’i. s .. i: 0.5%.»: 1.6.x . . ,... « Bur“... . In? a n :1 ~ ~\ ’1 .. . :5. u if... .5» «gm. .. .i . 3.37.?“ o f... .4 It .I flit}? .5... 5. ’fi.’ ”'91.. auqaxsuathua 3.5 it. s... s 1.. 3. In. I e . 3...... 19 a. x. . .. ‘ ‘ . 4 V . ‘ ‘ . §A . . , \fi . y . .1304 . x! «:5. ix," I I affixukfiuu‘ a». fungus... . . 3 x! 31. $5.? 293: a. a. V '.r“‘ K ‘1. 7 If. I ‘3. 5.3.. 3 3.3,... 3);- 9“.“ ‘. 51(34- . ix... w Saw...“ :. i~vsavx.,.rt.§ .. r 133.. L. r r... 73.. 5:.“ r... 99:15:...1. ..i...,...§:f a. . aa:\:~y.:« El 9 a a. .....v.x t. $3191. 5:»:1}. Air... . , n . . In . - ~ A I ‘ , . 2 fig «5...? Q3» 9 E...mm\..2 #5: .. Fun.» Sana}. R . :‘1 o ‘ . ‘ u ‘ . w .VI A . ‘ ‘ . .3: . 1. .... Z. V THESIS 133-1- “BRA" M'dfiaan State niversity This is to certify that the thesis entitled EFFECTS OF LEADERSHIP AND TASK DEMONSTRABILITY ON INFORMATION REPETITION IN DECISION-MAKING GROUPS presented by Isabel Cristina Botero has been accepted towards fulfillment of the requirements for MA. degree inmmmon 5 Major professor Date December 7, 2001 0-7639 MS U is an Affirmative Action/Equal Opportunity Institution PLACE IN RETURN 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 6/01 cJCIRC/DateDuepssp 15 EFFECTS OF LEADERSHIP AND TASK DEMONSTRABILITY ON INFORMATION REPETITION IN DECISION-MAKING GROUPS By Isabel Cristina Botero A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Communication 2001 ABSTRACT EFFECTS OF LEADERSHIP AND TASK DEMONSTRABILITY ON INFORMATION REPETITION IN DECISION-MAKING GROUPS By Isabel Cristina Botero This study examined the effects of leadership and task demonstrability on the repetition of shared and unshared information in decision-making groups. Participants worked in 3-person groups (one leader and two non-leaders) to either solve a murder mystery (intellective task) or rank the three murder suspects in order of likely guilt (judgmental task). After discussion, members of groups that construed the task as intellective chose the correct suspect more often than members of groups that construed the task as judgmental. As expected, leaders repeated more information than non-leaders. However, their repetitions focused largely on shared rather than unshared information. Task demonstrability did not qualify the effects of leadership on information repetition. ACKNOWLEDGEMENTS I would like to thank my family for always encouraging me to think different and to search for my dreams. Thank you to all my committee members for helping me think outside the box and learn from your experiences. Special thanks to Dr. Gwen Wittenbaum for all her time, her ideas and helping me find the light at the end of the tunnel. Dr. Frank Boster and Dr. Dan llgen, thank you for introducing different and valuable points of view to the project, they were very useful. And last, but not least, thank you TOM. Your ideas help me understand different points of view that are very valuable and your support was essential for getting this project done. Thank you all. I. C. B. TABLE OF CONTENTS LIST OF TABLES ............................................................................. vi INTRODUCTION ............................................................................. 1 CHAPTER 1 MODERATORS OF LEADERSHIP’S EFFECT ON THE CIS BIAS ............. 3 The present study and hypothesis ................................................ 7 CHAPTER 2 METHOD ....................................................................................... 10 Design and Participants ............................................................ 10 Decision Task ........................................................................ 10 Procedure ............................................................................. 1 1 Pre-discussion Phase ..................................................... 11 Discussion Phase .......................................................... 12 Post-discussion Phase ................................................... 12 Discussion Content ................................................................. 13 CHAPTER 3 RESULTS ...................................................................................... 16 Manipulation Checks ............................................................... 16 Task Demonstrability ...................................................... 16 Member Status .............................................................. 16 Individual Preferences and Group Choices ................................... 21 Pre-discussion Choices ................................................... 21 Post-discussion Choices ................................................. 21 Group Decision ............................................................ 22 Discussion Content ................................................................. 23 Information Mentioned .................................................... 23 Information Repeated ..................................................... 24 CHAPTER 4 DISCUSSION ................................................................................. 27 Effects of Leadership on Information Repetition ............................ 27 Effects of Task Demonstrability on Member and Group Choices ....... 33 Implications and Interesting Directions ........................................ 34 Conclusion ............................................................................ 35 APPENDIX A .................................................................................. 40 Informed Consent - Murder Mystery APPENDIX B .................................................................................. 42 Pre-discussion, group discussion and post-discussion Ballot Intellective Tasks APPENDIX C .................................................................................. 