Fred from the Fixed Interval has a section called accidentally behavior analysis. This story falls into that category. Here is a restaurant critic who delivers plenty of reinforcers and creates a wealth of positive unintended consequences.

I’m talking about Ertan Bek, a 34-year-old New York cabdriver of Turkish heritage. He posts as @NewYorkTurk on Instagram and TikTok, and as AmericanTurk on YouTube. He has become extraordinarily popular: The New York Times reported in April that he had amassed about two million followers on Instagram and TikTok in less than two years. His YouTube channel now has more than one million subscribers.

His point-of-view presentation is quite literal. He wears Meta Ray-Ban glasses that record what he sees and uses his phone for the shots in which we see him. Because restaurant employees increasingly recognize him, he sometimes arrives in a deliberately unconvincing wig and beard.

Bek originally made videos with friends at halal food carts. He then began visiting Michelin starred restaurants, using the hook: “Let’s see how expensive a solo dinner is at…” He shows the entire experience: the welcome, menu, food, portions, service, bill and tip. He isn’t a trained restaurant critic, and now jokingly calls himself an “ignorant gourmand.” His strength is that he approaches fine dining as a regular working New Yorker. He asks questions that an ordinary diner might hesitate to ask and makes an intimidating restaurant seem accessible.

The struggling restaurant videos developed later. Viewers or family members write to him about a quiet family business, and he generally visits unannounced. He pays for his food, refuses complimentary meals when he can, tips generously and does not charge the restaurant for the video. Then he returns several days later to see what happened. Sometimes on his return visit he even helps serve customers or wash dishes when the restaurant is overwhelmed.

A well documented example is Plaka, a nearly empty Greek family restaurant in Queens operated by a couple in their seventies. Bek ordered Greek salad, spanakopita, moussaka and baklava, paid his $58.60 bill and left a $100 tip. Three days after he posted the video, he returned to find the restaurant packed. The owners said they had been overwhelmed for three consecutive days and had temporarily stopped delivery orders because they were running out of food. They appeared with him later on the Today show. The Plaka story is described here.

One reason the chefs and owners seem genuinely moved is that he pays attention to them. He remembers names, praises the work involved, shows the cooks and servers, and sends customers to businesses that cannot afford conventional promotion. More importantly, after a positive review customers come in droves. As one owner said, “my restaurant was empty 3 days in a row. After your review I made the rent in one night. You’ve changed my life”.

He does earn money from the platforms, so when he tells an owner “Nobody pays me,” he means that the restaurant did not pay him. According to a New York Times report syndicated here, he spends more than $3,000 per month producing the videos. He reported receiving $4,024 from TikTok for one month with 10 million views, but only $2,521 for a month with 32 million views, illustrating how unpredictable the income is.

There is some showmanship and careful editing. A quiet dining room at one moment does not necessarily prove that a restaurant is failing, and the enormous initial rush following a viral video does not guarantee lasting survival. Nevertheless, the Plaka turnaround and several similar follow-up videos demonstrate a very real short-term effect. Bek seems to have found an unusually decent use for his influencer power: demystifying expensive restaurants while giving excellent, overlooked neighborhood restaurants a chance to be discovered. You will enjoy him, the owners’ reactions are often remarkably touching. Check out the video below.

Paul Coyne PhD

Share

Leave a comment

Subscribe now

I recently read this passage from Claude Goldenberg, professor emeritus at Stanford, in a footnote on his Substack:

“I’m no fan of the term ‘science of reading,’ since it’s been mis-used to signify different things to different people, leading to counterproductive and misinformed discourse about science, what counts as science, and assorted other dead ends. I prefer to think in terms of literacy research comprehensively understood. But I also realize this does not save us of from [sic] rhetorical and ideological bogs.”

Goldenberg may have a point. The term can be stretched. Publishers want to sell books. So I wondered, if there was an agreed-upon definition of the science of reading? What should a curriculum include, or exclude, to meet that definition? And do popular curricula meet the standard? Here is what I found.

A Definition

In 2022, The Reading League, a national organization founded in 2016 to promote evidence-based reading instruction, published Science of Reading: A Defining Guide. The guide defines the term this way:

“The science of reading is a vast, interdisciplinary body of scientifically-based research about reading and issues related to reading and writing. This research has been conducted over the last five decades across the world, and it is derived from thousands of studies conducted in multiple languages. The science of reading has culminated in a preponderance of evidence to inform how proficient reading and writing develop; why some have difficulty; and how we can most effectively assess and teach and, therefore, improve student outcomes through prevention of and intervention for reading difficulties.” (p. 6)

The guide was developed by The Reading League with its contents created by the Defining Movement Coalition, a group of researchers and practitioners who met weekly from September 2020 to July 2021. The coalition included Louisa Moats, Maria Murray, Stephanie Stollar, Mary Dahlgren, Linda Diamond, and Margie Gillis, among others. It was designed as a consensus document within the coalition, drawing on several disciplines and professional perspectives rather than expressing the view of a single author. It may be the most influential current definition, although it has not eliminated disagreements about the term or its instructional implications.

I found no coalition member prominently identified with behavior analysis. The Reading League has, however, featured scholars whose work is compatible with behavioral instruction, including Amanda VanDerHeyden, Matthew Burns, Anita Archer, and Charles Hughes. Their work emphasizes explicit instruction, observable student responding, frequent assessment, corrective feedback, treatment integrity, and data-based instructional decisions.

What the Curriculum Contains

Following the Simple View of Reading and Scarborough’s Reading Rope, the guide organizes skilled reading around two broad domains: word recognition and language comprehension.

