The Autism Test That Doesn't Travel

Here's a question worth asking about any diagnostic tool: does it work the same way everywhere, or does it quietly measure culture instead of the thing it claims to measure?

A cross-cultural study of a widely used autism screening tool — the Autism Spectrum Quotient, tested across India, Japan, and the UK — found that individual screening items performed completely differently depending on where they were used. One item was a strong predictor of autism in Japan and a poor one in the UK and India. Another worked well in the UK and poorly everywhere else. Items asking whether a child "enjoys social occasions" or understands "the point of a joke" didn't predict a diagnosis accurately in India the way they did elsewhere.

Eye contact is the clearest example of what's going on underneath these numbers. Standard screening treats reduced eye contact as a warning sign. But for many Chinese children, avoiding eye contact with an adult is a mark of respect, not disengagement. In Ethiopia, formal social greeting behavior — something a Western-trained clinician might flag as scripted or ritualized — is a valued social skill, and Ethiopian parents themselves are, by their own report, far less likely to have been alarmed by social interaction differences than by other behaviors entirely.

None of this means autism itself varies by culture. Researchers are consistent that the underlying condition doesn't change by geography — what changes is whether the tools built to find it can actually see past the cultural assumptions baked into their own questions. Lower diagnosis rates in a given region reflect a failure to identify, not a lower true rate.

This matters beyond the international comparison. If a screening instrument can mistake cultural norm for clinical sign at the level of an entire country, it's worth asking what else it might be missing closer to home — in immigrant families, in communities whose norms don't match the assumptions embedded in the tool a clinician was trained on. The instrument isn't just measuring the child. It's measuring how well the child matches a specific, unstated cultural baseline. That's a real problem for anyone whose job depends on getting the answer right.