Some network phenomena, like the small-world property, are applicable only for certain types of networks, e.g. social networks, and often need to be confirmed empirically. The Friendship Paradox, despite its name, works for all graphs regardless of domain and nature. This dissertation introduces a comprehensive generalization of the Friendship Paradox to edge weights and numerical node attributes and also introduces the Homophily Paradox, which is conceptually similar to the Friendship Paradox, but is also qualitatively different and hasn't been studied before.