Word recognition. Students receive explicit instruction in phonological and phonemic awareness, the ability to notice, think about, and produce the individual sounds in spoken words. They also receive systematic phonics and spelling instruction that teaches letter-sound correspondences directly, shows students how to blend sounds for reading, and shows them how to segment sounds for spelling.

Language comprehension. The guide calls for vocabulary instruction, both expressive and receptive, along with instruction in morphology (base words, roots, prefixes, and suffixes), syntax, and semantics. Because comprehension depends heavily on what a reader already knows, a curriculum that neglects history, science, literature, and the arts is neglecting comprehension.

Consider the sentence, “John, José, and Maria went on a picnic.” A child may decode every word but still understand little without knowing what people do at a picnic. Background knowledge gives the sentence meaning.

How It Is Taught

The guide is equally specific about instruction. Here its requirements align closely with the principles of behavioral science and Direct Instruction.

Instruction is explicit. The teacher directly explains each concept. The guide rejects the premise that students will discover essential word reading skills simply through exposure to literature.

Instruction follows a deliberate scope and sequence that moves from simpler concepts to more complex. Each new skill builds on skills already mastered. Ongoing assessment data identify when a student has not yet mastered a skill, and targeted intervention follows to remediate.

This is where the guide comes closest to the standard behavior analysts have long applied to teaching. R. Douglas Greer, whose Comprehensive Application of Behavior Analysis to Schooling model measures every teaching interaction, holds that instruction has occurred only when the learner's behavior has measurably changed.

What Is Absent

The Reading League followed the Defining Guide with a second document, the Curriculum Evaluation Guidelines (2023), which its reviewers use to audit curricula. For administrators conducting curriculum reviews, one of the most useful features of the Guidelines is its list of red flags. Red flags identify practices that conflict with the scientific evidence and may require materials to be rejected, supplemented, or substantially modified.

The first red flag is the three-cueing system, known as MSV for meaning, structure, and visual cues. Teaching beginning students to identify unfamiliar words by looking at pictures, using context, or guessing from sentence structure, rather than decoding the letters, is a red flag.

The second red flag is the use of predictable texts for early readers. Books built around repetitive sentence patterns and heavy picture support are likely to train students to guess, e.g., I see a hat, I see a dog, I see a horse. The Guidelines call for decodable texts, books whose vocabulary matches the phonics patterns students have been taught.

The third red flag is relying on incidental phonics. That is, using brief lessons prompted by particular words or student errors instead of following a planned phonics sequence.

The Reading League’s Guidelines are consistent with the work of Direct Instruction researchers such as Douglas Carnine and Russell Gersten, particularly in their emphasis on explicit and systematic instruction, carefully sequenced skills, frequent student responding, corrective feedback, assessment, and data-based intervention. I found no evidence, however, that either researcher formally endorsed the Guidelines or The Reading League’s definition of the science of reading.

The Reviewers

For years, EdReports drew criticism for awarding favorable ratings to balanced-literacy programs that lacked adequate foundational-skills instruction.

Emily Hanford’s 2022 podcast Sold a Story brought national attention to three-cueing, predictable texts, and the influence of balanced-literacy publishers. EdReports later revised its review methods to give greater weight to structured foundational-skills instruction and to identify nonaligned practices more clearly.

EdReports introduced revised K–2 foundational-skills review criteria in June 2024. In July 2025, it launched its broader version 2.0 review system. The new English language criteria include indicators that instructional materials must meet before they advance through the full review. They also flag practices such as teaching three-cueing as a word-reading strategy.

This history creates a problem when people cite an EdReports rating. A favorable review issued under an older version does not necessarily show that the program would pass the current review, and EdReports does not retroactively update completed reports when its methods change. The date and version matter. Consider Wilson Fundations: under EdReports’ original 2019 review, it only partially met the criteria. Under the current evaluation, it meets expectations. Same publisher, different review.

The States Weigh In

The push for evidence-based literacy is now legislated. According to Education Week’s tracker, forty-two states and the District of Columbia have passed laws or adopted policies requiring evidence-based reading instruction since 2013, as of this writing.

A growing number of states, including Ohio, Indiana, Texas, Wisconsin, and North Carolina, have prohibited three-cueing in curriculum, instruction, teacher preparation, or some combination of these areas.

California, where I live, tells a more complicated story. The state that popularized whole language in the 1980s passed a science of reading law, AB 1454, in October 2025. The law revises teacher preparation and literacy training requirements. It funds teacher training in evidence-based reading instruction and directs the State Board of Education to adopt an approved materials list for grades one through eight. It contains no ban on three-cueing, and districts may bypass the approved list by certifying that their chosen materials meet the criteria.

The publishing industry responded to all this pressure in two ways. The earnest revisers replaced predictable readers with decodable texts and rewrote their scope-and-sequence documents so that phonics was taught systematically. Others added a phonics patch onto their existing workshop models, kept the predictable texts and the discovery learning, and then rebranded the package as “aligned to the science of reading.” The EdReports and Reading League reviews exist to help tell these two groups apart.

Science of Reading Curricula

At the time of this writing under EdReports current review system, I found three K–2 foundational-skills programs that received “Meets Expectations” ratings:

• Really Great Reading,

• Wilson Fundations, and

• IMSE Orton-Gillingham Plus.

All three provide explicit and systematic instruction in phonemic awareness, phonics, word recognition, spelling, and fluency. According to EdReports, none constitutes a complete English language arts curriculum; schools must pair them with strong instruction in oral language, vocabulary, knowledge, literature, comprehension, and writing.

