Leadership Determinants of Trust Sim Sitkin Fuqua School of Business Duke University In collaboration with Allan Lind, Morela Hernandez, and Chris Long 1 What Influences Trust? Various influences have been identified Competence, benevolence, reliability, honesty, etc (e.g., Dirks & Ferrin, Avoilio and colleagues) Influences are generally consistent with the kinds of things leaders “should” do But the links between specific leadership behaviors and trust has remained largely unexamined Often asserted, rarely tested Conceptualization is often vague and atheoretical, or very broad Goal is to clarify the potentially important insights for both literatures Present newly developed leadership approach Make the leadership-to-trust links explicit and specific Present preliminary results of an early test Discuss some implications 2 The “Leadership Challenge” There is a substantial body of work on leadership, but there is not really a clear picture of what leadership is and how one can teach people to be better leaders. Some scholars (e.g., Meindl, Ehrlich, & Dukerich) have even contended that leadership is not an important factor in organizational performance, that it is simply a “romanticized explanation” of more complex management and environmental factors. 3 How to Address the Challenge Consider whether there is indeed something in the concept of leadership to theorize about, study, and teach Organize our understanding of leadership and related phenomena so that we can systematically examine how leadership evolves as organizations change 4 Contemporary research and theory on leadership Huge quantity and variety Variable in focus and quality Empirically rigorous, but more usually managerial than leadership-focused Atheoretical and/or vague Practitioner-oriented, but not very systematic or testable Narrow, focusing on only one or two aspects of leadership Theoretical, but are not very generative for rigorous scholarship or practical enough for managerial application Some examples Transformational leadership, Charismatic leadership, Symbolic leadership, Attribution theories of leadership, Relational leadership 5 A preview of approach Working with my colleagues Allan Lind and Chris Long (and additional colleagues, and including recent PhD students Jim Emery, Morela Hernandez, and Drew Carton), we have worked to organize and integrate this topic with an eye to empirical testing and implementation. We have found ourselves incorporating & extending many elements used by other theories, but we have been most influenced by relational views of leadership and trust. Thus, today’s focus . . . Before I begin presenting our theory and early results on the impact of leadership on trust, let me define it and give you a brief picture of our leadership framework. 6 Definition of Leadership A leader is: A person who influences others A person who exhibits specific leadership behaviors A person who accepts a leadership role and identity Leadership is: A set of behaviors and their effects A social role A perspective or identity Leadership is not: Formal authority or position Only positive (effective leaders can pursue evil) A set of traits that cannot be developed or modified or learned (“you can’t teach height” but you can teach leadership) 7 Leadership versus Management Not about leaders vs. managers, not different people Most individual roles involve elements of both leadership and management But both good leadership and good management are necessary for optimal organization performance Leadership is not just about top organizational heroes Includes leading up, down and laterally Applies to a variety of life roles – leading peers, family members, community, leading oneself 8 How is the Approach Distinctive? Focus on Behavior: Leadership is what you do, not just who you are. Change what you do and you can change your leadership style. Thus, the approach is testable and actionable. Focus on Effects: Each dimension is keyed to theorized effects of leadership behaviors. Focus on Breadth and Integration: Most leadership approaches focus on just a few aspects of leadership – ours tries to integrate the range of leadership dimensions – and effects. 