“…Some studies that reported high reliabilities employed pretraining sessions for raters and focused on more specific counselling behaviours (e.g., frequency of minimal encouragers or empathic responses), that constitute a small part of the knowledge and skills that trainees are supposed to acquire during placements (e.g., Aronson, Akamatsu, & Page, 1982;Hill, Charles, & Reed, 1981). Jones, Krasner, and Howard (1992) reported high reliabilities for supervisor reports, but were actually describing agreement between ratings on two aspects of therapy by the same supervisor, rather than high interrater reliability. Finally, a recent survey (Robiner, Saltzman, Hoberman, Semrud-Clikeman, & Schirvar, 1997) indicated that most supervisors (59%) acknowledged that their own ratings were biased and also believe that other supervisors' assessments were biased in some form, with only a small percentage (11%) believing otherwise.…”