The Reading League evaluates curricula against criteria similar to EdReports’, publishing its findings as Curriculum Navigation Reports. Its Curriculum Evaluation Guidelines provide criteria and a reviewer workbook for examining instructional materials, while its Curriculum Navigation Reports apply those criteria to specific curricula. The League is careful to say these reports are informational, not recommendations, since no curriculum is perfect. Its navigation reports found that the following curricula meet or mostly meet the criteria:

Amplify CKLA, K–5: meets or mostly meets most criteria; strong in explicit foundational skills, knowledge building, comprehension, and writing.

SRA Open Court Reading, K–5: meets or mostly meets most criteria; praised for systematic phonics, explicit instruction, and knowledge-building units.

Really Great Reading, K–3: meets or mostly meets most criteria, with minimal red flags; especially strong in phonemic awareness, decoding, spelling, and fluency. It is mainly a foundational-skills program, and it is the one program on this list that also earned a Meets Expectations rating from EdReports, a useful signal for administrators seeking convergent evidence.

Reading Mastery Transformations. For readers interested in the closest connection between the science of reading and behaviorally designed instruction, Reading Mastery deserves special attention. Developed within Siegfried Engelmann’s Direct Instruction tradition, it uses carefully sequenced examples, scripted presentation, frequent student responding, immediate correction, cumulative review, and mastery-based progression. The Reading League’s October 2025 report found that it meets or mostly meets most criteria with minimal evidence of red flag practices (they did criticize the assessment practices and the kindergarten sequence).

Direct Instruction tradition does not sequence content according to the conventional order used in most phonics programs. It sequences examples by discriminability, usefulness, ease of pronunciation, likelihood of confusion, and the number of words students can generate from what they have already learned. Carnine (2017) explicitly writes that sequencing requires compromises among competing instructional considerations. He might therefore argue that an unconventional order is not a defect unless evidence shows that students learn less efficiently. It could be superior.

Final Thoughts

Science of reading curricula should improve when several research traditions converge. Reading research identifies the skills that must be taught: phonemic awareness, phonics, fluency, vocabulary, language comprehension, background knowledge, writing, and assessment. Intervention researchers such as Gersten and colleagues add screening, progress monitoring, and data-based intervention. Direct Instruction researchers such as Engelmann and Carnine add the design logic: careful sequencing, clear examples, active student responding, immediate correction, cumulative review, and mastery. Behavior analysts such as R. Douglas Greer press the question still further: did the teaching interaction produce a measurable change in the learner’s behavior? The Reading League’s Guidelines reflect much of this convergence, but they are not the science itself. They are one attempt to translate the science into curriculum review criteria.

Thank you for reading. Comments are appreciated.

Paul Coyne, PhD

References

California State Legislature. (2025). Assembly Bill No. 1454: Pupil literacy: Credential program standards and professional development: Instructional materials. California Legislative Information. https://leginfo.legislature.ca.gov/faces/billNavClient.xhtml?bill_id=202520260AB1454

Carnine, D. W., Silbert, J., Kame’enui, E. J., Slocum, T. A., & Travers, P. A. (2017). Direct instruction reading (6th ed.). Pearson.

EdReports. (2024, June 12). EdReports launches new ELA K–2 foundational skills curriculum reviews to better align with science of reading research and policy. https://edreports.org/resources/article/edreports-launches-new-ela-k-2-foundational-skills-curriculum-reviews-to-better-align-with-science-of-reading-research-and-policy

EdReports. (2024, August 21). Really Great Reading, grades K–2. https://edreports.org/reports/overview/really-great-reading-2023

EdReports. (2025, January 28). How to use EdReports’ earlier reports and review tools. https://edreports.org/resources/article/how-to-use-earlier-reports-and-review-tools

EdReports. (2025, July 31). EdReports launches next generation of curriculum reviews. https://edreports.org/resources/article/edreports-launches-next-generation-of-curriculum-reviews

EdReports. (2025, August 1). EdReports version 2.0 reports and tools: Frequently asked questions. https://edreports.org/resources/article/edreports-version-2-0-tool-revisions-frequently-asked-questions

EdReports. (2026, May 7). Orton-Gillingham Plus, grades K–2. https://edreports.org/reports/overview/orton-gillingham-plus-2025

EdReports. (2026, May 7). Wilson Fundations, grades K–2. https://edreports.org/reports/overview/wilson-fundations-2025

Goldenberg, C. (2026, April 21). Cautious optimism. Again. We Must End the Reading Wars . . . Now.

Greene, T. (2024, January 10). From policy to action: Why eight states banned three-cueing from K–3 reading instruction. ExcelinEd in Action. https://excelinedinaction.org/2024/01/10/from-policy-to-action-why-8-states-banned-three-cueing-from-k-3-reading-instruction/

Greer, R. D. (2002). Designing teaching strategies: An applied behavior analysis systems approach. Academic Press.

Hanford, E. (Host). (2022). Sold a story: How teaching kids to read went so wrong [Audio podcast]. APM Reports. https://features.apmreports.org/sold-a-story/

National Institute of Child Health and Human Development. (2000). Report of the National Reading Panel: Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction (NIH Publication No. 00-4769). National Institutes of Health. https://www.nichd.nih.gov/sites/default/files/publications/pubs/nrp/Documents/report.pdf

Scarborough, H. S. (2001). Connecting early language and literacy to later reading (dis)abilities: Evidence, theory, and practice. In S. B. Neuman & D. K. Dickinson (Eds.), Handbook of early literacy research (pp. 97–110). Guilford Press.