9 Leadership domains ETHICAL INSPIRATIONAL PERSONAL SUPPORTIVE RELATIONAL CONTEXTUAL 10 Core focus of the domains Accepting responsibility Raising optimism & enthusiasm Preparing and projecting who you are Providing resources, feedback, and protection Showing concern and understanding Clarifying who we are and how we work together 11 Consequences of effective leadership STEWARDSHIP HIGH ASPIRATION CREDIBILITY INITIATIVE TRUST COMMUNITY 12 Leadership domains and effects STEWARDSHIP ETHICAL HIGH ASPIRATION INITIATIVE INSPIRATIONAL PERSONAL CREDIBILITY SUPPORTIVE RELATIONAL TRUST CONTEXTUAL COMMUNITY 13 Our Focus Today TRUST 14 Definition of Trust Trust is a psychological state comprising the intention to accept vulnerability based on positive expectations of the intentions or behavior of another (Rousseau et al., 1998) 15 Foundation of the Model Personal: Leadership emerges from the projection of your personal values, concerns, passions and world view. Relational: Leadership is rooted in projecting concern for and understanding of others in interpersonal and inter-organizational relationships. Contextual: Leadership simplifies and focuses by clarifying contexts. 16 Personal Leadership Demonstrate that you have the insight and knowledge to lead to success. Be real; let your values and personality show in your actions. Make your dedication to the team-and your courage in pursuing its goals--evident to all. 17 Effective personal leadership yields CREDIBILITY 18 Relational Leadership Attend to your leader-follower relationship with each person you seek to lead: Show concern, understanding, and respect for others. Be seen as fair. 19 Effective relational leadership yields TRUST 20 Contextual Leadership Create a sense of identity (pride and belonging). Focus and simplify to build a sense of coherence. Enhance clarity of roles and functions. 21 Effective contextual leadership yields a sense of COMMUNITY 22 But our focus today is on leadership determinants of trust TRUST What is the prediction from the literature about what influences trust? 23 So how might these three dimensions relate to trust? 24 Arguments Extrapolated from Trust Literature Personal Leadership Affects Trust Competence, shared values, personal interests are predictive of trust (Sitkin & Roth; Mayer et al) Attribution of leadership – passion, insight (Calder) Relational Leadership Affects Trust Leadership as forms of relationship (Weber; Lewin; Kouzes & Posner) Benevolence, caring, respect, fairness & understanding of the other (Bies; Lind & Tyler; Sitkin & Roth) Contextual Leadership Affects Trust Increased contextual control undermines trust in other party (Shapiro, Zucker) Symbolic leadership and the importance of congruent symbols (Pfeffer; Sitkin & Stickel) Without formal protection, risk of opportunistic exploitation is too high (Sitkin; Lewicki & Bunker; Bijlmsma-Frankema & Costa; Long & Sitkin) 25 Implicit Theoretical Model Personal Relational Contextual Trust 26 ETHICAL INSPIRATIONAL PERSONAL SUPPORTIVE RELATIONAL CONTEXTUAL 27 Key theoretical assumption: interdependence of leadership domains Third-Order ETHICAL Second-Order INSPIRATIONAL SUPPORTIVE Foundation PERSONAL More personal RELATIONAL CONTEXTUAL More structural 28 Leadership Framework Extends Argument to Make It More Directly Testable Dimensions of Leadership are Mutually Facilitative Personal & Contextual Leadership Affects Trust, but only through their link to Relational Leadership Direct Links of Personal and Relational to Trust implied in the literature may be spurious, as indirect routes of influence never tested 29 Theoretical Model Personal Relational Contextual Trust 30 Hypotheses Higher levels of Relational Leadership are associated with greater Trust in the leader (H1). Personal (H2A) and Contextual (H2B) Leadership will be significantly associated with Trust. Personal (H3A) and Contextual (H3B) Leadership will be significantly associated with Relational Leadership. The effects of Personal (H4A) and Contextual (H4B) Leadership on Trust will be mediated by Relational Leadership. 31 Sample and Procedure Participants completed the 360-degree survey online as part of executive leadership courses (n=129) Weekend EMBA (n=55) Cross Continent EMBA (n=52) Open Enrollment (n=22) Ratings supplied by supervisors, peers and direct reports (n= 700+) Permission to use data for research requested, nearly all consented Survey distributed & completed online (approx. 