46 Pre-discussion, group discussion and post-discussion Ballot Intellective Tasks APPENDIX D .................................................................................. 50 Verbal Instructions Intellective Task APPENDIX E .................................................................................. 57 Verbal Instructions Judgmental Task APPENDIX F .................................................................................. 64 Post-Discussion Questionnaire APPENDIX G .................................................................................. 69 Information Sheet APPENDIX H .................................................................................. 71 Coding Instructions BIBLIOGRAPHY .............................................................................. 83 LIST OF TABLES TABLE 1 ........................................................................................... 35 Member Status Manipulation Check Questions, and Factor Loadings from a Varimax Rotated Factor Analysis. TABLE 2 ........................................................................................... 36 Pre-discussion and Post-discussion Preferences and Group Choices as a Function of Task Demonstrability. TABLE 3 ........................................................................................... 38 Mean Proportion of Information Mentioned During Discussion as a Function of Member and Task Demonstrability TABLE 4 ........................................................................................... 39 Mean Proportion of Information Repeated During Discussion as a Function of Member and Task Demonstrability vi INTRODUCTION In many political and organizational contexts, important decisions are entrusted to groups. During the process of decision-making, group members come together to exchange information concerning the best possible decision. The diversity of knowledge, experience, and points of view that each member brings can benefit the group decision. If members effectively communicate their unique knowledge, the group may reach a better decision compared to that of a single member. Therefore, groups are assumed to be better decision-makers than individuals because group members have access to and can communicate unique information from various expertise domains. However, research has shown that groups tend to mention and repeat information that all members already know (shared information) in a greater proportion than the unique knowledge of single members (unshared information; e.g., Larson, Foster- Fishman, & Franz, 1998; Larson, Foster-Fishman, & Keys, 1994; Stasser & Stewart, 1992; Stasser, Taylor & Hanna, 1989; Stasser & Titus, 1987; Stasser, Vaughan, & Stewart, 2000; Winquist & Larson, 1998; Wittenbaum, 2000). Wittenbaum, Hubbell, and Zuckerman (1999) referred to this bias as the Collective Information Sampling (CIS) bias (see Stasser, 1999, and Wittenbaum & Stasser, 1996, for reviews). Research has tried to identify the factors that reduce the bias and increase group discussion of unshared information. One factor that may increase groups’ use of unshared information is leadership. As explained by Bass (1990), in group process the leader is the center of the group interaction, motivating members to obtain individual and group goals. The leader plays an important role for structuring and presenting information in a group, stimulating discussion, clarifying Information, and assessing consensus (Yukl, 1998). In decision-making groups, leaders play an active role in the management and use of information, coordinating the retrieval of information between members. In sum, leaders have the responsibility to keep a group focused on the task, stimulate members’ contributions, and integrate contributions to solve the problem (Maier, 1967). Not surprisingly leaders help group members to more thoroughly exchange information, particularly unshared information. To demonstrate leadership’s effect on information sampling, Larson, Christensen, Abbot, and Franz (1996) formed 3-person medical decision making teams consisting of a resident (who was accountable for the group decision), an intern, and a medical student. Teams received information about two hypothetical medical cases and diagnosed each case. Larson et al. found that leaders (i.e., the residents) repeated more unshared information than other members, and over time leaders increased the amount of unshared information repeated (see also Larson, Christensen, Franz & Abbott, 1998; Larson, Foster-Fishman & Franz, 1998). In particular, leaders repeated the unshared information communicated by other members. By doing so, leaders kept the unshared information “alive” during discussion and within the collective focus of attention (Larson et al., 1996). Therefore, leaders can benefit the information pooling process in groups. The purpose of the present study is to understand the conditions under which leadership benefits information sampling. CHAPTER 1 MODERATORS OF LEADERSHIP’S EFFECT ON THE CIS BIAS One factor that moderates the way leaders pool information in decision- making groups is leadership style. Participative leaders share their power with subordinates by actively including them in the decision making process, whereas directive leaders place less value in member input and instead direct members’ preferences toward a particular solution. Larson, et al. (1998) found that participative leaders were the ones who encouraged the discussion of more information (shared and unshared) than groups with directive leaders. Alternatively, directive leaders repeated more unshared information, particularly their own, when compared to participative leaders. Directive leaders can also influence the quality of