WAI Scoring Guide · Working Alliance Inventory

What does the WAI measure?

The Working Alliance Inventory (WAI) is a therapeutic alliance questionnaire developed by Adam O. Horvath and Leslie S. Greenberg at Simon Fraser University ^[1].

It operationalises Bordin's pantheoretical model of the working alliance, which defines the alliance by three elements ^[4]:

1. Bond — The quality of the personal attachment between client and therapist; feeling understood, cared for, and trusted
2. Task — Mutual agreement and understanding of the specific activities undertaken in therapy
3. Goal — Consensus on the targets and outcomes of treatment

The WAI can be completed by the client (client form), therapist (therapist form), or an observer rating a session (observer form). All three perspectives have been validated and are used in both research and clinical training.

How do you give the WAI?

The WAI is completed after a session by the client (client form), the therapist (therapist form) or an observer rating a recorded session (observer form). It is longer than session-by-session alliance checks such as the SRS, so it is usually given at set points in treatment (for example, sessions 3 and 8) rather than every session. Items are rated on a 7-point scale from 1 (Never) to 7 (Always).

WAI versions and items

Original WAI ^[1]

36 items, 7-point Likert scale (1 = Never to 7 = Always). Three subscales of 12 items each (Bond, Task, Goal). Total score range: 36–252.

WAI-S — Short Form ^[2]

12 items selected from the original 36, 4 items per subscale. Total score range: 12–84. Factor structure replicates the three-factor model.

WAI-SR — Short Revised ^[3]

12 items, a refined short form first validated in US psychotherapy outpatients and later tested in German outpatients and inpatients, where it measured the same Bond, Task and Goal constructs in both settings.

How do you score and interpret the WAI?

Items are rated on a 7-point scale:

• 1 = Never
• 2 = Rarely
• 3 = Occasionally
• 4 = Sometimes
• 5 = Often
• 6 = Very often
• 7 = Always

Some items are reverse-scored — consult the scoring key for your specific version (WAI, WAI-S, or WAI-SR). Reverse-scored items represent low alliance (e.g., "I am not sure what we are trying to accomplish in therapy").

Subscale scores are sums of the respective items. Total score = sum of all subscale scores.

Interpreting WAI scores

The WAI does not have a single universally accepted clinical cutoff like the ORS or SRS. Interpretation relies on comparison with normative data and within-client change over time. Higher scores mean a stronger alliance, and a falling score between administrations is worth raising with the client.

Subscale patterns matter

Each subscale profile points to a different clinical response:

• Low Bond. The client does not feel personally understood or cared for.
• Low Task. Client and therapist disagree on what they are actually doing in sessions.
• Low Goal. The client's treatment goals are not aligning with the therapist's understanding.

Psychometric properties

• WAI-SR reliability and validity. In German outpatients (n = 88) and inpatients (n = 243), reliability was good (alpha > 0.80), as was convergent validity with the Helping Alliance Questionnaire (r > 0.64) ^[3].
• Factor structure. Confirmatory factor analysis supported the Bond-Task-Goal model in both settings; inpatients reported lower alliance and less differentiated components ^[3].
• Prediction of outcome. A meta-analysis of 295 studies covering more than 30,000 participants found an overall alliance-outcome correlation of r = .278 in face-to-face psychotherapy, consistent across assessor perspectives, alliance measures, treatment approaches and countries ^[5].

Limitations

• No validated cutoff. Interpretation relies on norms and within-client change, not validated thresholds.
• Version differences. The 36-item WAI, WAI-S and WAI-SR use different item sets and reverse-scoring keys, so scores are not interchangeable.
• Setting effects. Alliance ratings and the separation of Bond, Task and Goal differ between outpatient and inpatient settings.
• Perspective matters. Client, therapist and observer ratings can differ; record which form was used.

Does the alliance predict therapy outcome?

A meta-analysis of 295 studies found a consistent moderate association between alliance and outcome (r = .28), holding across treatment approaches ^[5]. Because the alliance predicts outcome whether CBT or another approach is used, the WAI fits any therapy model. A later review describes the alliance, typically measured at the third or fourth session, as a robust predictor of outcome even when prior symptom improvement is taken into account ^[4].

WAI in clinical training and supervision

The WAI is widely used in:

• Clinical training — tracking alliance development across a trainee's early sessions
• Supervision — comparing client and therapist forms to identify alliance blind spots
• Research — alliance as mediator/moderator in clinical trials
• Periodic check-ins — repeating the WAI at set points in treatment to see how the alliance is developing