On mosaics and melting pots: conceptual considerations of comparison and matching strategies.

Burack et al. (2004) wrote a think-piece, not an experiment.

They looked at how autism studies pick control or comparison groups.

The authors say every matching choice must fit the exact question you are asking.

There is no single “right” comparison group for autism research.

Matching kids with autism to typical peers, to other diagnoses, or to siblings can all be wrong or right.

You must tell readers why you chose your match and what limits that choice brings.

Elsmore et al. (1994) set the stage. They told clinicians to treat each JEAB study as a sample and each client problem as a comparison—an early call for study-specific matching.

Katz et al. (2003) gives a real example. They compared kids from families with one autistic child to kids from families with many. No behavior difference showed up, proving that a matching choice can lead to a null finding.

Brown et al. (2011) supply the numbers side. They list nine ways to size up single-case data. Burack et al. (2004) warn that even the best effect size is weak if the comparison group makes no sense.

Next time you read an autism study, do not just glance at p values. Ask: “Why these groups?” If you run a study, write a short paragraph that starts with: “We matched X to Y because…” Your supervisor and future readers will thank you.