Schwartz, S. (2022, July 20; updated April 24, 2026). Which states have passed “science of reading” laws? What’s in them? Education Week. https://www.edweek.org/teaching-learning/which-states-have-passed-science-of-reading-laws-whats-in-them/2022/07

The Reading League. (2022). Science of reading: A defining guide. https://www.thereadingleague.org/wp-content/uploads/2022/01/Science_of_Reading_Defining_Guide_eBook.pdf

The Reading League. (2023). Curriculum evaluation guidelines. https://www.thereadingleague.org/wp-content/uploads/2023/03/The-Reading-League-Curriculum-Evaluation-Guidelines-2023.pdf

The Reading League. (2024). Curriculum navigation report: Amplify Core Knowledge Language Arts curriculum, second edition, grades K–5. https://www.thereadingleague.org/wp-content/uploads/2024/05/The-Reading-League-Curriculum-Navigation-Report-Amplify.pdf

The Reading League. (2024). Curriculum navigation report: Really Great Reading, grades K–3. https://www.thereadingleague.org/wp-content/uploads/2024/12/The-Reading-League-Curriculum-Navigation-Report-Really-Great-Reading.pdf

The Reading League. (2024). Curriculum navigation report: SRA Open Court Reading curriculum, grades K–5. https://www.thereadingleague.org/wp-content/uploads/2024/04/The-Reading-League-Curriculum-Navigation-Report-Open-Court.pdf

The Reading League. (2025, October). Curriculum navigation report: Reading Mastery Transformations. https://www.thereadingleague.org/wp-content/uploads/2025/10/The-Reading-League-Curriculum-Navigation-Report-Reading-Mastery.pdf

https://www.nytimes.com/2026/05/23/health/autism-therapy-clinics.html?searchResultPosition=1

There are other leaders in our field who are better suited for this task, such as those who govern ABAI, or are involved in certification of BCBAs or people like Ron Leaf who currently operate such programs. I haven’t operated an ABA program for children with autism since 2014. I am no longer familiar with current billing practices. When I was in business we were paid by California through the Regional Center system and various school districts. We had medical billing but it was a relatively small part.

What I’d like to comment on is depicted in the photo below.

I would not develop a center based program for children with autism that looked like this. I prefer to arrange the teaching in natural environments, such as the home, or in the classroom, or in the community, where ever the child’s goals and objectives dictated. In the home the therapists worked with the child on a 1:1 basis. Certain sessions were designed to train the parents in ABA teaching techniques. We taught them discrete trial teaching and pivotal response teaching. We taught them how to give good instructions, how to use prompts, how to reinforce, when to ignore behavior, how to arrange activity schedules, how to react to screaming, hitting and other behavior problems. We taught them how to collect data and more. I felt that if you trained the parents so they used the teaching techniques in naturally occurring situations the effects of the behavior therapy would extend beyond the therapy sessions. You did not need to provide 40 hours a week of therapy to obtain excellent results.

Regional Center provided funding for many of the children younger than 3 years old through Early Start. For that program we typically provided 2 hours a day of therapy 5 days a week with 2 hours of supervision. There were some children who received 4 hours a day 5 days a week with 4 hours supervision. Few if any received more. \The therapy was provided at home. Almost all rooms in the home were used for therapy, not just the bedroom. Some sessions occurred outside. We were not limited. We were looking for generalization. I have talked about the outcome data of our Early Start program here:

School districts provided the funding for children older than 3 years old. Sometimes their program was funded also by insurance or by Regional Center. If they were not in preschool or elementary school we provided a similar program to our Early Start program. If they were in preschool or elementary school we provided therapy at home and at school. However, there were some cases where we only provided service at home or at school not both.

The PALS program.

Mary Alice Coyne BCBA and I owned and operated The Learning Village Preschool in Tustin CA as well as Coyne & Associates. The preschool enrolled typical children, not children with special needs. However, we did have an arrangement with Coyne & Associates where The Learning Village offered a 600 square foot classroom and the outdoor play yard within the preschool to offer a Play And Language Skills program (PALS) for children with autism. The room included 2 kidney bean shaped tables seating 4 children at each table. Table activities included art, science, and snack. There was also space for floor activities, such as: free play with cause and effect toys, story time, dance and creative movement. The large room was open with no partitions. Incidental teaching techniques and natural environment teaching techniques were used to develop play, language and social skills. Also taught were readiness skills for functioning in a school setting, such as: using a cubby, participating in group, lining up, using a drinking fountain and restroom. PALS children interacted with the typical preschoolers during playground time.

The children with autism did not participate in the PALS program every day. In general, the children attended PALS two afternoons a week for 2 1/2 hour sessions. All behavior therapists were required to meet the requirements of Community Care Licensing who monitored our program.

Here is a staff memo written for the PALS program. At that time we called our behavior therapists, Early Childhood Interventionists or ECIs. Think of them as registered behavior therapists (RBT) which is the common label used today. Also, notice the “Centers” reference. The program was divided into centers: play yard activities, table top activities, and floor activities without using partitions. The ECI sometimes worked 1:1 with their child, sometimes they supervised whoever was in their section or zone. In basketball parlance we sometimes used a man on man defense, sometimes we used a zone defense.

Staff Memo

Data Collection at PALS

ECIs are assigned two children for whom to collect data; ECIs summarize and graph data for those children at the end of the session.

ECIs collect data during all opportunities with the following exceptions:

• Most of the time, ECI will be unable to collect data when ECI is Activity Coordinator for a Lesson.