the group decision. They may share information that supports their own preference, resulting in a low quality group decision when the leader’s preference is suboptimal (Cruz, Henningsen, & Smith, 1999). Another factor that may moderate how leaders sample information is task demonstrability. According to Laughlin (1980) and Laughlin and Ellis (1986) tasks can be defined along a continuum from intellective to judgmental depending on the demonstrability of the task solution. A task solution is said to be demonstrable when sufficient information is available for the completion of the task, the group is motivated to perform the task, members have the ability and opportunity to propose and defend solutions, and there is a shared system for identifying the correctness of the solution. Intellective tasks are considered to have a correct answer, and presumed to have an objective standard against which to evaluate the group’s decision. When group members believe that their task is to solve a problem, the goal of the group is to identify the necessary information and reasoning to find the correct answer. Judgmental tasks (also called decision making tasks) are those that are perceived as not having one demonstrably correct answer, or members believe that they have insufficient information for identifying the correct solution. As a result, the group has to reach a consensus for identifying which solution is preferred. Larson, Christensen et al. (1996; 1998) found that leaders were more likely than non-leader members to repeat unshared information using a task with intellective qualities. Members of medical decision-making teams attempted to diagnose correctly hypothetical medical cases. For both studies, teams listed possible diagnoses for medical cases, and the leader completed a team diagnosis report that asked for the probability that each diagnosis was the correct one. In addition, leaders were held responsible for the team’s accuracy. It is possible that the quest to find the right diagnosis motivated leaders to repeat information, particularly that which was unshared. Likewise, the task used by Larson, Foster-Fishman, and Franz (1998) had intellective qualities. Three-person groups were asked to identify which of three hypothetical professors would be the best one to teach an advanced undergraduate course in personality psychology. The experimental materials were constructed such that there was a superior professor, and group members’ task was to correctly identify that professor. The search to find the best professor may have motivated leaders to repeat unshared information. Other research suggests that intellective tasks, as compared to judgmental tasks, inspire group members to pool information more thoroughly. Stasser and Stewart (1992) proposed that when groups work on judgmental tasks, the way that members share information in discussion is guided by an attempt to reach a consensus. Thus, if the shared information leads to agreement, members may not explore unshared information. Alternatively, when a task is intellective the assumption of demonstrability may lead members to search for a critical set of information that allows group members to identify and defend a superior choice. To test this assumption they asked university students to read and review evidence in a homicide investigation. There were three possible suspects, but two of these could be ruled out and the third one could be implicated. In the judgmental conditions, the participants were told that their task was to decide which suspect “was the most likely to have committed the crime” and noted that they might not have enough information to determine definitely who was the guilty suspect. Under the intellective conditions, they told participants that their task was to determine who was the guilty suspect. Stewart and Stasser found that groups discussed more unshared information when they believed that the case could be solved, supporting the idea that information exchange in decision making groups is affected by a task’s perceived demonstrability. When a task is judgmental, there is a tendency to pool shared information, instead of unshared information. When the task is intellective, groups will take into account more unshared information when reaching a decision. Given that intellective tasks facilitate groups’ use of unshared information, relative to judgmental tasks, Larson and colleagues’ finding that leaders repeat unshared information may be pronounced for tasks such as theirs, with intellective qualities. Leadership may facilitate the pooling of unshared information when groups work on intellective tasks for at least three reasons. First, intellective tasks may require that members effectively pool information in order to find the correct solution. Although information pooling is not necessary to solve all types of intellective tasks (eg. math problems), successful completion of tasks used to study the information sharing in groups have required information exchange. That is, the optimal decision alternative was apparent only when all unshared information was communicated. Leaders should be more sensitive to task demands than non-leaders because it is the responsibility of leaders to motivate the group to perform the task. If leaders recognize the demand to pool information when the task is intellective, they may take responsibility for facilitating information exchange and repeating information that is critical for task solution. Likewise, leaders should recognize that judgmental tasks require consensus seeking rather than information pooling and should emphasize the former instead of the latter. Second, group discussions last longer when the task is intellective compared to judgmental (Stasser & Stewart, 