Statistical methods related to Rasch modeling, social network analysis, and latent variable structural equation modeling for experimental designs are critiqued with regard to three important issues: (1) measuring constructs; (2) modeling processes of change; (3) operationalizing social relations. An argument is made for adequate conceptualization of research problems. (Author/EMK)