Mixing by degree in signed social networks

Social networks have empirically been found to be assortative (i.e., the degree of neighboring nodes are positively correlated), while other networks (e.g., technological, biological) show the opposite pattern (disassortative). Why is that so?
How do these patterns change in signed networks, where relations indicate trust/distrust, friendship/enmity? Do individuals who dislike many others tend to dislike each other, or do they dislike those who dislike only very few others?

V. Ciotti, G. Bianconi, A. Capocci, F. Colaiori, and P. Panzarasa (2014), Degree correlations in signed social networks.