CATQ Subscale Profile Analysis: Comparing Uneven and Repeated Scores
By Silas Rowland · Updated November 3, 2026
A CATQ subscale profile shows how your reported camouflaging differs across three dimensions. Before interpreting an uneven pattern, check what the numbers represent: a larger raw score, a higher item average, and a percentile are different claims. A later change in the pattern raises another question: did one score rise, another fall, or both move?
If you have completed the CAT-Q self-reflection questionnaire, save the three results with their date and the version you used. You can then compare the actual numbers without relying on a profile label or the relative length of three bars.
The CAT-Q measures self-reported camouflaging, including among non-autistic adults. Its original development study did not establish diagnostic cutoffs or clinical types based on combinations of subscale scores. Use results for reflection, with professional support when difficulties affect daily life. [1]

Put the scores on a comparable numerical basis
For the three definitions, item counts, scoring key, and the overlap that makes Assimilation especially difficult to interpret in isolation, see the guide to Compensation, Masking, and Assimilation.
Consider a fictional result of Compensation 45 and Masking 40. Dividing each correctly scored sum by its item count gives 45 ÷ 9 = 5 and 40 ÷ 8 = 5. The raw difference is five points, but the item averages are identical. The larger sum alone does not establish greater endorsement per item. These numbers illustrate arithmetic; they are not thresholds or targets. [1]
Expressing those sums as percentages of their maxima also produces a tie: 45 ÷ 63 and 40 ÷ 56 are both about 71.4%. For these subscales, dividing by the maximum is equivalent to dividing the item average by seven. Changing the display from an average to a percentage does not add information about how you compare with other people. [1, 5]
That is the difference between a percentage of the maximum and a percentile rank. The former uses the scale's possible maximum as its denominator. A percentile requires a comparison distribution from a specified group. A display of 71.4% therefore does not mean that you scored above 71.4% of respondents. The site's basic results show subscale scores and percentages, without a total-score classification. [5]
An item average solves the unequal-item-count problem in the example. It does not make the three constructs interchangeable. A one-point average difference is still a difference between responses to different sets of questions, rather than a direct measure of extra effort, harm, or need for support. The original validation does not supply a clinical threshold for the gap between two subscales. [1]

Audit the claim attached to your profile
Use this table when a results page or explanation makes a claim about your numbers. Work from the claim you want to check, rather than choosing a row based on which dimension is highest.
| Claim you encounter | What to check | What the check can establish |
|---|---|---|
| “This is your highest score.” | Whether the comparison uses raw sums or item averages | Whether unequal item counts account for the ranking |
| “You are above most people.” | Whether a named comparison group and percentile method are supplied | Whether there is a population comparison, rather than only a percentage of the maximum |
| “The gap is significant.” | What evidence and definition of significance support the gap | Whether the phrase means more than a visible numerical difference |
| “Your pattern changed.” | Each subscale at each date, using the same scoring basis | Which individual scores moved, rather than only which became highest |
| “This pattern predicts burnout.” | Whether the evidence tests that outcome, that combination, and future change | Whether the cited research actually addresses the prediction being made |
For example, the fictional 45-versus-40 result supports “the raw sums differ.” After conversion, it also supports “both item averages are five.” It does not support “Compensation is significantly dominant.” Each sentence makes a different claim, even though all refer to the same two numbers.
If an explanation uses “significant,” ask whether it means statistically supported, clinically important, or simply noticeable. Do not let the word substitute for a method. Likewise, a label such as “high Compensation, low Masking” needs a stated basis for “high” and “low”; the label is not supplied by the arithmetic itself.
You can stop the audit with a narrower, accurate statement. “My Compensation sum is larger, but the averages are equal” is already a useful result. You do not need to turn it into a named profile. If a report adds a percentile, keep the reference group beside it so that a later comparison does not silently switch populations.
Separate association, future change, and personal prediction
Research about camouflaging and mental health asks several different questions. Distinguishing them helps you decide whether a finding applies to the claim attached to your profile.
Do people reporting more camouflaging also report more difficulties at the same time? Hull and colleagues' 2021 study found associations between higher total camouflaging scores and anxiety and depression in autistic adults. This was a cross-sectional analysis. It could describe a relationship at one time, but could not establish which came first. It also examined the total score, rather than validating combinations of the three subscales as risk profiles. [2]
Does an earlier score predict later change in difficulties? Van der Putten and colleagues studied camouflaging and mental health longitudinally. They did not find strong evidence that camouflaging predicted the development of mental health difficulties. Initial Assimilation did not predict subsequent change; exploratory analyses of initial Masking and Compensation also found no meaningful predictive effects on change. The study had two measurement occasions, which the authors noted were insufficient for strong causal conclusions. [3]
These findings address different questions. An association with current distress can coexist with a lack of strong evidence for predicting later change. Neither finding permits the conclusion that a specific person's subscale pattern caused their distress. The longitudinal result also does not establish that camouflaging is harmless. [2, 3]
Can your particular combination forecast what will happen to you? That is a further claim. Neither the total-score association nor the longitudinal findings validate a prediction rule for your three-score combination. A statement such as “this profile means burnout is approaching” would need evidence for that specific prediction. A familiar-sounding explanation is not that evidence. [2, 3]
Prediction of outward social performance is another separate issue. Cook and colleagues' systematic review distinguishes self-reported camouflaging effort from measures comparing internal characteristics with outward presentation. Your profile is not a direct reading of how successfully you concealed something or what other people noticed. [4]
When discussing a difficult period, keep the result and the difficulty side by side: “These were my scores, and I was feeling exhausted.” That preserves both observations. Saying “the highest subscale explains the exhaustion” adds a causal conclusion. You can ask for help with the exhaustion while leaving that explanation open.
Compare repeated profiles one subscale at a time
A profile can look more uneven even when its highest score has not increased. In a fictional example, Compensation stays at 45 while Masking changes from 40 to 32. Their item averages move from 5 versus 5 to 5 versus 4. Compensation now looks relatively higher because Masking fell, not because Compensation rose.
The reverse problem is also possible. Suppose the sums change from 45 and 40 to 54 and 48. Both item averages rise from five to six, while the relationship between them remains equal. The relative pattern looks unchanged, but the reported levels are different. Neither example identifies a cause or a desirable direction of change; they show why comparing only the highest bar loses information. [1]
Before interpreting two results, check their comparability:
- Keep the questionnaire version and language with each result. Do not assume an adapted quiz or a different translation is interchangeable with the version used earlier. Findings from the original adult validation should not be assumed to apply unchanged to every translation or age group. [1, 4]
- Keep the scoring basis consistent. Compare a sum with the same subscale's sum, or an item average with its counterpart. Keep percentages distinct from percentiles, and retain any named reference group used for percentile results.
- Record the date and any difference in how you understood the questions. If an item now means something different to you, note that beside the changed answer. It leaves an important interpretation question visible rather than assigning the entire difference to changed behavior.
- Describe each subscale's movement before interpreting the profile. “Compensation unchanged; Masking lower” says more than “I became more of a compensator.”
The original development paper reported preliminary test–retest evidence. That supports studying the measure's consistency across repeated administrations; it does not make every individual score movement meaningful or provide a threshold for meaningful change in your subscale gap. [1] Do not use a research finding about repeated scores as a rule that every personal difference is either real progress or measurement noise.
A useful comparison note might read: “Same version and language. Compensation stayed at 45; Masking changed from 40 to 32. The larger gap came from the Masking result. I also understood one question differently this time. I want to discuss that change before interpreting the pattern.” This fictional note separates calculation, comparability, and interpretation without assigning a clinical meaning.