20 minutes to complete survey) Participants received feedback as part of a course Raters assured anonymity; ratings aggregated to preserve confidentiality of individual raters 32 Web-based Leadership Instrument 33 MDLI Feedback 34 MDLI results (graphics) 35 6 domains, multiple rater groups 36 MDLI results (Personal Leadership table) 37 Some Preliminary Findings Today’s analysis based on data collected in July, August, September of 2003. Able to clean data and create appropriate scales Analyzed using structural equation models (AMOS) Still consider findings to be tentative - newly redesigned variables & results in but not yet analyzed. But results seem quite robust, so optimistic 38 Scale Reliabilities “The locus of leadership . . . involves behaviors . . . produced by leaders as these elements are interpreted by followers.” (Lord & Maher, 1993; p.11) IV Scales created from “follower” perceptions of leader behaviors DV Scales created from “follower” ratings of leader effects across three rater groups Intraclass Correlation Coefficient (ICC) ICC useful for determining the extent to which variance of individual responses are attributed to group membership ICC assesses reliable differentiation between groups (Bliese et al., 2002; Castro, 2002) ICC especially useful if between group variability is potentially theoretically important and requires detailed examination Supervisors + Peers + Direct Reports: Peers + Direct Reports: ICC = .862 ICC = .723 39 Independent Variables Personal Leadership Behavior (α = .80) Authenticity Vision - Creativity - Passion & Courage Expertise Relational Leadership Behavior (α = .81) Concern Respect Reliability Contextual Leadership Behavior (α = .77) Coherence Coordination Identity 40 Personal Leadership Behavior (α = .80) Authenticity Lets you know what he/she is really like. Lives his/her values. They are who they appear to be. Vision Provides a clear vision for the organization or unit. Formulates clear goals. Articulates where the organization/unit is going. Expertise Really understands our work. Is smart about what we do. Has deep expertise. Creativity Finds innovative solutions to business problems. Is open to exploring new ideas. Thinks outside the box. Passion and Courage Is passionate about the work we do. Displays courage in the face of uncertainty. Is not afraid to show his/her feelings. Is not afraid of being wrong. Is committed to doing what he/she thinks is right. 41 Relational Leadership Behavior (α = .81) Concern Displays concern for me. Is sensitive to my needs. Cares about my priorities and interests. Is interested in understanding me. Shows compassion. Respect Shows respect for people regardless of their level in the organization. Makes an effort to seek out others' opinions on important issues. Takes the time to explain decisions. Is a good listener. 42 Contextual Leadership Behavior (α = .77) Coherence Makes sure his/her employees understand business issues. Promotes a shared understanding about complex issues. Cuts through complex or ambiguous problems to make them easier to understand. Explains why things are being done a particular way. Coordination Helps coordinate actions of unit or organization. Resolves conflicts constructively. Creates processes that facilitate the work. Ensures that we take the needs of others into account as we do our work. Makes clear how responsibilities are being divided. 43 Dependent Variables: Leadership Effects Personal Leadership: LOYALTY (α = .86 ) I feel loyal to ____. ____ can depend on me. I would go out of my way to help ____ if he/she asked me to. Relational Leadership: TRUST (α = .80 ) I trust ____ to be fair. ____ deals fairly with me. ____ is unbiased in his/her decisions. Contextual Leadership: COMMUNITY (α = .87 ) We are like family. People here are concerned with the success of the whole organization. I feel like I’m really part of the team around here. In this organization, we know we can depend on each other. Being a good organizational citizen is part of our organization culture. 44 Hypotheses Higher levels of Relational Leadership are associated with greater Trust in the leader (H1). Personal (H2A) and Contextual (H2B) Leadership will be significantly associated with Trust. Personal (H3A) and Contextual (H3B) Leadership will be significantly associated with Relational leadership. The effects of Personal (H4A) and Contextual (H4B) Leadership on Trust will be mediated by Relational leadership. 