Thèse


A Minority influence Interpretation of Members' Interpersonal Processes and Team Level Outcomes using Bayesian Estimation and Bayes Factors


par Johnson, Alan R.

Édité par emlyon business school 2013 - 295 p. - En anglais

Résumé

The purpose of the study was to investigate how members’ interpersonal processes observed at the individual level determine team outcomes at the cluster level. I focus on two dimensions of interpersonal processes from previous research and theory, the extent to which members discuss and disagree about their work. I labeled the first construct dimension task debate, which is based on the information sharing literature (Stasser & Titus, 1985, 1987), and the second construct dimension task conflict, which is based on the intragroup conflict literature (Jehn & Mannix, 2001; Jehn, 1995). From minority influence theory (Moscovici, 1980, 1985a), I hypothesized that task conflict moderates (generally positive) relations between task debate and team outcomes. However, I extended De Drue and West’s (2001) contingent model by proposing that the moderated effects are (a) in alternate directions for early, middle, and late episodes and (b) generalize to several important outcome variables in the team effectiveness literature; performance, satisfaction, and creativity. Specifically, in early episodes, task conflict is a positive moderator; in middle episodes, it is negative; and in late episodes, it is positive again. I explore my thesis using a sample of 60 student teams (360 individuals) working together during 5 months (21 weeks) to write a first business plan for a new venture. I surveyed members on weeks 1, 9, and 17 about their interpersonal processes and collected data from archival, observer-rated, and self-reported sources on outcomes in week 21. I used Bayesian estimation for my latent variable modeling approach; (a) to indicator measurement error, (b) to individual selection error, and (c) to missing data on the predictor variables. Finally, a Bayes factor analysis of my informative hypothesis revealed moderately convincing support from the data for my model where no such support was forthcoming from traditional graphing and post hoc probing of the moderated relations.



Signalez un lien brisé

Chargement des enrichissements...