1992). Because leaders’ tendency to repeat unshared information increases over time, while non- leaders’ tendency to repeat unshared information decreases over time, increases in discussion length should increase the disparity between leaders’ and non- leaders repetition of unshared information. Thus, if discussions last longer for intellective than judgmental tasks, the tendency for leaders to repeat unshared information more than non-leaders will be greater for intellective than judgmental tasks. Third, confident members have more influence in decision-making groups when the task is intellective compared to judgmental (Zarnoth & Sniezek, 1997). Because leaders likely have more task confidence than non-leader members, leaders may exert greater influence on information pooling when the groups’ task is intellective rather than judgmental. In sum, the tendency for leaders to keep unshared information alive during discussion will be stronger in groups that work on intellective rather than judgmental tasks. The Present Study and Hypotheses The present study examined the effects of leadership and task demonstrability on information repetition in decision-making groups. Participants received information about a homicide investigation and the perceived task demonstrability manipulation used by Stasser and Stewart (1992) and Stewart and Stasser (1998). Participants individually read and reviewed evidence in a homicide investigation, and then determined the guilty suspect in three-person groups. Each member read unshared clues that other members did not read, and shared clues that all members read. The evidence in this mystery case suggested three possible suspects, and instructions led them to perceive a task as intellective orjudgmental. For the judgmental condition, participants were told that they should decide which suspect “was the most likely to have committed the crime” and there were indications that they did not have sufficient information to determine the guilty suspect. For the intellective condition, participants were told that their task was to solve the mystery by uncovering the correct guilty suspect. For both conditions, one member played the role of the group leader. The leaders were older and had more experience, knowledge, responsibility and education in the decision-making process compared to non-leader members. Leadership style was not specified given that, overall, leaders repeat more unshared than shared information, regardless of their leadership style (Larson, Foster-Fishman, & Franz, 1998). Stasser and Stewart (1992) and Stewart and Stasser (1998) found that when group members perceived that the homicide case could be solved (intellective task) they were more likely to choose the guilty suspect compared to group members who perceived that they did not have enough information to solve the case (judgmental task). As a replication of the work of Stasser and Stewart and Stewart and Stasser, the following hypothesis was formulated: Hypothesis 1: A higher proportion of groups in the intellective condition will choose the correct compared to groups in the judgmental condition. The present study also aimed to replicate the findings from Larson and colleagues (1996, 1998) where leaders repeated more information than non- leaders and especially more unshared information than the other group members. The following hypotheses predict replications of past research. Hypothesis 2: Leaders will repeat a higher proportion of total clues compared to non-leaders. Hypothesis 3: Leaders will repeat a higher proportion of unshared clues compared to non-leaders. Although the research of Larson and colleagues (1996, 1998) used a decision-making task, the researchers emphasized decision accuracy when describing the task requirements to group members. Perceiving the decision- making task as being intellective may have enhanced leaders’ use of unshared information. The present study expected that the perceived task demonstrability would affect the tendency for leaders to repeat unshared Information. A perceived intellective task should create greater demand for leaders to repeat unshared information compared to a perceived judgmental task. The following hypothesis was formulated: Hypothesis 4: The tendency for leaders to repeat a higher proportion of unshared clues compared to non-leaders will be stronger when the task is perceived as intellective compared to when it is perceived as judgmental. CHAPTER 2 METHOD Design and Participants The present study used a 2 x 2 mixed factorial design. The factors studied were Task Demonstrability (intellective vs. judgmental), and Member (leader vs. non-leader), with the latter factor occurring within groups. A total of 223 undergraduate students at a large midwestern university participated in partial fulfillment of a research requirement for their communication courses. Participants were randomly assigned to mixed-sex, three-person groups comprised of two students from an introductory communication class (i.e., non- leaders) and one student from a leadership and group communication class (i.e., leaders). There were a total of 50 three-person groups, twenty-four in the intellective task condition and twenty-six in the judgmental task condition. Sixty- two participants remained after groups were formed. These participants worked on individual tasks; their data are excluded from this report. Decision Task Participants read a 27-page booklet containing interviews from a homicide investigation. The booklet had interviews with the three suspects and related characters, a newspaper article, area maps, and a hand-written note. This information contained 24 critical clues that incriminated or exonerated the three suspects in the case (E, B, and M). Six clues incriminated