Blogs

Two Denominators, One Question: Reading Autism Rates by State

State autism rates are published against two different denominators. Plainsheet walks through what each one measures, why the state spread reaches 3.7x, and which comparisons are invalid.

Plainsheet is a research blog published by Special Needs Care Network.

State-level autism rates circulate as a single ranked list. The underlying data does not support a single list, because two federal collections measure two different populations against two different denominators.

The measurement problem

The CDC Autism and Developmental Disabilities Monitoring Network estimates clinical prevalence. Surveillance year 2022, published 2025, covers children aged 8 across 16 sites and reports 32.2 per 1,000, or 1 in 31. An ADDM site covers a defined community, often part of a single metro or county. An ADDM site figure is not a state figure.

The U.S. Department of Education IDEA Section 618 child count is an administrative census. It counts children ages 5 to 21 served in public schools under the autism eligibility category, school year 2023-24. Nationally that is about 1 in 72.

PropertyCDC ADDMIDEA child count
TypeSurveillance estimateAdministrative census
PopulationChildren aged 8Ages 5 to 21
Geography16 defined communitiesAll states and territories
DenominatorCommunity child populationResident population ages 5 to 21
What a high value meansHigher detected prevalenceHigher formal identification

Comparing a state's IDEA rate against another state's ADDM site rate is invalid. The two are not the same instrument.

Why the state spread reaches 3.7x

Within the IDEA count alone, Massachusetts identifies about 1 in 51 and Montana about 1 in 188. The ratio between the extremes is roughly 3.7 to 1.

Biological prevalence does not vary that way across state lines. The variance is administrative. Four inputs drive it:

  1. Evaluation capacity, meaning how many qualified evaluators a state has per child.
  2. Referral practice, meaning how readily schools and pediatricians initiate an evaluation.
  3. Eligibility interpretation, meaning how a state applies the federal autism category against other categories.
  4. Reporting convention. Iowa and New Mexico report special education non-categorically and publish no autism-specific count, so both are excluded from any ranking rather than scored as zero.

A state that identifies fewer children is a state where more children are unidentified. Reading a low rate as a favorable outcome inverts the finding.

The valid comparisons

  • IDEA rate against IDEA rate, across states, for the same school year.
  • A single state's IDEA rate against the national clinical figure, to size the identification gap.
  • A state's IDEA rate over time, to detect a change in identification practice.

The dataset

Special Needs Care Network published the full state table with both figures side by side at Autism Rates by State, including a downloadable CSV of all 51 rows. Sources named on the page: CDC ADDM 2022, IDEA Section 618 for 2023-24, and U.S. Census population estimates for 2023.

Special Needs Care Network operates a directory of special needs schools and therapy providers in the United States at specialneedsusa.com. Further reports are collected at specialneedsusa.com/data.

Figures verified August 2026.