• ECI will be unable to collect data when leading a Center and designated children are at a different Center.

Example:

• ECI = Tiffany

• Tiffany is responsible for data collection for two children, Colin and Quinn

Tiffany’s data collection responsibilities for the PALS session are outlined below in BOLD

1:30 Supervisor arrives (Celia) - organize and begin set up

2:15 ECIs arrive, prepare datasheets

• Tiffany arrives and prepares data sheets for Colin and Quinn. She reviews their binders and graphs to make sure that the current teaching step for each program is indicated on the data sheet for the day. Tiffany affixes the data sheets onto her clipboard.

2:30-2:55 Drop off and Outside Centers (3 centers, 7-8 minutes each)

Instructors 1-3, Activity Coordinator, Supervisor

Activity Coordinator (Not Tiffany)

• Tiffany is assigned to one of the Outdoor Centers: Play House

• During the first center rotation, Colin and another child are at Tiffany’s Center, the Play House.

• Tiffany runs the Center and when possible, creates opportunities to work on Colin’s target objectives.

• Tiffany records data for each teaching opportunity for Colin.

• During the second rotation, Quinn and another child are at Tiffany’s Center, the Play House.

• Tiffany runs the Center and when possible, creates opportunities to work on Quinn’s target objectives.

• Tiffany records data for each teaching opportunity for Quinn.

• During the third rotation two other children are at Tiffany’s Center, the Play House.

• Tiffany runs the Center with those children.

2:55- 3:15 Opening Circle

Activity Coordinator Not Tiffany

Tiffany supports children in her “zone” during Circle Activities.

During activity, whenever possible, Tiffany does one of the following:

• Creates teaching opportunities for Colin and Quinn

• Prompts Activity Coordinator to create teaching opportunities for Colin and Quinn

• Prompts another ECI to create teaching opportunity for Colin or Quinn

• Capitalizes on naturally occurring teaching opportunities.

Tiffany collects data on Colin’s performance and on Quinn’s performance during these opportunities.

3:15-3:20 Wash Hands

3:20-3:30 Snack time

Activity Coordinator Tiffany

• Tiffany coordinates the snack activity; acts as “Teacher.”

• Tiffany responds to directives from other ECIs to create teaching opportunities for children.

• Data for Colin and Quinn are not collected UNLESS Tiffany is able to manage the clipboard and conduct teaching trials with Colin and Quinn during a lull in the activity (e.g., all children are eating).

• As Activity Coordinator, Tiffany cleans up after the snack activity before re-joining group at the next scheduled activity.

3:30-3:45 Group Motor/ Movement Activity

Activity Coordinator Not Tiffany

(Pull out for 1:1 instruction begins – 15 minutes per child for 4 children)

Tiffany supports children in her “zone” during Gross Motor Activity.

During activity, whenever possible, Tiffany does one of the following:

• Creates teaching opportunities for Colin and Quinn

• Prompts Activity Coordinator to create teaching opportunities for Colin and Quinn

• Prompts another ECI to create teaching opportunity for Colin or Quinn

• Capitalizes on naturally occurring opportunities.

Tiffany collects data on Colin’s performance and on Quinn’s performance during these opportunities.

3:45-4:00 Tabletop activity- Art, Sensory

Activity Coordinator Not Tiffany

Tiffany supports children in her “zone” during Art Activity.

During activity, whenever possible, Tiffany does one of the following:

• Creates teaching opportunities for Colin and Quinn

• Prompts Activity Coordinator to create teaching opportunities for Colin and Quinn

• Prompts another ECI to create teaching opportunity for Colin or Quinn

• Capitalizes on naturally occurring opportunities.

Tiffany collects data on Colin’s performance and on Quinn’s performance during these opportunities.

4:00-4:05 Wash Hands

4:05-4:20 Story time and Songs

Activity Coordinator Tiffany

• Tiffany coordinates the Story Time Activity; acts as “Teacher.”

• Tiffany responds to directives from other ECIs to create teaching opportunities for children.

• Data for Colin and Quinn are not collected UNLESS Tiffany is able to manage the clipboard and conduct teaching trials with Colin and Quinn while conducing group activity.

• As Activity Coordinator, Tiffany cleans up after the story time activity before re-joining group at the next scheduled activity.

4:20-4:40 Cooperative Learning Centers (3 centers, 7-8 minutes each)

Instructors 1-3, Activity Coordinator, Supervisor

Activity Coordinator (Not Tiffany)

• Tiffany is assigned to one of the Cooperative Learning Centers: Cariboo

• During the first center rotation, Colin and another child are at Tiffany’s Center playing Cariboo.

• Tiffany runs the Center and when possible, creates opportunities to work on Colin’s target objectives.

• Tiffany records data for each teaching opportunity for Colin.

• During the second rotation, Quinn and another child are at Tiffany’s Center playing Cariboo.

• Tiffany runs the Center and when possible, creates opportunities to work on Quinn’s target objectives.

• Tiffany records data for each teaching opportunity for Quinn.

• During the third rotation two other children are at Tiffany’s Center, playing Cariboo.

• Tiffany runs the Center with those children.

(Pull out for 1:1 Instruction ends)

4:40-4:50 Goodbye Circle

Activity Coordinator Not Tiffany

Tiffany supports children in her “zone” during Circle.

During activity, whenever possible, Tiffany does one of the following:

• Creates teaching opportunities for Colin and Quinn

• Prompts Activity Coordinator to create teaching opportunities for Colin and Quinn

• Prompts another ECI to create teaching opportunity for Colin or Quinn

• Capitalizes on naturally occurring opportunities.

