The main problem isn't that it isn't a science, but that to the extent that science is used in sociology, it's mostly badly done.

Findings don't replicate, data is used to tell lies, policy built on scientific findings is mostly unrelated to the findings, focus is primarily guided by culture instead of discovery or curiosity. I read a lot of sociology, and the general level of critical thinking and concern for the truth in journal articles is much lower than biology, engineering, or even economics.