each suspect, but 3 clues exonerated each of Suspects B and M. The critical clues showed that Suspect E had the motive and opportunity to commit the crime and that he had 10 attempted to frame Suspect B. Of the 24 critical clues, 15 were distributed to all members as shared information. The remaining 9 clues were critical in that they were necessary in identifying the guilty suspect. Of these 9 critical clues, three exonerated Suspect B, three exonerated Suspect M, and three incriminated Suspect E. Each member of the group received 3 of these critical clues as unshared information - one critical clue about each suspect. To assure that leaders’ behavior was not due to the kind of critical clues they read, leaders received each of the three sets of critical clues an equal number of times in each task demonstrability condition (i.e., unshared information was counterbalanced across members). There were also 28 additional pieces of information (i.e., M) that were shared by all members but were not needed to solve the mystery. Procedure Pre-discussion phase. Groups were formed of two introductory communication students and one leader from a leadership and group communication course. As students arrived, they were welcomed and assigned to a group. The experimenter then asked participants to sign a consent form (see Appendix A) that explained the study’s purpose and procedures. After signing the consent form, participants received the murder case material and pre-discussion ballot (see Appendices B and C). At this point, the experimenter introduced the leader by saying that “member number one” was assigned as the group leader because he or she was a student in the leadership and group communication course, and that this student had the most leadership expertise of anyone else 11 the group. Then verbal instructions were varied according to task demonstrability (judgmental or intellective; see Appendices D and E). The judgmental groups were told “the detectives in this case did not have sufficient evidence to charge anybody for the homicide, so we are asking you to use the available evidence and rank the suspects in order of who seems more likely to have committed the crime." The pre-discussion ballot for the judgmental groups asked members to rank-order the suspects from most to least likely to have committed the crime. The intellective task participants were told that “only one of the suspects could have committed the crime, and we would like you to read over the material carefully and correctly choose the guilty suspect.” The pre-discussion ballot asked members to select individually the one suspect who committed the crime. When all members finished reading the case information, the materials and ballots were collected. Discussion phase. Participants had up to 30 minutes to discuss the mystery and determine the guilty suspect for the intellective task, or the likelihood of guilt, for the judgmental task. Leaders were responsible for completing a group discussion ballot (see Appendices B and C) that asked for the order of the suspects for the judgmental condition, or the guilty suspect for the intellective task. These discussions were videotaped. Post-discussion phase. After discussion had finished, each participant individually completed a post-discussion ballot (see Appendices B and C) again asking for members’ suspect choice(s), followed by a post-discussion questionnaire that asked for members’ impressions of one another and checks 12 on the manipulations (see Appendix F). Participants were debriefed, given an information sheet (see Appendix G) and credit, and an opportunity to receive by e-mail the final results and solution to the mystery. Discussion Content Three coders, who were blind to the experimental hypotheses, independently viewed two thirds of the group discussions. Two different coders coded each group discussion (See the coding instructions in Appendix H). Each coder recorded every informational utterance along with the member who made the utterance. The utterances included statements citing information from the murder mystery booklet. To be counted as correct information, a speaker needed to convey, explicitly or by context, the essential meaning of a clue and the suspect to whom that clue was related, as cited in the mystery booklet. An item was considered to be repeated if the discussion moved to another topic and then returned to a piece of information that was mentioned earlier. From these coding protocols the proportion of information mentioned and repeated were calculated. The dependent variables calculated from the codings were: proportion of shared clues, unshared clues and details mentioned and repeated. Measures were calculated for leaders and non-leaders of each group. The proportion of shared information mentioned was calculated by dividing the number of shared items mentioned at least once by the total number of shared items (Le, 15). The proportion of unshared information mentioned was calculated by dividing the number of unshared items mentioned at least once by the total number of unshared items available to each member (i.e., 3). The proportion of 13 details mentioned was calculated by dividing the number of details mentioned at least once by the total number of details (i.e., 28). The proportion of total information mentioned by each member was calculated by first adding the number of shared items, unshared items and details mentioned by each member, and then dividing by the total number of clues available to each member (i.e., 46). The repetitions were calculated as a proportion of shared clues, unshared clues, details, or total clues that were mentioned, which were later repeated.1 The proportion of shared items repeated was calculated by dividing the