used more productively. 1 2 3 5 6 7 The parties help each other in emergencies. l 2 3 5 6 7 When conflict occurs. one party often tries to come out on top. 1 2 3 5 6 7 The parties have disagreements over fundamental issues. 1 2 3 5 6 7 The parties monitor each other's inventory levels. 1 2 3 5 6 7 The parties frequently share information. 1 2 3 5 6 7 The parties look out for each other when resolving conflict. 1 2 3 5 6 7 The conflict that occurs helps to defuse the more serious and destructive conflict. 1 2 3 5 6 7 This relationship meets or exceeds our expectations. 1 2 3 5 6 7 A small percentage of individuals involved in this relationship set the relationship policies. 1 2 3 5 6 7 Unless payback is immediate. help is not offered to the other party. 1 2 3 5 6 7 {\lthough information is shared between the parties. it is often unusable because it is in a l 2 3 5 6 7 format that cannot easily be understood by the other party. When we meet to settle a dispute with this party. we often refer to common ODJCCIIVCS in 1 2 3 5 6 7 order to overcome these disputes. The parties refer to written and/or verbal instructions to guide decision making. 1 3 3 5 6 7 The parties help each other in value analySis. cost reductions. problem solvmg. etc. i 2 3 5 6 7 Long-range forecasts are readily shared. 1 2 3 5 6 7 Conflict in this relationship centers on how to achieve goals. not the goals themselves. 1 2 3 3 6 7 The parties assess relationship performance through a formal evaluation program. 1 2 3 5 6 7 The parties often disagree on what is in the best interest of the relationship. 1 2 3 5 6 7 Relationship conflict is handled in a manner that attempts to make both parties better off. 1 2 3 5 6 7 Reflecting on the relationship conflict that has occurred, I would characterize the overall impact of that conflict as: DESTRUCTIVE (damaged relationship) 1 2 3 4 5 (improved relationship) CONSTRUCTIVE 312 Considering the relationship you identified. Strongly Strongly please respond to the Ellowing statements: 0"“3’“ 48"“ This firm does not keep our interests in mind when making decrsmns or taking actions 1 2 3 4 5 6 7 Thisfirmkeepsthepromisesitmakestoourfinn l 2 3 4 5 6 7 We could not achieve our corporate goals without being pan of this relationship. I 2 3 4 5 6 7 My firm intends to maintain this relationship indefinitely. l 2 3 4 5 6 7 We believe the information this party provides us. 1 2 3 4 5 6 7 We find it necessary to be cautious with this firm. 1 2 3 4 5 6 7 If this relationship were terminated. our firm could easily replace the other party. I 2 3 4 5 6 7 This relationship deserves our firm's maximum et’fon to maintain. l 2 3 4 5 6 7 The other party is not always honest with us 1 2 3 4 S 6 7 We are nor very committed to this relationship. 1 2 3 4 5 6 7 Compared to a year ago, this relationship can best be described as: Less Productive l 2 3 4 S 6 7 More Producuve Less Trusting l 2 3 4 5 6 7 More Trusting Lower Quality 1 2 3 4 5 6 7 Higher Quality Less Friendly 1 2 3 4 5 6 7 More Friendly Based on the relationship you identified, please consider the following statements: Strongly Strongly Disagree Agree Conflict has pushed this relationship to higher performance levels. 1 2 3 5 6 7 lnforrnation / data is exchanged infrequently. l 2 3 5 6 7 Superiors must review all important relationship decrsrons before the decrsrons take effect. I 2 3 5 6 7 This relationship has performed poorly. l 2 3 5 6 7 Conflict has made us review past actions and make improvements. l 2 3 5 6 7 Formal. written contracts gurde relationship decision-making. l 2 3 5 6 7 Impending design changes are readily shared. 1 2 3 5 6 7 Conflict has allowed the parties to vent frustration. but hasn't really hurt the relationship. 1 2 3 5 6 7 Much of the conflict centers on disagreements over the basrc goals of the relationship. 