ABA Therapy Statistics 2026: 40 Numbers, Traced to Their Primary Sources

Here is how an ABA statistic usually gets made. A vendor publishes a blog post with a number in it. A second vendor cites the first. A third cites the second. Six months later the number is “industry consensus,” it is in a conference deck, and nobody on the chain has ever seen the study — because in a fair number of cases there isn’t one.

So we did the unglamorous thing. We took the forty numbers that matter most to an ABA clinic owner in 2026 and chased each one back to whoever actually measured it: the BACB’s own certificant data, the CDC’s ADDM surveillance report, a March 2026 study in JAMA Pediatrics, federal Medicaid spending data, and a peer-reviewed access study in the Journal of Developmental & Behavioral Pediatrics.

Everything below is linked to its primary source, dated, and labeled with what it actually measures. Where a number is a vendor estimate rather than research, we say so. And at the end you will find three statistics this industry repeats constantly that do not survive contact with their sources — including one we had to correct on our own site.

Use any of it. Cite it, screenshot the charts, quote the corrections. That is what it is here for.

The money: Medicaid built this industry, and Medicaid is now squeezing it

Every state Medicaid program now covers ABA. That mandate, layered onto rising diagnosis rates, produced one of the fastest spending curves in American healthcare — and a predictable backlash. The rate cuts of 2025 and 2026 are not random state budget decisions. They are the second half of a story whose first half looks like this.

403%

Growth in Medicaid spending on direct ABA therapy, protocol modification, and family training across all states, 2019–2024. Federal data analyzed by Behavioral Health Business, June 2026.

$347M → $2.2B

Combined Medicaid autism therapy spending across the eight states that publish ABA-level data (Indiana, Minnesota, Colorado, Massachusetts, Nebraska, North Carolina, Wisconsin, Maine) — a 561% increase. Cato Institute analysis.

$21M → $611M

Indiana’s Medicaid ABA spending, 2017 to 2024. A single state, a 29-fold increase in seven years. This is the number that made other state budget offices start reading their own ABA line items.

You can argue about whether that curve represents unmet need finally being served or a payment model being exploited. Both readings have honest advocates. What is not arguable is what state Medicaid agencies did next.

−48%

Nebraska’s cut to technician-delivered ABA, taking the rate from roughly $144/hour to $74.80/hour. The steepest single-state cut on record. Becker’s Behavioral Health.

−25%

New York’s authorized reduction to the technician rate, split into two 12.5% steps — the first on October 1, 2025, the second on April 1, 2026. Providers got a year to watch it coming. Acuity News.

~$68/hr

Indiana’s new flat rate, replacing a 40%-of-billed-charges model, effective January 2026 — paired with a lifetime cap on covered hours. The most structurally aggressive correction in the country: it changes not just the price but the ceiling.

74%

Share of New York Medicaid ABA providers who said they would leave the program if the second 12.5% reduction took effect, in a joint NYSABA/CASP survey. Treat this as a stated intention under advocacy pressure, not a measured outcome — but the direction is corroborated by what actually happened next.

What actually happened next, in the last eight months: three Arizona Medicaid managed care organizations terminated contracts with the state’s two largest ABA providers between November 2025 and March 2026, affecting close to 1,000 children. Autism Learning Partners ended ABA services in Texas on March 21, 2026, citing rates and authorization friction. Stepping Stones Behavioral Solutions shut down entirely in June 2026. And Centria Autism is laying off 48 employees in Fresno, with the reduction complete by the end of this month.

The workforce: bigger than you think, and leaving faster than you think

This is where the sourcing gets bad, and where getting it right matters most — because “there aren’t enough BCBAs” is the single most repeated claim in the field, and the number attached to it is usually years stale.

85,587

BCBAs holding active certification as of July 1, 2026, per the BACB. If you have seen “approximately 48,000 certified BCBAs” cited recently — and it is everywhere — that figure is badly out of date. The real number is nearly double it.

260,174

RBTs holding active certification as of July 1, 2026. Up from 89,122 in 2020: 176% growth in six years.