45 Theoretical Model Personal Relational Contextual Trust 46 Statistical Methods Structural Equation Model Analyses (AMOS 5.0 in conjunction with SPSS 11.5) 4 models are presented Standardized Regression Weights are shown Details (error terms etc) are not shown on models to simplify display for readability Model Fit Indices TLI and CFI were considered This is still very much a work in progress 47 Model e1 e2 Vision e3 Creativity Expertise e10 e4 Passion e5 e6 Authenticity e7 Respect Concern e11 e9 Coordination Coherence e12 Relational Ldrshp Personal Ldrshp e8 Contextual Ldrshp e13 TRUST 48 Direct Effects Test of Trust Literature Personal Ldrshp Relational Ldrshp .45*** .34** Contextual Ldrshp .26* TRUST Model Fit: TLI = .54; CFI = .66 49 Direct Effects Test of Trust Literature Personal Ldrshp Relational Ldrshp .45*** Contextual Ldrshp .26* .34** TRUST Model Fit: TLI = .54; CFI = .66 •Significance consistent with trust literature (H1, H2A & H2B) •Fit of model based on literature is quite low. 50 Test of Hypothesized Mediated Model r = .87 Personal Ldrshp .35** ns Relational Ldrshp .78* .62*** Contextual Ldrshp ns TRUST Model Fit: TLI = .91; CFI = .94 51 Test of Hypothesized Mediated Model r = .87 Personal Ldrshp .35** ns Model Fit: TLI = .91; CFI = .94 Relational Ldrshp .78* .62*** Contextual Ldrshp ns TRUST •Personal & Contextual Leadership significantly associated with Relational Leadership (H3A & H3B) •Direct effects predicted by trust literature mediated, as predicted (H4A & H4B) •Fit of model substantially improved & quite high. 52 Results of Hypothesis Tests Higher levels of Relational Leadership are associated with greater Trust in the leader (H1). SUPPORTED Personal (H2A) and Contextual (H2B) Leadership will be significantly associated with Trust. SUPPORTED WHEN EXAMINED IN ISOLATION; REJECTED WHEN MODELLED MORE COMPLETELY Personal (H3A) and Contextual (H3B) Leadership will be significantly associated with Relational Leadership. SUPPORTED The effects of Personal (H4A) and Contextual (H4B) Leadership on Trust will be mediated by Relational leadership. H4A. SUPPORTED H4B. SUPPORTED 53 Implications of Hypothesis Tests H1 and H2: All three foundational dimensions appear to be determinants of trust if examined independently. Personal and Contextual Leadership were found to be significantly associated with Relational Leadership (H3A and H3B) Relational Leadership is a key determinant of Trust. As a direct determinant of Trust (H1) Personal and Contextual Leadership effects on Trust are mediated (H4A and H4B) by Relational Leadership. 54 Including Self Ratings “The locus of leadership is not solely in a leader or solely in follower. Instead, it involves behaviors . . . produced by leaders as these elements are interpreted by followers.” (Lord & Maher, 1993; p.11) Although our theoretical focus was based on “follower” reactions to leadership behavior (focusing on effects, not just leader attributes), we wanted to examine whether self ratings of leaders matched the ratings of “followers” and whether leader self perceptions predicted “follower” trust in the leader. 55 Examining Trust Predictions with Self Ratings Personal Ldrshp (Self) ns Relational Ldrshp (Self) ns Contextual Ldrshp (Self) .12** Model Fit: TLI = .30; CFI = .58 TRUST 56 Examining Trust Predictions with Self Ratings Personal Ldrshp (Self) ns Relational Ldrshp (Self) ns Contextual Ldrshp (Self) .12** •Self ratings of leadership behavior are distinct from others’ rating, are less predictive of trust, & have terrible model fit. Model Fit: TLI = .30; CFI = .58 TRUST •Could they have effects through others’ perceptions of 57 leadership? Do Self Ratings Have an Indirect Effect on Trust? r = .88 r = .78 Personal Ldrshp (Self) .21** r = .69 Relational Ldrshp (Self) ns Contextual Ldrshp (Self) ns ns .18* ns Personal Ldrshp (Others) Relational Ldrshp (Others) .54*** ns Model Fit: TLI = .75 ; CFI = .81 Contextual Ldrshp (Others) .67*** .66** TRUST ns 58 Do Self Ratings Have an Indirect Effect on Trust? Relational Ldrshp (Self) Personal Ldrshp (Self) Contextual Ldrshp (Self) .21** .18* Personal Ldrshp (Others) Relational Ldrshp (Others) .54*** Contextual Ldrshp (Others) .67*** .66** •Results are consistent with an indirect effect, but . . . TRUST Model Fit: TLI = .75 ; CFI = .81 Why the striking relational leadership gap? 59 Self and Other Perceptions of Relational Leadership Behavior Familiar social psychology finding of a gap between self perception of how fair, concerned, understanding, etc we are compared with how others see us (e.g., Messick et al, “Why we are fairer than others” JESP, 1985 ) Why observe in this situation? Possible that relational leadership behaviors are harder to discern Possible that we are not as relationally positive as we think Possible that we are but are lousy at projecting our concern or our efforts But could self perceived relational leadership behavior still have an indirect effect on trust via other self perceptions? 