Tiffany collects data on Colin’s performance and on Quinn’s performance during these opportunities.

4:50-5:00 Pick up and Structured Play (outside)

At 5:00, Tiffany summarizes and graphs data in Colin’s and Quinn’s binders.

Two Issues to Consider:

1. References to Tiffany’s “Zone” are indicated above. This implies that Tiffany is not supporting Colin and Quinn exclusively during those activities, but instead is positioned in an area and supports the 2-3 students in that area. We will need to flesh this out a little bit so staff know where they should be positioned.

An alternative to the “Zone” would be having the ECIs support the children for whom they are collecting data (e.g., Tiffany supporting Colin and Quinn during all activities except activities during which Tiffany is Activity Coordinator).

I think that there are advantages and disadvantages to both systems. We should put this issue on the agenda for the next PALS meeting.

2. The 1:1 Pull Out was not explicitly discussed above. Those details need to be worked out. In the example above, for instance, Tiffany might be responsible for conducting the 1:1 with Colin and with Quinn during their scheduled 1:1 sessions. Or Celia could do the 1:1 to start. Or one ECI does all the 1:1 for all the 1:1 sessions on a given day (to minimize staff movement in the group). We should put this issue on the agenda for the next PALS meeting.

Final Thoughts

The PALS program was available to children who lived near Tustin, CA. The children we served in San Diego, for example, did not have access to a PALS program. The description of the PALS program came directly from an internal memo provided to staff. I thought it illustrated how we ran our center based program within The Learning Village Preschool. It seems different than the center based program operated by Compleat Kids, as described by the NYTimes. The photo below is not the classroom where the PALS program operated. This is a different part of the school, however it shows you the open design and the various activity centers in the program. Unfortunately, I don’t have a photo of the PALS program without children’s faces easily identified. Here the child’s faces are obscured. I’ve included also a photo of the play yard the PALS program used. It is a typical preschool environment. It is not a partitioned program in a corporate building designed to maximize billable hours. Notice that the PALS program began at 2:30 p.m., i.e., after children’s nap time was over. Children took naps at the Learning Village with blankets and cots from 12:30 - 2:00 p.m. We certainly didn’t wake them up after 7 minutes so we could bill for therapy, as described in the NYTimes article.

Written by Paul Coyne PhD and Mary Alice Coyne MAEd.

Notes:

No AI was used in writing this article.

The Learning Village is now closed. Coyne & Associates was sold to private equity by my previous employees.

No references.

Thanks for reading Paul’s Substack of Useful Information! Subscribe for free to receive new posts and support my work.

Tired of making
hundreds of copies?

We are too.

Finally, Morningside Math Facts are now available as consumable student workbooks!

$19.50

Preorder for Fall 2026!

Improvements include:

- New, leaner sequence (Bye Bye 0s and 1s!)

- Reformatted Box Sheets

- Integrated Data Collection (Student Daily Charts, Timings Trackers, and Progress Tracker)

- Increased guidance for teachers and students

- Digital freebies including wall charts, support posters, and more!

These are designed to be consumable student workbooks.

Less copying. More fluency.

Share

I’m recovering from spinal fusion surgery faster than I expected. Initially, I was prescribed hydrocodone, a narcotic. It reduced my pain, but I was still very sore. Plus, I was drowsy constipated and it conflicted with other drugs I was taking. I tried switching to Tylenol alone, but Tylenol wasn’t strong enough.

So, I remembered an article I read in the Wall Street Journal about a non narcotic drug, that is not addictive, does not make you drowsy and is not in conflict with other drugs I’m taking, Journavx. It’s made by Vertex. I may buy a few shares of the stock.

I asked my Doctor to prescribe Journavx. He was willing to provide the prescription but he warned me that is was not covered by my insurance. I’d have to pay out of pocket. He estimated that it would cost about $500 for a month’s supply. I haven’t played golf in 6 months, so with the money I saved I thought I’d splurge on Journavx.

I purchased Journavx at the Vons grocery store pharmacy for $98. Money well spent. I am on it now several days and I am 90% pain free. I take one every 12 hours. For me, it was more effective than the hydrocodone plus no noticeable side effects. And certainly more effective than Tylenol.

Journavx was not offered to me. I had to request it. It’s only because I happen to read an article about Vertex and their new holy grail medication that I obtained it. If you are ever going into surgery ask your doctor about it. Remember, I got it for $98.

Paul Coyne PhD

UPDATES

I want to amend my pain free percentage. I said 98%. Probably 85-90% is more accurate once I started to get out of bed and walk around. At times it is not strong enough.

May 29th. I made an error on the cost of the medication. I was told that my insurance would not pay for Journavx. So when I was charged the $98 I thought that was the cost. I was wrong. I have since found out that my insurance paid for about $300 of the $400 cost of the medication, so my co-pay was $98. Now I’m up for a refill and my insurance will no longer pay anything, I will pay the $400 myself for a 2 week supply. For me, the cost is worth it. Your miles may vary.

I went on the Journavx website and there is a coupon you can use to obtain a two week supply for $30. I’ve used that twice.

Leave a comment

Subscribe now

If you miss my comments here is a list of articles I wrote. I like these articles and I hope you might find them an interesting read until I get back. Keep scrolling there are some sleepers toward the end.

1. One way to fight addiction, homelessness, and unemployment.

There is more. Check out the website. And thanks for all people who have read my articles and committed. I am very appriciatative, Any thoughts, please comment.