5,246

BCaBAs holding active certification. This is the field’s only shrinking credential — down from 5,623 in 2021, and down 5.6% in 2025 alone, while BCBAs grew 10.1% and RBTs grew 25.1%. The mid-tier rung of the career ladder is quietly disappearing, which is worth sitting with given that “limited opportunities for professional growth” is a top-three reason RBTs leave.

77–103%

Annual staff turnover by organization size (77.4% small, 89.3% mid-size, 103.3% enterprise). Attribute this one carefully: it comes from CentralReach’s market report, a vendor that sells retention software. It is not peer-reviewed research, and most articles citing it do not tell you that.

1 year

Median tenure for a behavior technician, with 65% average direct-care turnover, per the BHCOE compensation and turnover report. Also a vendor source (BHCOE sells accreditation), also paywalled, and the underlying data is from 2021.

So the two numbers everyone quotes about ABA turnover both come from companies selling solutions to turnover. That does not make them wrong. It does mean you should know it. Which brings us to the best dataset in this entire article, and the one almost nobody has written about.

In early 2025 the BACB surveyed 30,018 people whose RBT certification had expired during 2024 without their holding any other BACB credential. 1,386 responded — a 4.6% response rate, which is thin, so weight it accordingly. But this is the field’s only major workforce dataset with no commercial interest behind it. The BACB is not selling you anything.

57%

Cited inadequate pay as influencing their decision to leave — the top answer, and the one everyone expects. BACB December 2025 newsletter.

42% / 41%

Cited concerning issues in the workplace, and limited opportunities for professional growth. Neither is a pay problem. Both are a management problem, and both are cheaper to fix than a raise.

34%

Cited unpredictable pay — a separate answer from inadequate pay. Cancelled sessions and unbilled hours are their own retention issue, and one that gets worse as reimbursement tightens.

58%

Said “I like the work, but life took me in a different direction.” The single most common sentiment among people who left was not that the job was bad. It is that the job could not hold them.

~30 / 83%

Median age and share female among RBTs whose certification lapsed. Any retention strategy that ignores what happens to a 30-year-old woman’s career and caregiving demands is not a retention strategy.

On compensation, one caveat worth more than any number: the BLS does not track BCBAs or RBTs as occupations. Every “BLS says BCBAs earn X” claim is someone mapping to a proxy code. The honest version is that the closest BLS category (substance abuse, behavioral disorder, and mental health counselors) had a median wage of $59,190 in May 2024 and is projected to grow 17% through 2034 — but that category is not BCBAs, and you should not present it as though it were.

Private equity: 574 sites, 147 deals, and one number that gets misread

Until March 2026, almost everything written about PE in autism services traced back to a single 2023 advocacy report. Then researchers published an actual peer-reviewed count in JAMA Pediatrics. It is now the defensible source, and it is the one to cite.

574

Autism service delivery sites acquired by private equity across 42 states, via 147 separate deals, 2015–2024. Arnold DR et al., JAMA Pediatrics, March 2026.

79.6%

Share of those acquisitions that happened in a five-year window, 2018–2022. The roll-up was not a decade-long trend; it was a stampede, and it is largely over.

24%

Increased likelihood of PE entry in areas with top-tercile autism prevalence (p<0.01). PE bought where the demand was — which is either rational capital allocation or cherry-picking, depending on your priors.

85%

Share of all autism-segment M&A conducted by PE firms, 2017–2022 — which the authors call a rate not found in any other segment of healthcare or any other industry. Appelbaum & Batt, CEPR, 2023.

$1.65B

Roper Technologies’ net purchase price for CentralReach, announced March 24, 2025. Roper’s own release puts CentralReach at roughly $175M revenue and $75M EBITDA for the twelve months ending June 30, 2026, serving 200,000+ professionals.

That last one is routinely cited as evidence that private equity keeps buying ABA. It is the opposite. Roper is a strategic public-market acquirer (NYSE: ROP), and it bought CentralReach out of PE ownership from Insight Partners. The $1.65B is also a net figure that includes roughly $200M in tax benefits. If you use this deal in a PE argument, you are using it backwards.

