Although Sociology was born with scientific vocation, its current status as a scientific discipline seems not to enjoy good health. Departing from this idea, the purpose of the paper is threefold: first, several arguments for explanations of social phenomena through mechanisms will be reviewed. Such explanations can offer a suitable way to Sociology to be able to develop as an authentic science. Second, the idea that such explanations can be computational programmes created by agent-based modelling will be justified. Finally, it will build new arguments about how such explanations can avoid the ambiguity and the obscurity conceptual that a part of sociology suffers; reduce its theoretical fragmentation and strengthen the multidisciplinary dialogue with the purpose of researching in depth social reality as well as being able to design effective policies against several social problems.