60 Full Model e1 e2 Visio n e3 Creativity e4 Expertise e5 Passion e6 Authenticit y e7 Respect Concern e8 e9 Coordinatio n Coherence e10 Personal Ldrshp (self) e11 e12 Visio n e13 Creativity Expertise e20 Relational Ldrshp (self) e14 Passion e15 e16 Authenticit y Contextual Ldrshp (self) e17 Respect e18 Concern e21 e19 Coordinatio n Coherence e22 Personal Ldrshp Relational Ldrshp Contextual Ldrshp (other) (other) (other) e23 TRUST 61 .94*** Personal Ldrshp (Self) .22** Relational Ldrshp (Self) ns .85*** ns Contextual Ldrshp (Self) ns .18* ns Personal Ldrshp (Others) .56*** ns Relational Ldrshp (Others) .65** Contextual Ldrshp (Others) .69*** ns TRUST Model Fit: TLI = .73; CFI = .80 62 Personal Ldrshp (Self) .94*** Relational Ldrshp (Self) .85*** .22** Personal Ldrshp (Others) Contextual Ldrshp (Self) .18* .56*** Relational Ldrshp (Others) Contextual Ldrshp (Others) .69*** .65** TRUST Model Fit: TLI = .73; CFI = .80 63 Results of Hypothesis Tests Higher levels of Relational Leadership are associated with greater Trust in the leader (H1). SUPPORTED Personal (H2A) and Contextual (H2B) Leadership will be significantly associated with Trust. SUPPORTED WHEN EXAMINED IN ISOLATION; REJECTED WHEN MODELLED MORE COMPLETELY Personal (H3A) and Contextual (H3B) Leadership will be significantly associated with Relational Leadership. SUPPORTED The effects of Personal (H4A) and Contextual (H4B) Leadership on Trust will be mediated by Relational leadership. H4A. SUPPORTED H4B. SUPPORTED 64 Implications of Exploration of Self versus Others’ Perception of Leadership Leader self perceptions do not directly affect trust – but seem to have an indirect effect (sometimes very indirect) via others’ perceptions of leader behaviors Followers’ perceptions of personal and contextual leadership behaviors appear to affect trust through followers’ perceptions of relational leadership behavior Perceived relational leadership is the key to influences on trust in leaders The relationship between self perceptions of leadership behavior and follower perceptions is not uniform – gap seems to be in the relational leadership dimension. 65 Builds on existing theory Consistent with trust, justice and leadership literatures Provides more systematic theoretical framework for linking leadership to trust through specific influencing actions Competence and other personal attributes Relational features of fair treatment, consideration, respect, benevolence Structural features can enable but only through their relational effects 66 Additional Implications/Future Directions Leadership behavior can be systematically broken down and its effects on trust tested. Leadership behavior does appear to be a significant influence on trust. Examine what does not affect trust, as well as what does Does contextual leadership behavior really not affect trust if relational features are controlled? Test more complex, embedded models of how trust arises and is influenced by leaders Specific behaviors, not just broader dimensions Different organizational and cultural conditions Different leader attributes Test other effects in the leadership model 67 Conclusion and Next Steps Book under development Have tightened measures & tests, coupling with experimental studies Will examine rest of model Articles on specific tests and applications AMR piece in development on theory Under development on domain effects, crisis, organizational founding, and co-leadership, etc Case studies and instructional materials Practitioner pieces & cases Improve measurement instruments Longitudinal field & lab data collection on perceived behavior – plus “hard” performance and behavior measures 68 Duke Leadership Research Program: Additional Studies In Progress Sim