Response to Freedman's critique of path analysis discusses appropriate ways to represent social processes in numerical form and the use of path analysis to achieve that goal. Discusses two main issues: the selection of variables and whether multiple regression can be used to model social processes. (RB)