*3.2. Measures*

**Task conflict** was evaluated with four items adapted from a conflict scale presented in Jehn (1995). For the first evaluation, at the onset of the simulation the content of the items were adapted to capture

expectations ("To what extent do you expect disagreements in your interest party related to the task?", "To what extent do you expect differences of opinion in your stakeholder party?", "How often do you think the members of your interest party will disagree about how things should be done?", "How often do you think the members of your interest party will disagree about which procedure should be used to do your work?"), while for the subsequent evaluation, the items referred to experienced task disagreements within one's stakeholder party (e.g., "To what extent are there disagreements in your stakeholder party that are related to the task," etc.). The answers were recorded on a 5-point Likert scale (1 to 5) and Cronbach's alpha for this scale was 0.94 at time 1 and .88 at time 2 for the group level aggregated items. In order to support aggregation, we used computed RWG (James et al. 1984) that ranged between 0.82 and 0.98 supporting the aggregation of the scores at the group level of analysis. The change from expectations of task conflict as evaluated before the interactions started to the experienced task conflict will therefore reflect the entwinement of positive interdependence at the stakeholder party level.

**Relationship conflict** was evaluated with four items adapted from the same conflict scale (Jehn 1995) and, like with the task conflict scale, at time 1 referring to expectations ("How much jealousy or rivalry do you expect to see among the members of your interest party?", "How often do you expect to have personality conflicts in your interest party?", "How much tension do you think will exist among the members of your stakeholder party?", "How often do you think people will ge<sup>t</sup> angry while working in your interest party?"). At time 2 the items referred to experienced interpersonal frictions (e.g., "How much jealousy or rivalry is there among the members of your stakeholder party," etc.). The same 5-point Likert scale was used (1 to 5) to record the answers. Cronbach's alpha for this scale was 0.95 at time 1 and .85 at time 2 for the items aggregated at the group level. The RWG scores ranged from 0.73 to 1.00 supporting the aggregation of the scores at the group level. The change from expected relationship conflict to experienced relationship conflict will therefore be indicative of the entwinement of negative interdependence at the stakeholder party level.

**Collaborativeness** was evaluated using a round robin technique in which each participant was asked to evaluate the extent to which he or she perceived each of the participating stakeholder parties as being collaborative. At time 1 participants were asked to estimate the collaborativeness of each stakeholder party based on the reading of the briefing material ("Please evaluate how collaborative you think each stakeholder party will be." The evaluations range from 0 = not collaborative at all to 5 = very collaborative"), while at time 2 they were asked to evaluate the experienced collaborativeness ("Please evaluate the extent to which [each stakeholder party] is collaborative." The evaluations range from 0 = not collaborative at all to 5 = very collaborative"). To capture bottom-up influences (what individual parties bring into the larger system), we aggregated evaluations for each stakeholder party as a referent—how each stakeholder party is perceived by the other stakeholder parties in the system. As an index of collaborativeness for bottom-up influences, we used the average score for the collaborativeness of each stakeholder party as perceived by the other stakeholder parties (total system-level score excluding one's own party)—that is the average collaborativeness ascribed to a particular stakeholder party by the others in the system. To capture top-down processes (how the multiparty interactions impact the within-party climate), we have aggregated evaluations using the whole system as a referent—how each stakeholder party perceives all the other stakeholder parties. As an index of collaborativeness for the top-down influences we have used the average score for how collaborative each party sees all the other stakeholder parties in the system (again excluding one's own party)—that is the collaborativeness ascribed to all other stakeholders in the system by a particular party. In other words, we have use collaborativeness of each stakeholder party as evaluated BY others as an index in the bottom-up analyses and collaborativeness attributed TO all the others by each stakeholder party as an index in the top-down analyses. Change from the expected to the experienced collaborativeness reflects the entwinement of positive interdependence and the two indices will be separately used to capture the bottom-up and top-down influences.

**Conflictuality** was evaluated in a similar fashion as collaborativeness, using a single item that referred to expected conflictuality of each party at time 1 ("Please evaluate how conflictual you think each party will be. The evaluations range from 0 = not conflictual at all to 5 = very conflictual") and evaluated experienced conflictual relations at time 2 ("Please evaluate the conflictuality of ... [each of the other stakeholder parties]". The evaluations range from 0 = not conflictual at all to 5 = very conflictual"). To capture the bottom-up influences, a conflictuality score was computed by averaging all scores received by each stakeholder party from all the other stakeholder parties (total system-level score, excluding one's own party)—that is, conflictuality ascribed to a particular stakeholder party by all other parties in the system. To capture the top-down influences, we have used the average score for how each stakeholder party perceived the conflictuality of all the other stakeholder parties in the system (excluding one's own party)—that is conflictuality ascribed by a particular party to all the other parties in the system. We have therefore used conflictuality as evaluated BY others as an index in the bottom-up analyses and conflictuality in the system (ascribed TO others) as a whole for the top-down analyses. As for collaborativeness, the change from expected to experienced conflictuality indicates the entwinement of negative interdependence and the two indices are used to capture bottom-up and top-down influences.

To summarize, in order to capture the top-down and bottom up influences, we have used different aggregation procedures of the data collected through the round-robin procedure for collaborativeness and conflictuality. For the bottom-up influences the aggregation reflects what each party brings into the system (collaborativeness and conflictuality of each party as seen by others), while for the top-down influences, the aggregation reflects what each party perceives in the system as a whole (collaborativeness and conflictuality ascribed to others in the system).