1 2 3 5 6 7 The parties share performance reports regularly. I 2 3 5 6 7 When changing conditions have caused friction. the parties have been able to make the l , 3 5 6 7 necessary adjustments in order to restore balance to the relationship. " Based on the objectives we hold for this relationship. we are very satisfied with the l , 3 5 6 7 performance of this relationship. ' The frontline employees involved in managing this relationship are free to make decisions 1 ., 3 5 6 7 involvmg the relationship. ' The parties are able to exchange data or information easin and in usable formats. 1 2 3 5 6 7 313 Considering the relationship you have with the primary contact at this Strongly Strongly firm. please respond to the following statements: 0““3'“ ‘3’“ This indivrdual always treats me fairly. l 2 3 4 5 6 7 I consider this individual to be a friend as well as a business associate. 1 2 3 4 S 6 7 There is a great amount of personal conflict between us. 1 2 3 4 S 6 7 l have great respect for this individual. 1 2 3 4 5 6 7 Based on the relationship you identified, St ngl St “81 . . _ to y ro y please consrder the following statements. Di Agm The parties seek to understand each other‘s positions before attempting to resolve conflict. 1 2 3 4 5 6 7 The parties have different expectations for this relationship. 1 2 3 4 5 6 7 If the past had more open conflict. there would be less frustration with each other now. 1 2 3 4 5 6 7 When making a relationship deciSion. the parties fall back on standard courses of action. 1 2 3 4 5 6 7 Conflict has forced the parties to create innovative solutions to relationship problems. 1 2 3 4 5 6 7 The parties advise each other of potential problems in meetings each other's needs. 1 2 3 4 S 6 7 When we approach conflict resolution. each party pursues its own goals. 1 2 3 4 5 6 7 In response to the relationship conflict. the parties have come to a better understanding of 1 2 3 4 5 6 7 what is. and what isn‘t. acceptable relationship behavior. This relationship has not performed up to our expectations. 1 2 3 4 5 6 7 There can be little action taken until relationship decrsmns have been approved through 1 2 3 4 5 6 7 the proper channels. The relationship makes use of many controls. 1 2 3 4 5 6 7 When dealing with disputes. the parties only look out for their own interests. 1 2 3 4 5 6 7 After conflict has been resolved. the relationship returns to the status quo Without making I 2 3 4 5 6 7 the changes required to avoid the reoccurrence of similar conflict. In general conflict in this relationship occurs mostly over very important issues 1 2 3 4 5 6 7 When faced With a relationship deciSion. the parties rely on behavior dictated by l 2 3 4 5 6 7 formalized roles and responsibilities. Financial planning information is rarely shared in this relationship. 1 2 3 4 5 6 7 When resolving conflict. the parties look for mutually beneficial solutions. 1 2 3 4 5 6 7 This relationship is characterized by strong disagreements about strategic issues. 1 2 3 4 S 6 7 When meetings occur to discuss conflict. the parties often leave feeling even more upset. l 2 3 4 S 6 7 The parties often agree on what constitutes appropriate relationship behavior. 1 2 3 4 5 6 7 314 Please assess the conflict that has characterized this relationship over the last year. Conflict over this Conflict over this In this relationship, this ISSUE issue has occurediluiv No! vets“ has been: Very M" very issue is: Vt” Never Frequently Intense Intense Important Important Minimum order quantities 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Payment terms 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Product quality 1 2 3 4 5 6 7 l 2 3 4 S 6 7 l 2 3 4 5 6 7 Level of sales support 1 2 3 4 5 6 7 1 2 3 4 5 6 7 l 2 3 4 5 6 7 Level of service support 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Advertising support 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Performance objectives 1 2 3 4 5 6 7 1 2 3 4 5 6 7 1 2 3 4 5 6 7 Terms of contract 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 Inventory policies 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Warranty policies 1 2 3 4 5 6 7 l 2 3 4 S 6 7 l 2 3 4 5 6 7 Delivery terms 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Delivery