<div>Hi Group.</div><div><br></div><div>I want to test that the baseline tendency of subgroups on behavior z can be explained by influence from peers on the behavior z. That is, I want to perform a mediation test.</div><div>
<br></div><div>Here is a plan: Assume network selection effects are suitable; Include a covariate effect on behavior for each subgroup to model the baseline tendency of that group on the behavior; Estimate this model; Include either a single influence effect (avg. similarity) or an influence effect plus moderators (i.e. interaction of covariate by avg. similarity) per group; Estimate this model; Calculate the difference in effect size of the covariate effect on behavior effects between the two estimated models.</div>
<div><br></div><div>What important details am I not thinking about? I appreciate any input.</div><div><br></div><div>Nathan Doogan</div><div>Doctoral Student</div><div>The Ohio State University</div><div><br></div><div>P.S. I believe I sent this with the wrong email account initially, so the previous attempt can be deleted by a moderator. Thank you.</div>