AAQ-II: 7 questions about what you do with painful feelings
Everyone has thoughts they would rather not have. The AAQ-II asks a narrower question: while the thought is there, do you keep doing what matters, or does the thought end up running the day? Bond and colleagues published the second version in 2011 as a measure of psychological inflexibility, the ACT term for getting tangled up in your own inner life. It has since been translated into more than 40 languages.
Seven statements, rated 1 to 7, total 7 to 49. Higher means more tangled. Nothing is reverse-scored, so the arithmetic is just addition, and the whole thing takes about three minutes.
Why there is a second version
The original AAQ was widely used and quietly unreliable. Its internal consistency often sat below the usual bar, reported around alpha .70, and people who had never read the ACT literature found the items hard to parse. The AAQ-II correlates .97 with the original, so it measures the same thing while behaving better in the data, and the Association for Contextual Behavioral Science now recommends it over the original.
Reliability
| Measure | Value |
|---|---|
| Internal consistency | mean alpha .84 (range .78 to .88 across 6 samples, 2,816 participants) |
| Retest, 3 months | r = .81 |
| Retest, 12 months | r = .79 |
Confirmatory factor analysis in three different samples, university students, people in treatment for substance misuse, and bank employees, supported a single factor: RMSEA at or below .06, CFI at or above .96, SRMR at or below .04. Measurement invariance held across the samples too.
What it correlates with
| Measure | r |
|---|---|
| BDI-II (depression) | .70 to .71 |
| BAI (anxiety) | .61 |
| DASS depression | .61 |
| DASS anxiety | .49 |
| DASS stress | .57 |
| GHQ-12 | .30 to .53 |
| SCL-90-R global severity | .70 |
| AAQ-I | .97 |
Social desirability is not a problem here: the correlation with the Marlowe-Crowne scale was -.09 and not significant. After controlling for demographics, the AAQ-II still predicted GHQ-12 distress and days of work absence a year later.
Where the cutoff argument comes from
Bond and colleagues never drew a line. They reported where established cutoffs landed when regressed onto the scale: a BDI-II score of 14 predicted an AAQ-II of 28, a GHQ-12 of 4 predicted 24 to 25, and a global severity index of 0.6 predicted 28. That is a band, not a threshold, and the paper says so.
Later work moved it up. An item response theory analysis with ROC curves landed on 28 to 32, where 28 gives sensitivity .903 and specificity .807, and 32 gives .766 and .897. The same analysis found item-level lines of 4 or higher for items 1, 3 and 4, and 5 or higher for items 2, 5, 6 and 7. Other samples go their own way: a Malay version tested in cancer patients used above 17.5, and one widely used scoring guide puts the clinical range at 25 or higher, roughly the 86th percentile. Below about 17 the items carry little information, so low scores separate people poorly.
Totals on this page are read against four descriptive ranges:
| Total | Reading |
|---|---|
| 7-16 | Comparatively flexible |
| 17-24 | Middle of the range |
| 25-32 | Inflexibility at a level linked to clinical distress |
| 33-49 | High inflexibility |
What real samples average
People with no history of mental health treatment averaged 18.5 (SD 7). People entering treatment for substance misuse averaged 28.3 (SD 9.9), a difference of d = 1.12. The Spanish version found the same gap in different numbers: 32.64 (SD 9.12) in clinical samples against 21.22 (SD 7.76) in non-clinical ones, t = 12.69.
The validity fight is worth knowing about
Wolgast (2014) ran factor analysis on the items alongside measures of distress and acceptance, and found they tracked distress more closely than acceptance. The conclusion was blunt: the AAQ-II may be measuring general suffering rather than experiential avoidance. Ruiz and colleagues (2024) answered with three Colombian samples (N = 2398, 358 and 444), where the items formed their own factor, and argued the criticism comes from using the scale as a positive measure of flexibility when it was built as a negative measure of inflexibility. Knudsen and Levin's 2025 handbook chapter takes the middle position: the overlap with general distress is real and the scale is still useful. Practically, a high score should not be read as a clean measure of one process.
Meta-analytic evidence
Across 151 studies, 25 versions of the AAQ and 43 measures of well-being, psychological inflexibility correlated with well-being at r = -.47 (95% CI -.49 to -.45). Diagnostic status, which version was used, and how well-being was measured all changed the size of that relationship.
Chinese version
Cao, Ji and Zhu (2013) validated a Chinese version with 1749 university students. Alpha was 0.88 and the one-month retest was 0.80. One factor explained 62.5% of the variance with loadings from 0.72 to 0.84, and confirmatory fit was CFI 0.99, TLI 0.97, RMSEA 0.06, SRMR 0.02. Scores correlated with the SDS and SAS between 0.26 and 0.69, and the scale still explained anxiety and depression after mindfulness and rumination were controlled (change in R2 from 0.07 to 0.13).
Structure is still argued about
Most studies assume one factor, and fit is often mediocre. A 2025 study of two Spanish-speaking student samples (n = 829 and n = 830) found a three-factor solution fit better, splitting painful memories from emotional distress and emotional control. A bifactor model put 81% of the variance on a general factor, which supports reading the total score even when the sub-factors are unstable. The original paper's 10-item version correlates .96 with the 7-item one, so published work using it remains comparable.
What this page is not
The AAQ-II is a process measure from ACT, not a diagnostic screen. A high score means you spend a lot of effort managing inner discomfort, and that effort tends to make the discomfort stickier. ACT-based therapy targets exactly that pattern. The scale is free to use in clinical and research settings, and commercial use needs the authors' permission. If distress is interfering with work, sleep or relationships, that is worth raising with a professional. This page is not a diagnosis and does not replace one.