We focus on characteristics of social networks, i.e. assortativity. We investigate the relationship  between the presence of community structure in network (in particular, we consider explicitely identified groups) and degree correlation. In our model we pre-set some node degree distribution, reflecting the nodes individual sociability. The resulting degree distribution may be, however, altered - the number of created links depends both on the individual preferences and the number of formal groups an individual is assigned to. We show that the existence of groups does not guarantee itself high degree correlation.