N typical); hence, we employed the college imply to impute missing values. Accordingly, we treated each handle variable as a continuous measure and, instead of utilizing nodematch to assess similarity, we used the absdiff term, which measures absolute differences amongst members of each and every dyad. We expected larger absolute differences to lower the likelihood of ties, resulting in unfavorable coefficient estimates. An exception was created for race/ethnicity, which really is usually a categorical measure. Here we specified missing values as their own category and utilised the nodematch term, permitting coefficient estimates to differ across categories. One particular benefit of ERGMs is the fact that they let one particular to manage for network processes that also drive structure. The indegree and outdegree effects manage for the distributions of incoming and outgoing ties within the network, respectively. The edgewise shared partners term estimates the likelihood of a tie based around the quantity of partners linking the two men and women. The dyadwise shared partners term controls for the likelihood of men and women devoid of a tie sharing numerous partners. Every single in the 4 aforementioned network terms is geometrically weighted, enabling superior estimation of models.34 Finally, the reciprocity impact captures the tendency to reciprocate purchase IMR-1 friendship nominations. The ergm package in R version three.1 (R Foundation for Statistical Computing, Vienna, Austria) was used in estimating models. We fit every single model specification separately by college after which performed a meta-analysis to combine benefits.RESULTSWe start with two preliminary models that tested irrespective of whether the patterns anticipated in the avoidance and homophily mechanisms had been present. Our 1st preliminary model estimated the likelihood of overweight students being chosen as a pal. Constant with the avoidance mechanism, the BMI indegree coefficient was negative (b = .23; 95 self-confidence interval [CI] = .27, .19; P .001), indicatingthat overweight students have been much less most likely than nonoverweight students to be selected as a pal. The estimated coefficient represents the change within the log odds of a friendship for alters who’re overweight versus not overweight. Exponentiation revealed that the odds of picking an overweight pal have been 79 (e.23) from the odds of picking a nonoverweight pal, or 21 reduce. Our second preliminary model estimated the likelihood of students picking a buddy using the similar weight status as themselves. Consistent with homophily, the estimated BMI nodematch coefficient was positive (b = 0.14; 95 CI = 0.10, 0.18; P .001). Students were 15 (e0.14) extra probably to befriend a peer whose weight status matched their very own than to befriend a peer having a unique weight status. The PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/20070607 preliminary models indicated that each the homophily and marginalization patterns were present. However, what we truly want to know is how most likely friendships have been for each and every combination of ego and alter BMI. This goal required that we simultaneously model all 3 BMI effects: BMI nodematch, indegree, and outdegree. Model 1 included only these 3 effects, whereas models 2 and three also incorporated controls for endogenous selection mechanisms (model two), homophilous choice on individual attributes (models two and three), and indegree and outdegree based on person attributes (model 3). Ideally, we would involve all controls in 1 model; having said that, difficulties with model convergence prevented that. Alternatively, we present numerous model specifications and compare outcomes. As shown in Table 3,.