Refining the Analysis of Mechanism-Outcome Relationships for Anxiety Treatment: A Preliminary Investigation Using Mixed Models.

The team tracked adults in exposure therapy for anxiety. They used mixed models to test which numbers predict later symptom drops.

They looked at three things each session: self-efficacy, within-session fear ratings, and between-session habituation slopes.

They also checked if attention-bias-modification scores linked to outcome.

Self-efficacy growth and habituation slopes across sessions predicted improvement.

Surprisingly, the fear ratings taken during each exposure did not predict gains.

Attention-bias scores showed no link to outcome at all.

Carr et al. (1985) already warned that simple anxiety ratings bounce around and agree poorly across methods. Vassos et al. (2023) now show these same ratings also fail to forecast who gets better.

Geckeler et al. (2000) found that initial heart-rate, not fear ratings, predicted return of claustrophobic fear. Together the two papers suggest you should trust physiology or slope data more than moment-to-moment “How anxious are you?” scores.

McGeown et al. (2013) showed earlier exposure sessions can shorten total treatment. Vassos et al. (2023) add that watching the slope of habituation across those early sessions tells you if the client is on track.

Stop using single fear ratings to decide if exposure is working. Instead, graph self-efficacy and habituation across sessions. A rising self-efficacy line plus a downward anxiety slope means keep going. Flat lines mean tweak the task or dose. This small shift gives you an early warning system and clearer data for supervision.