References
- Hull L, Mandy W, Lai M-C, et al. 2019. Development and Validation of the Camouflaging Autistic Traits Questionnaire (CAT-Q). Journal of Autism and Developmental Disorders. doi: 10.1007/s10803-018-3792-6.
- Hull L, et al. 2021. Is social camouflaging associated with anxiety and depression in autistic adults? Molecular Autism. doi: 10.1186/s13229-021-00421-1.
- van der Putten et al. 2025. Camouflaging in autism: A cause or a consequence of mental health difficulties? Autism. doi: 10.1177/13623613251347104.
- Cook J, Hull L, Crane L, Mandy W. 2021. Camouflaging in autism: A systematic review. Clinical Psychology Review. doi: 10.1016/j.cpr.2021.102080.
- Cat-q. n.d. Questionnaire, scoring implementation, and basic results. Cat-q.
Finish with a comparison you can explain
A useful CATQ subscale profile analysis ends with a statement you can trace back to the results: which units you compared, which scores differed, and what remains a question. Save that statement with the dated results rather than relying on an identity label generated from them.
For wider questions about numerical results, review the CAT-Q scoring and cutoff guide. You can bring your comparison note to a qualified professional alongside the difficulties you want help with. A lower score or a more even profile need not be a prerequisite for seeking support.
If distress interferes with daily life, speak with a qualified health professional. If you face immediate danger or cannot stay safe, contact local emergency services or a crisis service for immediate support.
Frequently Asked Questions About CATQ Subscale Interpretation
What does it mean if my compensation score is significantly higher than my masking score?
Check whether “higher” refers to raw sums or item averages, then ask what supports the word “significantly.” In the fictional 45-versus-40 example, the item averages are equal. Even a remaining average difference does not establish a clinical type or identify the cause of a difficulty. [1]
Can my subscale profile change over time?
Repeated results can differ. Compare each subscale separately using the same scoring basis, and keep the versions, languages, dates, and any changed interpretation of questions. A changed gap may reflect a change in either score. Preliminary test–retest evidence does not establish the meaning of every individual movement. [1]
How accurate are these subscale profiles in predicting my social challenges?
Evidence supporting the measurement of self-reported camouflaging does not establish the accuracy of forecasts attached to a particular score combination. Current mental-health associations, later mental-health changes, and outward social performance are different outcomes. The cited studies do not validate a personal forecast from your three-score pattern. [1, 2, 3, 4]