Tests · anxiety sensitivity
Anxiety Sensitivity Test (ASI-3)
The Anxiety Sensitivity Index-3 (Taylor et al., 2007) measures the fear of anxiety sensations themselves: a racing heart, not being able to concentrate, other people noticing that you're on edge. A high score is linked to panic. Rate how much you agree with each statement.
- 18 questions
- Taylor et al., 2007
- About this test
Answer all 18 questions.
What's next
In a meta-analysis of 24 randomized trials, CBT lowered anxiety sensitivity, with a large effect in people seeking treatment (Smits et al., 2008). The usual tools are interoceptive exercises, which bring on the feared sensations on purpose, and checking the thoughts you have about them. The app has exercises for anxiety and panic, a journal, and a chat with the AI so you can talk through your result.
About the ASI-3
How much you fear the body, mind and social effects of feeling anxious, such as a racing heart, a mind that goes blank or other people noticing you're nervous.
- Answers
- 0 (very little) to 4 (very much)
- Subscales
- Physical, cognitive and social concerns, six items each
- Time frame
- None. You rate each statement as a general description of yourself.
- Developed by
- Steven Taylor, Michael J. Zvolensky, Brian J. Cox and 13 co-authors, 2007
Where it comes from
Steven Taylor and 15 co-authors published the Anxiety Sensitivity Index-3 (ASI-3) in 2007 in Psychological Assessment. Anxiety sensitivity is the fear of arousal-related sensations, based on beliefs that they lead to harm such as death, insanity or social rejection. The older 16-item Anxiety Sensitivity Index wasn't designed to measure separate areas. The authors wanted a measure of the three areas that factor analyses found most often: physical, cognitive and social concerns.
They chose items from a 36-item revised index, keeping those that fit only one of the three areas. The final 18 items include five from the original index. The scale was built on 2,361 nonclinical US and Canadian participants, mostly university students. The authors then tested it in six other samples.
What the research found
The three-factor structure held up in the 2,361-person development sample and in six replication samples. Five were nonclinical groups from the United States and Canada, France, Mexico, the Netherlands and Spain, 4,494 people in all. The sixth was a clinical group of 390. According to the authors, most fit indices showed a good fit. Cronbach's alpha for the three subscales ranged from .73 to .91 across samples. The paper didn't test how stable scores are over time, and the authors listed that as work still to do.
Average totals followed the clinical picture. Across 4,720 US and Canadian participants without a clinical diagnosis, the mean was 12.8 (SD 10.6). Patients tested before treatment averaged higher. Panic disorder (120 people) averaged 32.6 (SD 14.3), social anxiety disorder (38) 31.4, generalized anxiety disorder (30) 27.5 and obsessive-compulsive disorder (102) 26.3.
The authors published no cut-offs. A later study put lines at 17 and 23 on the total score, for moderate-to-high and high anxiety sensitivity. It used 579 adult smokers recruited for a smoking-cessation treatment study (Allan and colleagues, 2014).
How scoring works
| Score | Label | What it means |
|---|---|---|
| 0 to 12 | Low | Close to the average in the original paper: 12.8 (SD 10.6) across 4,720 nonclinical US and Canadian participants. The authors gave no cut-off. |
| 13 to 22 | Moderate | Between that average and the clinical group averages of 26.3 to 32.6. Allan and colleagues (2014) put their moderate-to-high line at 17. |
| 23 to 72 | High | Matches the high cut-score of 23 from Allan and colleagues (2014), a study of 579 adult smokers. It isn't a diagnostic threshold. |
| Each subscale: 0 to 4 | Low | Our page's rough guide for a 24-point subscale. We found no published subscale cut-offs. |
| Each subscale: 5 to 7 | Moderate | Same guide, same caveat. |
| Each subscale: 8 to 24 | High | Same guide, same caveat. |
The ASI-3 authors published no cut-offs. The 23 comes from one later study of smokers. The 13 sits at the published average of 12.8 and isn't a cut-off from any source. The subscale ranges are proportional guides, not published norms.
What it can't tell you
Most of the development data came from university students. The authors also note that the ASI-3 items sat inside a longer questionnaire when they were tested. They couldn't say how the scale behaves on its own. A 2016 study of 1,507 Italian community members found one strong general factor. Its authors concluded that the total score could be more appropriate than the three subscale scores.
A total score can't point to one condition. In the original data the averages for four anxiety disorders ranged from 26.3 to 32.6, with wide spreads (SD 11.9 to 16.8). The test is self-report, and it isn't a diagnosis.
How to use your result
Use the total as your main number and treat the three subscale scores as rough pointers. Retake the test after a few weeks if you're working on how you react to anxiety sensations. See a doctor or therapist if you have panic attacks or if fear of your own sensations is shrinking what you do. A new chest pain, fainting or a racing heart should be checked by a doctor first, whatever the score says. For an emergency, call 911.
Our pages on interoceptive exposure, CBT for panic attacks and physical symptoms of anxiety explain what these sensations are and how people practice with them.
More on this topic: Interoceptive exposure, CBT for panic attacks, Physical symptoms of anxiety, Panic attacks.
Your answers
Your answers are scored in your browser and aren't sent or saved anywhere. Close the page and they're gone.
Sources 5
- Taylor et al., 2007, Psychological Assessment (PubMed): Abstract: 18 items, three factors, development sample of 2,361, replication samples of 4,494 and 390. The alphas, group means and stated limits come from the full text, which we read in a copy hosted by a university lab.
- Wisconsin Addiction Research Center, ASI and ASI-3 scoring page: 0 to 4 coding, no reversed items, six items per subscale and the item numbers. States the lab has no copyright authority and that users should contact the author for permission.
- Allan et al., 2014, Journal of Anxiety Disorders (PMC): Cut-scores of 17 and 23 in 579 adult smokers recruited for a treatment study. One sample, so the numbers may not carry over to other groups.
- Ghisi et al., 2016, Frontiers in Psychology: Italian community sample of 1,507. Found a strong general factor and suggested the total score may be more appropriate than the subscales.
- Smits et al., 2008, Behaviour Research and Therapy (PubMed): Abstract: 24 randomized trials, 1,851 participants; Hedges' g 1.40 in treatment-seeking samples and 0.74 in at-risk samples. Used for the sentence on the result page that CBT lowers anxiety sensitivity.