Measurement tools and target symptoms/skills used to assess treatment response for individuals with autism spectrum disorder.

Bolte et al. (2013) counted every tool used to track change in 195 autism trials. They found 289 different measures. Most tools were used only once.

The team wanted to see if the field agrees on how to judge success. No standard set appeared.

The review shows chaos, not consensus. Each lab picks its own yardstick. Data from different trials cannot line up.

Without shared measures, we cannot tell which program truly helps.

Egli et al. (2002) saw the same mess in preschool studies. Elizabeth widened the lens and proved the problem spans all ages.

Matson (2007) warned that weak measures make ABA look better than it is. Elizabeth gives the numbers behind that fear.

Gitimoghaddam et al. (2022) looked at 770 ABA papers. Their data live inside Elizabeth’s count, so the chaos keeps growing.

Gandhi et al. (2022) zoomed in on one slice: maintenance probes. They found timing and labels are just as scattered, showing the trouble runs deep.

You need to pick tools others also use. If you choose a lone measure, your data will sit on an island. Stick to common sets like the Vineland or ADOS. Share them across teams. Only then can we stack studies and see what really works.