Investigation of causes of ceiling effects on working alliance measures

The presence of ceiling effects on measures of working alliance is important because they (a) may moderate the observed size of the alliance-outcome correlation and (b) have implications for how quickly the alliance is formed and when. Despite this, little is known about ceiling effects on alliance measures, particularly about potential causes. This study attempted to replicate findings of ceiling effects using a 7-item version of the Working Alliance Inventory (WAI) (Horvath and Greenberg, 1989) accessed in an archival database of 616 parolees enrolled in a drug abuse treatment study. Item response patterns on alliance and related measures were examined to explore potential methodological and theoretical factors that could produce ceiling effects. Analyses revealed ceiling effects on alliance measures assessing relationships with counselors and parole officers as well as floor effects (indicating highly positive appraisals) in measures of outcome expectations with counselors and parole officers. No ceiling effects were found with measures of drug use problems or negative affect. Item responses on the alliance and outcome expectations measures evidenced high consistency where many respondents endorsed the same choice on the 5-point response format across all items on the scale. Ceiling effects offer a potential marker of the working alliance at the scale level, while consistent response choice may provide a specific behavioral marker at the item level. Discussion focuses on theoretical implications and directions for future research in psychotherapy.