performance 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Communication of problems 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Performance expectations 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Frequency of communication 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Coordination of strategy 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Control of resources 1 2 3 4 5 6 7 l 2 3 4 5 6 7 1 2 3 4 5 6 7 Sharing of market information 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Design issues 1 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Price and/or margins l 2 3 4 5 6 7 l 2 3 4 5 6 7 l 2 3 4 5 6 7 Ease of communication 1 2 3 4 5 6 7 1 2 3 4 S 6 7 l 2 3 4 5 6 7 Considering the relationship conflict, I believe that overall the conflict has: Been very infrequent l 2 3 4 5 6 7 Been very frequent Not been very intense l 2 3 4 5 6 7 Been very intense Caused bitterness l 2 3 4 5 6 7 Reduced bitterness Lead to poor decisions 1 2 3 4 5 6 7 lead to better decisions Lead to wasted resources 1 2 3 4 S 6 7 Lead to better use of resources Pulled parties apart 1 2 3 4 5 6 7 Brought parties closer together Hindered productivity 1 2 3 4 5 6 7 Lead to productivrty improvements PERSONAL INFO: Your position in the firm (Title) Time in that position: (yrs) Time With current firm: (yrs) Time in Industry: (yrs) Time managing this relationship: (yrs) 315 Thank you for your time and c00peration RETURN TO: James A. Eckert Michigan State University Department of Marketing & Supply Chain Management N370 North Business Complex East Lansing, MI 48824-9902 FAX: 517—432-1 112 316 't‘gr ‘-_' ii:..,'.:,|t_-'I it'll“)! "f'-'l3VlflU . - I) “1311ch r ;.:h'_) Ream '. _ -il -‘- ‘-183& ill!" i". 9.0 4 ‘ I ‘ A I . ‘ REFERENCES 317 IUEFTHIEEKIEEI Achrol, Ravi (1991), “Evolution of the Marketing Organization: New Forms for Turbulent Environments,” Journal of Marketing, 55(October), 71-93. Achrol, Ravi, Torger Reve, and Louis W. Stern (1983), “The Environment of Marketing Channel Diads: A Framework for Comparative Analysis,” Journal of Marketing, 47(Fall), 55— 67. Alderson, Wroe (1954), “Factors Governing the Development of Marketing Channels," Marketing Channels for.Manufactured Products, Richard M. Clewet, ed. Homewood, IL: Irwin, 5-22. Anderson, Erin and Barton Weitz (1989), “Determinants of Continuity in Conventional Industrial Channel Dyads,” Marketing Science, 8(Fall), 310-323. Anderson, James C. (1987), “An Approach for Confirmatory Measurement and Structural Equation Modeling of Organizational Properties,” Management Science, 33 (April), 525-541. ————————————— and David W. Gerbing (1982), “Some Methods for Respecifying Measurement Models to Obtain Unidimensional Construct Measurement," Journal of Marketing Research, XIX (November), 453-460. ------------- and David W, Gerbing (1988), “Structural Equation Modeling in Practice: A Review and Recommended Two-Step Approach," Psychological Bulletin, 103 (May), 411- 423. ————————————— and James A. Narus (1984), “A Model of the Distributor’s Perspective of Distributor Manufacturer Working Relationships,” JOurnal of Marketing, 48(Fall), 62— 74. ————————————— and James A. Narus (1990), “A Model Of Distributor Firm and Manufacturing Firm Working Partnerships," Journal of Marketing, 54(Januaryl. 42-58. 318 Armstrong, J. Scott and Terry S. Overton (1977), "Estimating Nonresponse Bias in Mail Surveys," Journal of Marketing Research, 14(August), 396-402. Arndt, Johan (1979), “Towards a Concept of Domesticated Markets,” Journal of Marketing, 43(Fall), 69—75. Assael, H.(l969), “Constructive Role of Interorganizational Conflict,” Administrative Science Quarterly; 14(December), 573—582. Bagozzi, Richard P. and Youjae Yi (1988), “On the Evaluation of Structural Equation Models," JOurnal of the Academy of Marketing Science, 16(Spring). 