Paul Coyne

Subscribe now

I’m always looking for articles that discuss applications of behavior analysis outside of Autism. Don’t misunderstand me. I spent much of my career working with children diagnosed with autism. Nevertheless, I like to point out that ABA is much more than discrete trial teaching and pivotal response teaching. Thanks to Fred and his substack, The Fixed Interval, I came across an article by Mackenzie Denner et. al, (2026) that discusses a promising ABA based treatment program for addiction, unemployment and homelessness. This is what I found.

America has treated addiction, unemployment, and homelessness as separate problems requiring separate solutions. Clinics handle the addiction. Workforce programs handle the unemployment. Shelters and vouchers handle housing. Each operates on their own funding with their own strategy for change. The results have been discouraging. People complete substance use treatment and relapse. They get jobs and lose them.

A new study in the Journal of the Experimental Analysis of Behavior takes a different approach. Mackenzie Denner and her colleagues at Northern Michigan University and Johns Hopkins report on an intervention that treats these problems simultaneously rather than across three separate programs. The study is titled Employment outcomes of abstinence contingent wage supplements for adults experiencing homelessness and alcohol use disorder, and it can be read here: https://doi.org/10.1002/jeab.70104

The intervention is called Abstinence Contingent Wage Supplements, or ACWS. The principle is as follows. Pay people extra money (e.g.,$8) on top of their regular wages, but only if they stay sober and keep working. If they drink, the supplement drops to a dollar an hour. If they stay sober, the bonus climbs back up to $8 a dollar per day.

Limits of Earlier Programs

The authors discussed the failures of earlier conditional cash transfer programs, including Connecticut’s Jobs First initiative and Opportunity NYC Family Rewards. Both paid people for good outcomes. Both produced disappointing results. The authors argued that these programs lacked the precision that a behavioral analysis requires. For example, monthly bonuses for general good behavior are not as effective as daily, sensor-verified contingencies tied directly to drinking.

According to the authors, people in substance use treatment are more than twice as likely to complete it if they are employed. Work itself appears to have therapeutic value. Most treatment programs do not arrange for employment, and most employment programs do not address substance use. ACWS is built on the premise that combining the two produces something neither can produce alone.

Method

The trial enrolled 119 adults in Baltimore who met DSM-5 criteria for alcohol use disorder, were experiencing homelessness or housing instability, and had earned less than $200 in the previous month. After a 30 day induction, participants were randomly assigned to typical care or to ACWS. The intervention ran 26 weeks. The paper Denner and colleagues published is a secondary analysis of the 62 ACWS participants, focused on what paystubs and biosensors revealed about employment and income. The original trial was reported by Novak and colleagues in 2023 and showed that ACWS increased abstinence and employment relative to typical care: https://pubmed.ncbi.nlm.nih.gov/37541600/+2

Sobriety was verified continuously through transdermal alcohol monitors worn on the skin. The bracelets continuously measured alcohol secreted through sweat, leaving no room for self reporting errors. Employment was verified through official paystubs. Participants could earn up to $8 dollars an hour in supplements on top of their wages. A positive alcohol reading or a missed sensor day reset the hourly supplement to one dollar. Sobriety could rebuild it gradually, a dollar per day. Opiate- and cocaine-negative urine samples were also required to maintain maximum pay.

Results

Two-thirds of ACWS participants found verified employment during the intervention. the average time to first paycheck was six weeks. Once employed, participants worked an average of 33.7 hours per week and earned about $1,054 per biweekly pay period. Eighty percent earned enough during at least one pay period to qualify as living above the federal poverty line. While employed, they were above poverty during 79 percent of pay periods.

The authors also compared paystub income with what participants self-reported. The two correlated strongly, with a Pearson coefficient above 0.89. Self-reports ran slightly higher than verified earnings, likely reflecting under-the-table work. Eleven participants who claimed employment never produced paystubs. The discrepancy is small but instructive. It suggests that program evaluations relying on self-report alone may overstate their results.

Eight participants who found jobs never crossed the poverty line during any pay period. They tended to be in unpaid training programs, working part-time, or starting work too late in the intervention to accumulate enough income.

Implications

Providing money noncontingently does not reduce addiction. Money delivered on a precise contingency schedule, tied to objectively verified outcomes, with immediate consequences for backsliding, does.

Current treatment programs may not be taking advantage of the substantial treatment effects by not integrating wage-based contingencies into recovery. Policymakers who concluded that conditional cash transfers do not work in the this context may have drawn the wrong conclusion. The earlier programs failed because the contingencies were too loose and the verification was too weak.

The study has limitations. Twenty-six weeks is not a lifetime, and the durability question is open. The cost of biosensors, supplements, and employment specialists is not trivial, though a 2023 cost-effectiveness analysis of an earlier ACWS trial offers some preliminary numbers: “ACWS cost $11,310 per participant over the 12-month intervention period. Total intervention and healthcare costs per participant over the intervention period were $20,625 for usual care and $30,686 for ACWS. At the end of the intervention period an additional participant with a negative drug test cost $1,437 while an additional participant employed cost $915.”https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10207811/

Generalization to other substances, other cities, or less motivated participants remains an empirical question.

Still, the result deserves attention. Here is a program that gets two-thirds of homeless adults with alcohol use disorder into verified work. Most of them were above the poverty line for meaningful stretches. It seems, Denner, Toegel Hotyn and Silverman developed an effective program. Certainly, a step in the right direction.

