The effects of formal human leadership and computer-generated decision aids on problem solving via computer : a controlled experiment

Twenty-four groups of five professionals and managers within a variety of organizations were given the task of using a computer conference to reach agreement on the best solution to a ranking problem. The independent variable is the structure of the conferencing capability used. Two alternative means of structuring the conferences were employed, in a two-by-two factorial design. Groups with "Human Leadership" elected one of their members to lead the group in its decision making discussion. Groups with "Computer Feedback" were given periodic tables which displayed the current "group decision" in terms of the mean rankings of items, and the degree of consensus about each of these items. Dependent variables include: .Quality of decision .Degree of consensus .Amount of discussion and reranking activity .Equality of participation .Subjective satisfaction Covariates include initial (pre-discussion) quality of decision, typing speed, knowledgability of the leader, age, and sex. For this experiment, with small groups, human leadership was more effective than computer feedback for improving consensus and quality of decision. TABLE OF CONTENTS CHAPTER ONE AN OVERVIEW OF THE EXPERIMENT INTRODUCTION 1 Computer-Mediated Communication: Generalizations and Variations 3 Background: The Prior Experiment 5 EXPERIMENTAL DESIGN 8 The Independent Variables: Structuring the Group Process 9 Dependent and Process Variables 15 Hypotheses 17 Covariates 18 SUBJECTS AND PROCEDURE 19 Description of the Analysis of Variance Designs 22 SUMMARY 23 CHAPTER TWO QUALITY OF DECISION MEASURES OF QUALITY OF DECISION 24 The Decision Data 24 The Percentage Improvement Measure 25 "Collective Intelligence" 26 DIFFERENCES IN QUALITY OF THE GROUP DECISION 28 The Selection and Performance of Leaders 32 Influence of the "Best" Member 33 The Effect of Group Composition 34 "COLLECTIVE INTELLIGENCE," BY CONDITION 36 SUMMARY 39 CHAPTER THREE ABILITY TO REACH CONSENSUS REACHING A GROUP DECISION 41 RESULTS FOR FINAL INDIVIDUAL RANKINGS 46 FACTORS RELATED TO THE ABILITY TO REACH CONSENSUS 48 SUMMARY ••••50 CHAPTER FOUR VARIATIONS IN GROUP PROCESS RELATED TO STRUCTURE AND SUBJECT CHARACTERISTICS DOMINANCE 52 THE EFFECTS OF HUMAN LEADERSHIP AND FEEDBACK ON AMOUNT OF DISCUSSION AND RANKING 54 THE EFFECTS OF SUBJECT CHARACTERISTICS ON PERFORMANCE 57 PROCESS VS. OUTCOME 60 SUMMARY 61 CHAPTER FIVE SUBJECTIVE SATISFACTION EFFECTS OF EXPERIMENTAL CONDITION ON SUBJECTIVE SATISFACTION 64 VARIATIONS BY SUBJECT CHARACTERISTICS 67 The Effect of Age 67 Sex and Subjective Satisfaction 69 Typing and Subjective Satisfaction 70 Effect of Previous Computer Terminal Experience 70 GROUP DIFFERENCES 71 SUMMARY 72 CHAPTER SIX METHODOLOGY: THE AUTOMATED EXPERIMENT COMPUTER AND HUMAN ROLES IN CONDUCTING THE EXPERIMENT 74 Taking the Experiment Into the Field 78 Training and Monitoring Aids 79 Problems: Automated Errors 81 Automated Analysis 82 A SIMILAR AUTOMATED EXPERIMENT 83 CONCLUSION 86 CHAPTER SEVEN SUMMARY AND CONCLUSIONS SUMMARY OF FINDINGS 87 Decision Quality and Degree of Consensus 87 Group Process 89 Subjective Satisfaction 90 Variations Associated with Subject Characteristics 91 The Pervasive Influence of Group Differences 92 NOTES ON STRUCTURE 93 IMPLICATIONS FOR DECISION SUPPORT SYSTEMS 97 DDSS and the structure of Organizations: Some assertions 98 CONCLUSIONS 100 APPENDICES 102 REFERENCES .116 LIST OF TABLES TABLE PAGE 1-1 Text-Only Table Received in All Conditions (Example for the Practice Problem) 13 1-2 Sample of Computer Feedback Table (for the Practice Problem) 13 1-3 Design of the Experiment 14 2-1 Mean Initial Deviation Scores by Condition 27 2-2 Mean Group (Post Discussion) Deviations from Correct Answer, by Condition 30 2-3 Percentage Improvement in Deviation from Criterion, By Condition 31 2-4 "Collective Intelligence," by Condition 38 3-1 Initial (pre-discussion) Agreement, by Condition 44 3-2 Degree of Consensus on Final Group Ranking 45 3-3 Degree of Consensus on Final Individual Ranking 47 4-1 Variations in Number of Comments, by Condition 56 4-2 Mean Number of Run Re-Rankings, by Condition 57 4-3 Significant Pearson Correlation Coefficients Between Subject Characteristics and Performance Variables 59 5-1 Perceived Friendliness of the Group, by Condition 66 5-2 Perception of Having Reached Consensus, by Condition 66 5-3 Correlations Between Age and Subjective Satisfaction with CC 68 5-4 Satisfaction with One's Performance, by Age 68 5-5 Sex by Satisfaction with CC for Getting to Know Someone 69 5-6 Previous Computer Terminal Experience by Satisfaction with One's Performance 71