Homophily is a concept in social network analysis that states that in a network a link is more probable, if the two individuals have a common characteristic. We study the question if an observer can assess homophily by looking at the node-link diagram of the network. We design an experiment that investigates three different layout algorithms and asks the users to estimate the degree of homophily in the displayed network. One of the layout algorithms is a classical force-directed method, the other two are designed to improve node distinction based on the common characteristic. We study how each of the three layout algorithms helps to get a fair estimate, and whether there is a tendency to over or underestimate the degree of homophily. The stimuli in our experiments use different network sizes and different proportions of the cluster sizes.