**4. Results**

Means, standard deviations and correlations are presented in Table 1. In order to test the dynamic interplay between the two forms of conflict and the changing intergroup perceptions we used the MEMORE macro in SPSS version 22 (Montoya and Hayes 2017). The macro allows mediation tests in repeated measures designs and the procedure does not focus on the effect of a particular independent variable but rather models the effect of change induced by a particular event in a variable (mediator) on change in another variable (dependent variable). In our design, between the two successive evaluations, parties engaged in within and between-group interactions. This systematic change is considered as main independent variable. As we argued before, interactions within one's stakeholder party are expected to trigger changes in the conflict experienced between stakeholder parties, which in turn influences the way in which other stakeholder parties are perceived during intergroup interactions and vice versa. The macro is based on a bootstrapping procedure and it includes terms that capture change as well as the average scores of the mediator variables. Overall, as interactions in the simulation unfold, stakeholder parties seem to experience less within-group relationship conflict than originally expected (effect size = −0.24, *SE* = 0.10, *CIlow* = −0.45; *CIhigh* = −0.04) and they are rated as less conflictual by others than initially expected (effect size = −0.69, *SE* = 0.10, *CIlow* = −0.90; *CIhigh* = −0.49). In Hypothesis 1 we stated that as parties start interacting, changes in relationship conflict within stakeholder parties positively coevolve with changes in perceived conflictuality in the whole system. The results support this hypothesis as the indirect effect size is significant and the 95% confidence interval does not contain zero (indirect effect size = −0.14; *SE* = 0.06; *CIlow* = −0.24; *CIhigh* = −0.01). The results of the analyses carried out with MEMORE are summarized in Figure 2 following the example reported in Montoya and Hayes (2017). As we expected, change in relationship conflict within stakeholder parties has a significant positive influence on change in perceived system conflictuality (effect size = 0.58; *SE* = 0.17 *CIlow* = 0.24; *CIhigh* = 0.91), in other words, as relationship conflict experienced within stakeholder parties change, this change triggers changes in the perceptions of conflictuality at the system level.


**Table 1.** Presents the means, standard deviations and correlations among the variables used in further analyses.

Notes: \* *p* < 0.05, \*\* *p* < 0.01; TC—task conflict within parties, RC—relationship conflict within parties, CollBY—collaborativeness ascribed by others, ConflBY—conflictuality ascribed by others, ColTO—collaborativeness ascribed to others, ConflTO—conflictuality ascribed to others.

Overall, as interactions unfold in the simulations, parties seem to experience less task conflict within stakeholder parties than originally expected (effect size = −0.23; *SE* = 0.11, *CIlow* = −0.46; *CIhigh* = −0.01) and stakeholder parties are perceived as being less collaborative by others than initially expected (effect size = −0.61; *SE* = 0.08, *CIlow* = −0.77; *CIhigh* = −0.46). Hypothesis 2 stated that as intergroup interactions begin, changes in task conflict within stakeholder parties positively coevolve with the changes in perceived system collaborativeness. Although the results do not directly support the mediation hypothesis as the 95% confidence intervals for the indirect effect include zero (indirect effect size = −0.06; *SE* = 0.04; *CIlow* = −0.14; *CIhigh* = 0.01), the direct effect of change in task on conflict within stakeholder parties on change on perceived system collaborativeness is positive and significant (effect size = 0.26; *SE* = 0.11 *CIlow* = 0.03; *CIhigh* = 0.50). In other words, change in perceived system collaborativeness is directly and positively triggered by changes in task conflict within stakeholder parties. Moreover, as also illustrated in Figure 3 the change in collaborativeness is also positively predicted by the average task conflict in the two evaluation points. We can therefore conclude that Hypothesis 2 received partial support.

**Figure 2.** The overall mediation model for change in relationship conflict and conflictuality (bottom-up influence). Notes: \* *p* < 0.05; \*\* *p* < 0.01, \*\*\* *p* < 0.001, Δrelationship conflict = relationship conflict at time 2 minus relationship conflict at time 1 and describes the negative interdependence entwinement at the stakeholder party level, Δconflictuality = conflictuality as perceived by others at time 2 minus conflictuality as perceived by others at time 1 and reflects the negative interdependence entwinement at the multiparty level.

**Figure 3.** The overall mediation model for change in tach conflict and collaborativeness (bottom-up influence). Notes: \* *p* < 0.05; \*\* *p* < 0.01, \*\*\* *p* < 0.001, Δtask conflict = task conflict at time 2 minus task conflict at time 1 (positive interdependence entwinement at the stakeholder party level), Δcollaborativeness = collaborativeness as perceived by others at time 2 minus collaborativeness as perceived by others at time 1 (positive interdependence entwinement at the multiparty level).

As interactions started during the simulation, parties seem to perceive the others as less conflictual than they expected initially (effect size = −0.83; *SE* = 0.12; *CIlow* = −10.09; *CIhigh* = −0.58). Hypothesis 3 stated that changes in conflictuality at the system level are predictive of changes of changes in relationship conflict experienced within stakeholder parties. The mediation results support the mediation role for system conflictuality as the estimated confidence interval for the indirect effect does not include zero (indirect effect size = −0.23; *SE* = 0.12; *CIlow* = −0.49; *CIhigh* = −0.01), the direct relation between change in conflictuality on change in experienced relationship conflict within parties is positive and significant (effect size = 0.28; *SE* = 0.12 *CIlow* = 0.04; *CIhigh* = 0.52) supporting a top-down effect of system conflictuality on within party relationship conflict. The overall results supporting Hypothesis 3 are presented in Figure 4.

After interactions started during the simulation, parties perceive the others as less collaborative than expected (effect size = −0.67; *SE* = 0.08; *CIlow* = −0.84; *CIhigh* = 0.51). Hypothesis 4 stated that the changes in collaborativeness at the system level co-evolve with changes in task conflict experienced within stakeholder parties. The results of this mediation analyses are presented in Figure 5 and reveal no significant indirect effect as the estimated confidence interval include zero (indirect effect size = −0.18; *SE* = 0.18; *CIlow* = −0.60; *CIhigh* = 0.11) and the direct association between change in collaborativeness at the system level and task conflict experienced within parties is also not significant (effect size = 0.27; *SE* = 0.23 *CIlow* = −0.20; *CIhigh* = 0.75).

**Figure 4.** The overall mediation model for change in conflictuality and relationship conflict (top-down influence). Notes: \* *p* < 0.05; \*\*\* *p* < 0.001, Δrelationship conflict = relationship conflict at time 2 minus relationship conflict at time 1, Δconflictuality = the conflictuality as perceived by each party for the system as a whole at time 2 minus conflictuality at the system level at time 1.

**Figure 5.** The overall mediation model for change in tach conflict and collaborativeness (top down influence). Notes: \*\*\* *p* < 0.001, Δtask conflict = task conflict at time 2 minus task conflict at time 1, Δcollaborativeness = collaborativeness as perceived by each party for the system as a whole at time 2 minus collaborativeness as for the system as a whole at time 1.