One more distinction worth holding onto, because it is where most coverage of this topic falls apart. The CEPR report’s claims about post-buyout chains cutting staffing, training, and supervision are the authors’ analysis, not measured outcomes. The JAMA researchers, who did the counting, explicitly decline to make quality claims — they say it is unclear whether PE entry increased or decreased access. Those two things get blurred together constantly. They are not the same kind of evidence.

Prevalence and access: the demand is real, the follow-through is not

1 in 31

ASD prevalence among 8-year-olds across 16 ADDM sites (3.2%; 32.2 per 1,000), up from 1 in 36. This is the 2022 surveillance year, published in 2025: CDC MMWR Surveill Summ 2025;74(SS-2). There is no 2026 CDC prevalence figure. Anyone citing one is making it up.

47 months

Median age of earliest known ASD diagnosis in the ADDM network. Just shy of four years old — which is the single most important number in this article for anyone building an intake funnel.

1 in 103 → 1 in 19

The range across ADDM sites, from Laredo, Texas to California. A five-fold spread. Prevalence is not a national constant, and neither is your addressable market.

13%

Referred children who never initiated ABA at all, despite mandated insurance coverage (44 of 334). Choi KR et al., JDBP, 2022.

66% → 46%

Share of referred children still receiving ABA at 12 months, then at 24 months. Over half the referral cohort is gone within two years.

28%

Of the children who made it to 24 months, the share receiving a full ABA dose (80%+ of recommended hours). Caveat this one: single integrated health system, referrals from 2016–2018. It is dated. It is also the best peer-reviewed access data that exists, which tells you something about the state of research here.

Put the access numbers next to the prevalence numbers and the shape of the problem is clear. Diagnosis is happening at 47 months. Coverage is mandated in all 50 states. And 13% of referred families never start, while the majority who do start are gone inside two years. The demand is not the bottleneck.

Three statistics this industry keeps repeating that are wrong

These are not obscure errors. All three are on the first page of Google, in vendor blogs, in conference decks, and — in one case — they were on this website until we wrote this article.

Two honorable mentions. First, “approximately 48,000 certified BCBAs nationwide” is stale by years — the BACB says 85,587 as of July 1, 2026. Every “BCBA shortage” ratio built on the 48,000 figure overstates the gap by roughly half. We had that number in a chart on this site, sourced from a secondary write-up, and it was wrong. It is fixed now.

Second, ABA market size figures are not worth citing at all. We found estimates of $4.06B, $7.97B, $10.7B, $15.0B, and $18.9B for roughly the same period, with CAGRs from 4.56% to 12%. They disagree because they define “the market” differently and most come from firms whose product is the report. Pick one and state its definition, or leave it out. Presenting a range implies a precision that does not exist.

What all of this means if you own a clinic

Read the numbers together and the picture is not the one the industry tells itself. The story is not “there is a shortage of BCBAs.” The workforce grew 10.1% last year and stands at an all-time high. The story is that reimbursement is falling in the states with the most volume, the people delivering care leave inside a year for reasons that are mostly not about pay, families who are already referred and already covered evaporate before the second year, and the capital that funded the last decade of expansion has stopped buying.

Every one of those is a squeeze on the same thing: the number of families you can reach, convert, and keep. When the rate per hour falls and you cannot control the rate, the only lever left is volume and retention on the front end. That is not a clinical problem. It is a demand problem, and it is the part of this you can actually control — which is why we wrote ABA Lead Generation in 2026 about owning search demand instead of renting referrals.

Citing this page

Every statistic above links to its primary source, and we would rather you cite the primary source than us. Where this page is genuinely useful is the corrections — if you are writing about ABA workforce or access and you want to avoid repeating the three errors above, this is currently the only page we know of that documents them with the underlying sources side by side.

Suggested citation: ABA Clinic Marketing, “ABA Therapy Statistics 2026: 40 Numbers, Traced to Their Primary Sources,” July 16, 2026, abaclinicmarketing.com/blog/aba-therapy-statistics-2026. Charts are free to reuse with attribution and a link. If you find an error in here, tell us and we will fix it and say so — which, given the subject of this article, is the least we can do.