It would appear that this particular censustaker classed as "family" only what we would call a nuclear family (i.e., husband, wife, and children, with perhaps one of the parents absent). But if the people living in a dwelling house were unrelated (i.e., what we would call a group house) he did not class them as a family - this applies, for example, to dwelling 88, which has 4 unrelated laborers living there. This seems contrary to the instructions for 1870 census enumerators, which provided that all persons living under the same roof and provided for at a common table were to be considered a family; thus, assuming the four laborers in dwelling 88 ate together, they would have been considered a family under the instructions.
