Hypothesis 4: The more personal ties members have with an infractor, the more likely they are to choose as sanctioners other members with whom they are also personally close.
Each partner (or 'respondent') was thus asked to freely choose one or a set of colleagues (or 'sanctioners') who would be in charge of handling, at this early stage, the problem created by a specific partner (or 'infractor').
I can thus assess the dependence of the choices of sanctioners on the existence of interdependencies between respondent and sanctioner, or between respondent and infractor. These data are also used to locate the partners in the informal structure of their firm and for correlations between importance as a sanctioner and importance in the control of such resources.
To test Hypothesis 2, I used the sociometric data collected in the firm (see Appendix) to look at the effect of relations between respondent, sanctioner, and infractor on the choices of sanctioners.
Multiple regression analyses were replicated across the 36 individual 'sanctioner by infractor' matrices.
Across the 36 partners, two effects stand out as widely significant: first, the fact that 'infractor goes to sanctioner for advice' (positive and significant parameter for 18 partners out of 36) and the fact that 'infractor sees the sanctioner as a friend' (positive and significant parameter for 16 partners out of 36).
The first model shows that the fact that a respondent chooses a friend as a sanctioner (rather than someone who is not a friend) is indeed explained by the fact that the same respondent considers the infractor to be a friend.