number 1 Repetitions were also computed in three additional ways. First, to replicate Larson’s studies, the proportion of shared clues, unshared clues, details and total clues repeated were calculated by dividing the number of times items were repeated by the total items that were mentioned. This measure of repetitions differs from the one reported in the body of the paper in that it counts the number of times an item was repeated to achieve a measure of repetitions per item mentioned. Shared repetitions were calculated by counting the number of times shared items were repeated by each member and dividing it by the number of shared items that were mentioned by group. Unshared repetitions were calculated by counting the number of times that each member repeated unshared items and dividing it by the number of unshared items mentioned by the group. Details and total repetitions were calculated the same way. Second, repetitions were computed by counting the number of clues (shared, unshared, details and total) that were repeated one or more times (i.e., number of items repeated). Third, repetitions were calculated by counting the number of times each member repeated each type of clue (shared, unshared, details and total). This measure identified the number of times clues were repeated rather than the number of items repeated. Both of these later measures of repetitions were frequencies rather than proportions adjusted for information mentioned. The original and three additional measures of repetitions were submitted to the same significance tests. None of the statistical conclusions changed with the different analyses. 14 of shared clues that were repeated by each member by the number of shared clues that were mentioned by the group. The proportion of unshared clues repeated was calculated by dividing the number of unshared clues that each member repeated by the number unshared items previously mentioned. The proportion of details repeated was calculated by dividing the number of details repeated by each member by the number of details previously mentioned by the group. And, total information repeated was computed by dividing all the information (shared, unshared, and details) repeated by any by the total clues that were mentioned by the group. The coder reliability estimates were obtained by correlating the measures taken from two independent codings across 50 group discussions. Because the following analyses used the average of the measure taken from the two codings, these correlations were adjusted (via Speannan-Brown prophesy formula) to obtain the estimated reliability of the average. The resulting reliability estimates are given for each dependent measure. The reliabilities were .91 for shared items mentioned, .96 for unshared items mentioned, .97 for details mentioned,.98 for total clues mentioned, .91 for shared items repeated, .91 for unshared items repeated, .93 for details repeated, and .96 for total repetitions. 15 CHAPTER 3 RESULTS Manipulation Checks Task Demonstrability. Three items tested the success of the task demonstrability manipulation: (a) “The information I had was sufficient to determine who was the guilty suspect, " (b) “l was given enough information to find a solution to the mystery, " and (c) “The mystery was solvable.” Because these items were highly correlated, a composite measure was created by averaging the three items (Cronbach’s a=.95). This measure was analyzed in a Task Demonstrability (intellective vs. judgmental) x Member (leader vs. non- leader) mixed factorial analysis of variance (ANOVA) with the member factor occurring within groups. Intellective groups (M=6.01, S_D= 1.68) perceived that the information provided was more sufficient to solve the mystery compared to judgmental groups (Mares, S_D= 1.83), E (1,48): 11.05, p< .01, 112: .13. The effect of member and its interaction with task demonstrability were non- significant. Therefore, the task demonstrability manipulation was successful. Member Status. Several measures from the post-discussion questionnaire were used to assess the accuracy of the leadership manipulation. Leaders as compared to non-leaders, were expected to have more years in college, be older, and have taken more leadership and group communication courses. Overall seventy-eight percent of the leaders (3:38) were seniors, twenty two percent (fl=1 1) were juniors, and none were freshmen or sophomores. One leader failed 16 to answer this question. Of the non-leaders, thirty percent (l!=29) were freshmen, thirty-nine percent (51:38) were sophomores, fourteen percent (fl=14) were juniors, and seventeen percent (3:17) were seniors. Two non-leaders did not answer this question. Leaders were also older in years (M= 22.22, §Q= 4.26) than non-leaders (M: 19.51, SD: 1.07), E (1, 46)= 19.55, p<.001, 112:.16. Ninety- one percent of the leaders had previously taken a leadership and group communication course, while only 23 percent of non-leaders reported having taken a leadership course.2 Overall chi-square analyses showed that type of member (leader vs. non-leader) and having taken a leadership and group communication course (yes vs. no) were not independent, x2 (1 ,fl=143)= 57.49, p<0.001. Four leaders and three non-leaders did not answer this question. Direct comparisons using z-test for the difference in proportions showed that leaders reported having taken a leadership and group communication course (91%) more than expected by chance, g: 2.96, g<.001, whereas, non-leaders reported to have taken a leadership and group communication course (24%) less than expected by chance, g: - 7.58, p<.001. A direct comparison showed that leaders were more likely to take a leadership and group communication class than non- leaders, x2 (1,fl=65)= 5.55, g