74-94 Blake, Robert R. and Jane S. Mouton (1964), The Managerial Grid, Houston, TX: Gulf Publishing. Bollen, Kenneth (1989), Structurual Equations with Latent variables, New York, NY: John Wiley & Sons, Inc. Bowersox, Donald J. (1990), “The Strategic Benefits of Logistics Alliances," Harvard Business Review, (July— August), 36-45. ————————————————————————— , Roger Calantone, Steven Clinton, David Closs, M. Bixby Cooper, Cornelia Droge, Stanley Fawcett, Robert Frankel, David Frayer, Edward Morash, Lloyd Rinehart and Judith Schmitz (1995), WOrld Class Logistics: The Challenge of.Managing Continuous Change, Oak Brook, IL: The Council of Logistics Management. Brown, James and Ralph Day (1981), “Measures of Manifest Conflict in Distribution Channels,” Journal of Marketing Research, 18 (August), 263-274. Brown, James R., Robert F. Lusch and Laurie P. Smith (1991), “Conflict and Satisfaction in an Industrial Channel of Distribution," International Journal of Physical Distribution and Logistics Management, 21(6), 15-25 Bucklin, Louis P. (1965), “Postponement, Speculation and the Structure of Distribution Channels,” JOurnal of Marketing Research, 2(February). 26-31 Burnkrant, Robert E. and Thomas J. Page, Jr. (1982), “An Examination of the Convergent, Discriminant, and Predictive Validity of Fishbein’s Behavioral Intention Model,” Journal of Marketing Research, 19(November), 550—561. 319 Cosar, Lewis A. (1956), The Functions of Social Conflict, London: Free Press. ------------------ (1967), Continuities in the Study of Social Conflict, New York: Free Press. Churchill, Gilbert A. (1979), “A Paradigm for Developing Better Measures of Marketing Constructs,” Journal of Marketing Research, 16(February), 64-73. ------------------------- (1991), Marketing Research: .Methodological Foundations, 5th Edition, ChicagozThe Dryden Press. Converse, Paul D. (1949), “New Laws of Retail Gravitation," Journal of Marketing, 14(October) 379—84. Dant, Rajiv P. and Patrick L. Schul (1992), “Conflict Resolution Processes in Contractual Channels of Distribution," Journal of Marketing, 56(January). 38-54. Davis, Murray S. (1971) “That’s Interesting!,” Philosophy of Social Science, I, 309—344. Deutsch, Morton (1973), The Resolution of Conflict, New Haven, CT: Yale University Press. Dillman, Don A. (1978), Mail and Telephone Surveys: The Total Design.Method, New York, NY: John Wiley & Sons, Inc. Dwyer, Robert F. and Sejo Oh (1987), “Output Sector Munificence Effects on the Internal Political Economy of Marketing Channels," JOurnal of Marketing Research, 24(November), 347-58 Dwyer, Robert F., Paul H. Schurr and Sejo Oh (1987), “Developing Buyer-Seller Relationships,” Journal of Marketing, 51(Apri1), 11-27. ————————— and Orville Walker Jr.(1981), “Bargaining in an Asymmetrical Power Structure," Journal of Marketing, 45 (Winter), 104-115. --------- and Mary Ann Welsh (1985), “Environmental Relationships of the the Internal Political Economy of Marketing Channels," Journal of Marketing Research, 22(November), 397-414. xm El-Ansary, Adel I and Louis Stern (1972), “Power Measurement in the Distribution Channel," Journal of Marketing Research, 9(February) 47-52. Etgar, Michael (1979), “Sources and Types of Intrachannel Conflict,” Journal of Retailing, 55 (Spring), 61—78. Fornell, Claes and David F. Larcker (1981), “Evaluating Structural Equation Models with Unobservable Variables and Measurement Error," Journal of Marketing Research, 18(February), 39—50. Frazier, Gary L. and Kersi D. Antia (1995), Exchange Relationships and Interfirm Power in Channels of Distribution," JOurnal of the Academy of Marketing Science, 23(4), 321-26. Ganesan, Shankar (1994), “Determinants of Long-Term Orientation in Buyer-Seller Relationships,” Journal of Marketing, 58(Apri1), 1-19. Gaski, John (1984), “The Theory of Power and Conflict in Channels of Distribution," Journal of Marketing, 48 (Summer), 9-28. Gundlach, Gregory T., Ravi S. Achrol and John T. Mentzer (1995), “The Structure of Commitment in Exchange," JOurnal of Marketing, 59(January) 78-92. Hage, J. and M. Aiken (1967), “Relationship of Centralization to Other Structural Properties,” Administrative Science Quarterly, 12 (June), 72—92. Hayduk, Leslie A., (1987), Structural Equation Mbdeling with LISREL, The John Hopkins University Press. Heide, Jan B. (1994), “Interorganizational Governance in Marketing Channels,” Journal of Marketing , 58(January), 71- 8S. ---------------- and George John (1992), “Do Norms Matter in Marketing Relationships?," Journal of Marketing , 56(April), 32-44. ------------------------------------- (1990), “Alliances in Industrial Purchasing; The Determinants of Joint Action in Buyer-Supplier Relationships," Journal of Marketing Research, 27(February), 24-36. 