Table 2. Interpretations of parameters in cluster-mean centered self-included model and self-excluded model.

Model a	Parameter	Interpretations based on hypothetical interventions
Cluster-mean centered self-included model (model 2a)
	β 1	(group-level intervention) Expected change in the individual-level outcome of individual i of group j, following changes in “relative placement” of individuals in the group, such that the individual-level measure of individual i increases by 1 unit while the self-included measure of group j remains constant.
	γ 2	(group-level intervention) Expected change in the individual-level outcome of individual i of group j, following an intervention such that the self-included measure of group j, and the individual-level measure of individual i both increase by 1 unit.
Self-excluded model (model 3a)
	α 1	(individual-level intervention) Expected change in the individual-level outcome of individual i of group j, following an exclusive increase in individual-level measure of individual i by 1 unit, while individual i remains in group j and the self-excluded measure of group j remains constant.
	γ 3	(individual-level intervention) Expected change in the individual-level outcome of individual i of group j, following a shift of individual i to another group, such that the self-excluded measure of group j increases by 1 unit while the individual-level measure of individual i remains constant.

^aAs explained in the main text, y_ij is an observed outcome of individual i in group j and x_ij is an individual-level social capital score of individual i in group j. Furthermore, is a self-included measure that denotes the mean of social capital scores of all individuals in group j. It is calculated as , where n_j is the size of group j. Similarly, is a self-excluded measure denoting the mean of social capital scores of all individuals excepting individual i in group j, calculated as .