A total of 218 MBA students from Northwestern University completed the Bullard
Houses negotiation (Karp, et al., 2008). As in Study 1, one week prior to negotiating, students
were randomly assigned to a seller role or a buyer role. Each student was randomly assigned
negotiate as a solo or as a two-person team. Because Study 1 revealed no significant difference
between two-member and three-member teams, all teams in Study 2 contained two members.
The unit of analysis was the negotiation (N = 73 negotiations).
Immediately following the negotiation, all participants completed a post-negotiation
questionnaire in which they rated their counterpart’s behavior with five semantic differentials:
pleasant-unpleasant; friendly-unfriendly; cooperative-uncooperative; competitive-
noncompetitive (reversed); quarrelsome-harmonious (reversed). The items ranged from 1 to 5,
with higher scores indicating more disagreeable behavior. Because the unit of analysis was the
negotiation (not the individual participant), we averaged the ratings from all the participants in
the negotiation to form a composite index of perceived disagreeable behavior (α= .84).
We tested H1 with a 2 (buyer size: solo vs. team) X 2 (seller size: solo vs. team) logistic
regression (teams coded -.5; solos coded .5). We expected a significant effect of seller size such
that the presence of a seller team would increase the likelihood of impasse.
We tested H2 and H3 with a linear regression analysis that included three planned
orthogonal contrasts (J. Cohen et al., 2003). These contrasts were: (c1) solo-buyer/solo-seller
coded .75, team-buyer/solo-seller coded -.25, solo-buyer/team-seller coded -.25, team-
buyer/team-seller coded -.25; (c2) solo-buyer/solo-seller coded 0, team-buyer/solo-seller coded -
.33, solo-buyer/team-seller coded -.33, team-buyer/team-seller coded .67; (c3) solo-buyer/solo-
seller coded 0; team-buyer/solo-seller coded -.5; solo-buyer/team-seller coded .5; team