Sitkin, Allan Lind, and Colleagues Fuqua School of Business Duke University Winter, 2006 69 Leadership Behaviors as Determinants of Specific and Distinct Follower Responses Sim Sitkin & Allan Lind 70 Leadership domains and effects STEWARDSHIP ETHICAL HIGH ASPIRATION INITIATIVE INSPIRATIONAL PERSONAL CREDIBILITY SUPPORTIVE RELATIONAL TRUST CONTEXTUAL COMMUNITY 71 Leadership Determinants of Raised Aspirations Sim Sitkin, Jim Emery, Drew Carton, & Allan Lind 72 Initial Model for Aspirations PERSONAL Inspirational Aspirations Elements of Inspirational Leadership High standards, enthusiasm, confidence Initial Results Supported, but fit could be improved Path coefficient = .781, p<.001* RMSEA .107, CLI = .83 * 263 observations of 61 leaders, SEM results adjusted for non-independence of observations using Cluster option in MPlus Version 3.13 73 Additional Work on Aspirations INSPIRATIONAL Aspirations PERSONAL RELATIONAL Initial Results Individually, all 3 dimensions have positive and significant path coefficients predicting Aspirations, but Only Inspirational remains positive and significant when all three dimensions are included in the SEM model 74 Gender and Leadership: The Effect of Mental Models at Different Hierarchical Levels Ashleigh Rosette, Leigh Tost, Morela Hernandez, & Sim Sitkin 75 Gender Differences in Leadership Research Goals: Examine the ways that women top leaders may express gender biases toward their same-sex colleagues; Understand that potential rivalries may exist among elite women; Advance this area of inquiry beyond the search for general trends in gender-based biases that persist across actors Hypotheses: At the lower and middle levels of organizational hierarchy, women demonstrate an ingroup bias by favoring their women peers in their evaluations. However, at the top levels of organizational hierarchy, women demonstrate a bias against their female peers because the tokenism situations that are present in the highest levels of most organizations lead women to focus on perceived threat when considering peer women senior executives. 76 Study 1: Sample and Procedure Participants completed the 360-degree survey online as part of 2005-2006 executive leadership courses (N=61) 40 men; 21 women 36 were identified as top leaders and 24 were identified as middle managers Ratings supplied by 227 work peers (156 men; 71 women) Consent obtained from nearly all students Survey distributed & completed online; approx. 20 min. Participants received feedback as part of a course Raters assured anonymity; ratings aggregated to preserve confidentiality of individual raters The study consisted of a 2 (leader gender: male, female) x 2 (organizational level: top leader, middle manager) x 2 (rater gender: male, female) between-subjects factorial design 77 Study 1: Variables IV’s Organizational level: The raters’ position or rank within the company was coded into a dichotomous variable: Top managers and Middle managers DV’s Leadership Effectiveness: Relationship-oriented and task-oriented behaviors, relational and personal leadership, respectively. Relational leadership behaviors ( = .86). Concern, Respect, and Fairness Personal leadership behaviors ( = .85). Vision, Competence, and Creativity. 78 Personal Leadership 4.5 4 Top leader Middle manager 3.5 3 Figure 1. Mean ratings of leader effectiveness (personal leadership and relational leadership) by organizational level, leader gender, and rater gender. Male rater- Male raterFemale Male leader leader Female rater-Male leader Female raterFemale leader Relational Leadership 4.5 4 Top leader Middle manager 3.5 3 Male rater- Male raterFemale Male leader leader Female rater-Male leader Female raterFemale leader 79 Gender Differences in Leadership Results, confirmed predictions: Compared with women middle managers, women top leaders evaluated their work peers more negatively. This difference in evaluations was not observed between men top leaders and men middle managers. Negative evaluations only occurred when competition amongst the women top leaders was perceived to be high (as in Study 1) or when it was explicitly manipulated as high (as in Study 2). Implication: Managers and executives should be aware of how tokenism may negatively influence women leaders and how it may cause their perspectives to differ substantially from that of their male colleagues. 80