Motor Tests Used in Autism Research Were Not Built for It

What the research found

Five tests account for the whole published literature on standardised motor assessment in autistic children: the Movement Assessment Battery for Children, the Bruininks-Oseretsky Test of Motor Proficiency, the Peabody Developmental Motor Scales, the Test of Gross Motor Development and the Alberta Infant Motor Scale. All five were built for the general child population.

That count comes from a systematic review published in the journal Diagnostics in 2025. Adriana Piccolo, Carmela Settimo and colleagues at the IRCCS Centro Neurolesi “Bonino-Pulejo” and the University of Messina registered a protocol with PROSPERO, searched three databases to the end of December 2024, screened 3,551 records and retained 22 studies. Each retained study used a standardised, validated motor measure on participants aged eighteen or under with a formal diagnosis of autism. All 22 were observational rather than experimental, and 20 used a single instrument and nothing else.

The Movement Assessment Battery for Children appeared in twelve of the twenty-two studies, more than all the others combined: the Bruininks-Oseretsky test in five, the Peabody scales in four, the Test of Gross Motor Development in two, the Alberta Infant Motor Scale in one.

The proportions those studies report are high and consistent. Green and colleagues recorded definite motor impairment in 79 percent of the autistic children assessed, with a further 10 percent borderline. Liu placed 77 percent in the red zone of the movement battery, the band marking severe difficulty. Miller found more than 97 percent scoring below the sixteenth percentile, the cutoff used for the movement criterion in developmental coordination disorder, and Martín-Díaz reported 84 percent below average on the short form of the Bruininks-Oseretsky test. Alsaedi found strength the most affected domain, at a Cohen’s d of 2.66, a separation of more than two and a half standard deviations between groups.

The reviewers combined none of this statistically. They compared the studies narratively rather than pooling them, so no single figure exists for how far autistic motor skill falls below the norm, and the instruments do not reliably agree in any case. Odeh and colleagues gave both the movement battery and the Bruininks-Oseretsky test to the same children and obtained different results on equivalent subtests, the movement battery proving the less sensitive of the two on aiming and catching.

One design addressed the obvious alternative explanation. Staples and Reid matched autistic children against three comparison groups in turn: against peers of the same chronological age the autistic children performed worse, and against peers matched for cognitive ability they remained more impaired than that ability predicted, while matched on motor skill itself they resembled typically developing children roughly half their age. The reviewers read this as a motor difficulty that neither age nor cognitive delay accounts for.

What it means for you and your child

Motor coordination is not among the diagnostic criteria for autism. The manual defines the condition through social communication and restricted, repetitive behaviour, and no motor assessment is required at any point in the evaluation. The reviewers state directly that motor deficits are under-diagnosed for that reason, and are often recorded later, when they are recorded at all, as a separate diagnosis of developmental coordination disorder.

The percentages above come from separate samples recruited by separate methods and cannot be added together. They do share a floor. Even the most conservative of them describes a clear majority of the autistic children assessed, in samples ranging from 24 participants to more than fifteen thousand and across ages running from the first months of life to eighteen years. On this evidence, motor difficulty is a common feature of autism.

The evidence has real weaknesses and the reviewers name them. Judged against a standard tool for diagnostic accuracy studies, 86 percent of the 22 studies carried a high risk of bias in how participants were selected, most having relied on convenience samples. Only five of the twenty-two reported any reliability statistics at all, though where reliability was tested, agreement between raters and across occasions exceeded 0.95 on every instrument checked.

The practical consequence of five borrowed instruments is that a motor score answers a narrower question than its confidence suggests. Each test covers different domains, the review found no accepted gold standard for this population, and two batteries given to the same child can return different verdicts on the same skill. The reviewers also note that assessors modify administration more often for autistic children than for other groups, adding demonstrations or altering task demands, and that these adjustments erode the standardisation the score depends on.

A school or therapy meeting can be asked to produce two specific facts. The first is the name and date of any standardised motor assessment your child has been given, which in a great many files is nothing at all. The second is whether the administration was modified, and how, since the reviewers treat unrecorded adaptations as a threat to the validity of the result. Both questions ask the room to state what was measured, which is a matter of record rather than of judgement about your child.