An examination of the state of imitation research in children with autism: Issues of definition and methodology.

Melina and colleagues read every autism imitation paper they could find. They saw teams using different words for the same action. One study called it "imitation" when a child copied a hand flap. Another needed five different moves before it counted.

The reviewers grouped the papers by how they defined and measured imitation. They also noted who was tested, how old they were, and what tasks were used.

Half the field says kids with autism can't imitate. The other half says they can. The review shows the fight is mostly about wording and test design.

No shared checklist means every lab makes its own rules. Results clash, so progress stalls.

Shaked et al. (2004) saw the same mess in matching subjects. They warned that if age and IQ aren't reported the same way, outcomes can't be compared. Melina pushes the same fix, but for the imitation task itself.

Lord et al. (2005) asked for standard outcome measures in autism intervention trials. Melina narrows the plea to one core skill: imitation.

Jonsson et al. (2016) later showed that even good RCTs skip external-validity details. Together the three papers form a ladder: first match subjects, then define the skill, then prove it works in real life.

Rutter (2013) echoed the call three years later, saying the whole autism field needs new concepts and cleaner methods.

If you test imitation, pick one definition and stick to it. Write the exact prompt, response window, and scoring rule in your report. Share the task script so the next BCBA can repeat it. Over time we will finally know if our kids truly struggle with imitation or just with the way we ask.