Generalizability issues in observational studies of couples: Sample characteristics and task design.