321 ------------------------------------- (1992), “Do Norms Matter in Marketing Relationships?” Journal of Marketing, 56(April), 32-44. Hoyle, Rick H. (1995), Structural Equation Mbdeling: Concepts, Issues, and Applications, Thousand Oaks, CA: Sage Publications. Hunt, Shelby D. (1991), MOdern Marketing Theory: Critical Issues in the Philosophy of Marketing Science, Cincinnati, OH: South-Western Publishing Co.. ------------------- and John Nevin (1974), “Power in a Channel of Distribution: Sources and Consequences,” Journal of Marketing Research, 11 (May). 186-193. Jackson, Barbara Bund (1985), “Build Customer Relationships that Last,” Harvard Business Review, (November-December), 120-128. John, George (1984), “An Empirical Investigation of Some Antecedants of Opportunism in a Marketing Channel,” Journal of Marketing Research, 21(August), 278-89. Johnston, Russell and Paul R. Lawrence (1988), “Beyound Vertical Integration - The Rise of the Value-Adding Partnership, Harvard Business Review, (July-August), 94-101. Kerlinger, Fred N. (1986), Foundations of Behavioral Research, 3rd Edition, Fort Worth: Harcourt Brace Jovanovich College Publishers. Kuhn, Thomas S. (1970), The Structure of Scientific Revolutions, Chicago: University of Chicago Press. Leslie, L.L. (1972), "Are High Response Rates Essential to Valid Surveys?" Social Science Research, 1, 323-334. Lusch, Robert F. (1976a), “Sources of Power: Their Impact on Intrachannel Conflict,” JOurnal of Marketing Research, 13 (November), 382-390. ------------------- (1976b), “Channel Conflict: Its Impact on Retailer Operating Performance,” JOurnal of Retailing, 52(Summer), 3-12, 89-90. ——————————————————— and James R. Brown (1982), “A Modified Model of Power in the Marketing Channel," Journal of Marketing Research, 19(August), 312-323. 322 MacNeil, Ian R.(1978), “Contracts: Adjustment of Long-Term Economic Relations Under Classical, Neo-Classical, and Relational Contract Law,'.NOrthwestern University Law Review, 72, 854-905. --------- (1980), The New Social Contract, New Haven, CT: Yale University Press. ---------- (1983), “Values in Contract,” NOrthwestern University Law Review, 78 (April), 340-410. Mallen, Bruce (1963), “A Theory of Retailer-Supplier Conflict, Control, and Cooperation,” JOurnal of Retailing, 39(Summer), 24-32, 51. ------------------ (1967), The Marketing Channel: A Conceptual Viewpoint, New York: John Wiley & Sons, Inc. ------------------ (1973), “Functional Spin-Off: A Key to Anticipating Change in Distribution Structure,” Journal of Marketing, 37 (July). 18-25. March, James G. and Herbert A. Simon (1958), Organizations, New York, NY: Wiley. Morgan, Robert M. and Shelby D. Hunt (1994), “The Commitment-Trust Theory of Relationship Marketing,” Journal of Marketing, 58(July). 20-38. Narus, James A. and James C. Anderson (1986), “Turn you industrial distributors into partners," Harvard Business Review, 64(March-April), 66-71. Nevin, John R. (1995), “Relationship Marketing and Distribution Channels: Exploring Fundamental Issues,” Journal of the Academy of Marketing Science, 23(4), 327-334. Noordewier, Thomas G., George John and John R. Nevin (1990), “Performance Outcomes of Purchasing Arrangements in Industrial Buyer-Vendor Relationships,” Journal of Marketing, 54(October), 80-93. Nunnally, Jum C. (1978), Psychometric Theory,i?m‘ed. New York: McGraw-Hill Book Company. Palay, Thomas (1984), “Comparative Institutional Economics: The Governance of Rail Freight Contracting," Journal of Legal Studies, 13 (June), 265-288. 