The functions of assessment: implications for selection and development of devices for assessing repertoires in clinical, educational, and other settings.

Hawkins (1979) wrote a think-piece, not an experiment. The paper lists four jobs an assessment can do. It can screen many kids fast. It can pinpoint one skill to teach. It can watch progress over time. It can give data for policy.

The author warns against buying a shiny tool first. He says to pick the job, then pick the tool. The paper shows a funnel. Wide tools at the top, sharp tools at the bottom.

There is no new data. The paper gives a map. The map helps you match the tool to the question. If you need to screen, use a short checklist. If you need to teach, use a pinpoint probe.

Al-Hendawi et al. (2025) looked at dozens of preschool studies. They found the same funnel. CLASS works for wide screening. Custom DOS works for pinpoint teaching. The 1979 map still fits.

Lepper et al. (2023) moved the idea into numbers. They used BCBA pass rates to pick program parts. They turned the 1979 advice into a data rule.

Petursdottir et al. (2018) added a validity checklist. They took the 1979 idea and gave single-case researchers a guard-rail. The guard-rail keeps the tool matched to the job.

Next time you open a new assessment box, stop. Write the job on top of the page. Then open the manual. If the job is “find who needs help,” use a short screener. If the job is “teach one target,” use a pinpoint probe. This tiny pause saves hours of bad data.