Testing Math or Testing Language? The Construct Validity of the KeyMath-Revised for Children With Intellectual Disability and Language Difficulties.

Alaimo et al. (2015) ran a confirmatory factor analysis on the KeyMath-Revised. They wanted to know if the test truly measures only math skills in children with mild intellectual disability.

The sample was small. All kids had both ID and language difficulties. The team checked how math items and language items grouped together.

The math and language items loaded on the same factor. This means the test scores language ability as much as math ability for these children.

Low scores may look like a math problem when they are really a language problem.

Bouck et al. (2016) extend this idea. They show that working memory and flexibility also shape math scores in ID. So the KM-R is not just tangled with language; it is tangled with executive skills too.

Matson et al. (2004) conceptually replicate the warning. They found that false-belief tasks fail in ID when narrative language is weak. Same message: test scores can reflect language demands, not the skill you think you are testing.

Bae et al. (2015) seem to disagree at first glance. They found language confounds in math word problems for kids with autism, not ID. The contradiction is only surface-level. Both papers agree that if a child struggles with sentence comprehension, math scores will drop. The difference is the diagnosis, not the principle.

Before you write a math goal, check the student’s language scores. If expressive or receptive language is low, use a test with fewer verbal directions or allow AAC. Better yet, run a brief language probe alongside any standardized math test. This keeps you from targeting the wrong deficit and wasting instructional time.