We tested linear regression models to examine to what extent the average perceived leadership quality of both formal and peer leaders (across all roles) predicted the different outcomes. The graph below presents the results for team identification. The figures also show the strength of the relationships when only including the data of research teams and ATP-teams, respectively. The other groups (board teams and POC’s) were not included given that their sample size was too low to produce reliable results.
The significance levels are set at * p < .05, ** p < .01, and *** p < .001.
These linear regression models reveal that both formal leaders and peer leaders add unique contributions to each of the outcomes. Regarding team identification, team cohesion, and organizational citizenship behaviour, peer leaders have a larger impact, both in research teams and ATP-teams. For work engagement, peer leaders have a greater contribution in research teams, while formal leaders have a greater contribution in ATP-teams. The reverse is true for job satisfaction. In terms of team effectiveness, the contributions of both leaders are similar in research teams, but within ATP-teams, the contribution of peer leaders is noticeably greater.
Controlling these effects for moderators
In order to examine if the effects observed in the previous models were influenced by other factors, we investigated the moderating effects of the following variables:
To ensure an adequate sample size, we tested the potential moderation effects separately for formal leadership roles and peer leadership roles. This allowed us to include all participants who completed the questionnaire for either formal or peer leaders, rather than only a subset who responded to both sections.
First, we assessed whether the direct effects of perceived leadership quality for either formal leadership roles or peer leadership roles changed after including all moderators as predictors. Neither the formal leadership roles nor the peer leadership roles lost their significance as direct predictors for any of the outcomes in the models that included each of the moderators.
Next, we explored the relationship between the quality of leadership roles and the moderators by testing models that included either perceived quality of formal leadership roles or perceived quality of peer leadership roles, along with the moderators as predictors and their interaction effects with the perceived leadership quality. Based on these findings, we made the following conclusions:
Overall, we can conclude that the findings from the previously shown regression models were generally unaffected by the moderators mentioned above, with the exception that older individuals appeared to be a little less influenced by the leadership quality of their peers in terms of work engagement. However, no other age-related effects were found.