323 Peter, Paul J. (1981), “Construct Validity: A Review of Basic Issues and Marketing Practices," Journal of Marketing Research, 18(May), 133-145. Pondy, Louis R. (1967), “Organizational Conflict: Concepts and Models,” Administrative Science Quarterly, 12(September), 296-320. Pruit , Dean G. (1981), Negotiation Behavior, New York: Academic Press, Inc.. Rahim, Afzalur M. (1983), “Measure of Styles of Handling Interpersonal Conflict,” Academy of Management Journal, Reve, Torger and Louis W. Stern (1986), “The Relationship Between Interorganizational Form, Transaction Climate, and Economic Performance in Vertical Interfirm Dyads,” in Marketing Channels: Relationships and Performance, Luca Pellegrini and Srinivas K. Reddy, eds. Lexington, MA: Lexington Books, 75-102. Robicheaux, Robert A. and James E. Coleman (1994), “The Structure of Marketing Channel Relationships," JOurnal of the Academy of Marketing Science, 22(1), 38-51. Rosenberg, Larry J. and Louis Stern (1970), “Toward the Analysis of Conflict in Distribution Channels: A Descriptive Model,” JOurnal of Marketing Research, 34 (October), 40—46. ------------------------------------------ (1971), “Conflict Measurement in the Distribution Channel,” Journal of Marketing Research, 8(November),437—442. Rosenbloom, Bert (1973), “Conflict and Channel Efficiency: Some Conceptual Models for the Decision-Maker,” J0urnal of Marketing, 37 (July), 26—30. Ryan, Franklin W. (1935), “Functional Concepts in Market Distribution," Harvard Business Review, 13(January). 205- 224. Schul, Patrick L. and Emin Babakus (1988), “An Examination of the Interfirm Power-Conflict Relationship: The Intervening Role of The Channel Decision Structure," Journal of Retailing, 64(4), 381-404. Shaw, Arch (1912), “Some Problems in Market Distribution,” Quarterly JOurnal of Economics, 26(August) 706-765. 324 Sheth, Jagdish N. (1973), “A Model of Industrial Buyer Behavior," JOurnal of Marketing, 37(October), 50-56. Simmel, Georg (1955), Conflict, trans. Kurt H. Wolff, Glencoe, ILzThe Free Press. Stern, Louis W. (1969), Distribution Channels: Behavioral Dimensions, Boston, MA: Houghton Mifflin Co. ------------------ and Jay W. Brown (1969), “Distribution Channels: A Social Systems Approach,” in Distribution Channels: Behavioral Dimensions, ed. Louis W. Stern, BostonzHoughton Mifflin, 1-19. ------------------ and Adel I. El-Ansary (1977), Marketing Channels, Englewood Cliffs, NJ: Prentice-Hall, Inc. ------------------ and R. Gorman (1969), “Conflict in Distribution Channels: An Exploration," in Distribution Channels: Behavioral Dimensions, L.W. Stern, ed. Boston: Houghton-Mifflin, 156-175. ------------------ and Torger Reve (1980), “Distribution Channels as Political Economies: A Framework for Comparative Analysis," Journal of Marketing, 44(Summer), 52-64. Thomas, Kenneth (1976), “Conflict and Conflict Management,” in Handbook of Industrial and Organizational Psychology, Marvin D. Dunnette, ed. Chicago: Rand McNally, 889-935. Turner, Jonathan H. (1986), The Structure of Sociological Theory, 4a‘Edition, ChicagozThe Dorsey Press. Webster, Frederick E., Jr. (1992), “The Changing Role of Marketing in the Corporation,” JOurnal of Marketing, 56(October), 1-17. Weitz, Barton A. and Sandy D. Jap (1995), “Relationship Marketing and Distribution Channels,” Journal of the Academy of Markeing Science, 23(4), 305-320. Weld, L. D. H. (1916), The Marketing of Farm Products, New Yorszhe Macmillan Company. __________________ (1917), “Marketing Functions and Mercantile Organizations,” American Economic Review, 7(June), 306-318. 325 Wilkins, Alan L. and William G. Ouchi (1983), “Efficient Cultures: Exploring the Relationship Between Culture and Organizational Performance,” Administrative Science Quarterly, (28): 468-481. Williamson, Oliver E. (1975), Markets and Hierarchies, New York, NY: The Free Press. --------------- (1979), “Transaction Cost Economics: the Governance of Contractual Relations,” Journal of Law and Economics (22): 233-261. 326