Thanks for reading. I appreciate your comments. I wrote the article, however I used Gemini to provide the links and references. Paul Coyne PhD

Leave a comment

References

Bloom, D., Scrivener, S., Michalopoulos, C., Morris, P., Hendra, R., Adams-Ciardullo, D., & Walter, J. (2002). Jobs First: Final report on Connecticut’s welfare reform initiative. Manpower Demonstration Research Corporation.https://acf.gov/sites/default/files/documents/opre/ct_jobsfirst.pdf

Denner, M., Toegel, F., Holtyn, A. F., & Silverman, K. (2026). Employment outcomes of abstinence-contingent wage supplements for adults experiencing homelessness and alcohol use disorder. Journal of the Experimental Analysis of Behavior, 125(3), e70104. https://doi.org/10.1002/jeab.70104

Holtyn, A. F., Jarvis, B. P., & Silverman, K. (2017). Behavior analysts in the war on poverty: A review of the use of financial incentives to promote education and employment. Journal of the Experimental Analysis of Behavior, 107(1), 9-20. https://doi.org/10.1002/jeab.233

Novak, M. D., Toegel, F., Holtyn, A. F., Rodewald, A. M., Arellano, M., Baranski, M., et al. (2023). Abstinence-contingent wage supplements for adults experiencing homelessness and alcohol use disorder: A randomized clinical trial. Preventive Medicine, 176, 107655. https://doi.org/10.1016/j.ypmed.2023.107655

Orme, S., Zarkin, G. A., Luckey, J., Dunlap, L. J., Novak, M. D., Holtyn, A. F., Toegel, F., & Silverman, K. (2023). Cost and cost-effectiveness of abstinence contingent wage supplements. Drug and Alcohol Dependence, 244, 109754.https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10207811/

Riccio, J., Dechausay, N., Miller, C., Nuñez, S., Verma, N., & Yang, E. (2013). Conditional cash transfers in New York City: The continuing story of the Opportunity NYC-Family Rewards demonstration. MDRC.https://www.mdrc.org/sites/default/files/Conditional_Cash_Transfers_FR%202-18-16.pdf

Silverman, K., Holtyn, A. F., & Toegel, F. (2019). The utility of operant conditioning to address poverty and drug addiction. Perspectives on Behavior Science, 42(3), 525-546. https://doi.org/10.1007/s40614-019-00203-4

The Jobs First final report is here: https://acf.gov/sites/default/files/documents/opre/ct_jobsfirst.pdf

The Opportunity NYC-Family Rewards report is here: https://www.mdrc.org/sites/default/files/Conditional_Cash_Transfers_FR%202-18-16.pdf

Thanks for reading Paul’s Substack of Useful Information! Subscribe for free to receive new posts and support my work.

In 2007 Siegfried Engelmann wrote, Teaching Needy Kids In Our Backward System: 42 Years of Trying. The book is an autobiographical account of Engelmann’s efforts to provide effective instruction to kids “at risk,” particularly children of poverty. Much of the book is spent on Project Follow Through (1968–1977), which involved 180 communities and over 200,000 children in grades kindergarten through third grade where he tested the efficacy of Direct Instruction curricula. His results were compared to 22 other models of teaching, from discovery learning to behavioral approaches. Direct Instruction ranked first in every measured category. It beat every other model tested. Yet his approach was not adopted by many school districts.

So I thought, there must be some elementary schools that use DI curriculum in both general education and special education. I decided to locate a few of them and report back.

What Is Direct Instruction?

Direct Instruction refers to curriculum developed in the 1960s by Siegfried Engelmann and Wesley Becker, initially at the University of Illinois and subsequently further developed at the University of Oregon. Its commercial programs include Reading Mastery, Connecting Math Concepts, Spelling Mastery, Language for Learning, Corrective Reading, and Reasoning and Writing, among others. All are published today by McGraw-Hill and supported by the National Institute for Direct Instruction (NIFDI) at nifdi.org.

DI has three defining features. First, every lesson is carefully scripted. Second, students are grouped by skill level rather than age or grade, and they move through the sequence only when they demonstrate mastery. Third, teachers ask frequent questions, students respond in unison, and the pace is designed to keep every child engaged.

John Hattie's 2009 synthesis of educational research, Visible Learning, examined four meta-analyses of DI covering 304 studies and more than 42,000 students. The average effect size of 0.59 means the typical DI student outperformed roughly 72 percent of students in conventional classrooms, the strongest result Hattie found for any curriculum. The American Federation of Teachers named DI one of six promising schoolwide reform programs. The American Institutes for Research gave it its highest rating among whole-school reform models.

The Baltimore Curriculum Project

The Baltimore Curriculum Project (BCP) started in 1996. Today BCP operates six neighborhood charter schools serving more than 3,000 students in Pre-K through 8th grade.

Direct Instruction is used primarily for reading and language arts. BCP academic coaches meet weekly with school leaders to review student data, identify students who need additional support, and accelerate those who are ready to move ahead. The DI programs assess students every five to ten lessons.

Baltimore City Public Schools ranks among the lowest-performing in Maryland, with approximately 23 percent of students proficient in reading and 11 percent proficient in math across the district. Within that same city, Hampstead Hill Academy reports reading proficiency rates of 63 to 69 percent and math proficiency of 44 to 47 percent. In 2025, Hampstead Hill was the only Baltimore City Public School to receive a five-star rating on the Maryland State Report Card. It ranks in the top 30 percent of all elementary schools in Maryland and fourth among all elementary schools in Baltimore.

BCP has been candid about uneven results, particularly in mathematics. In recent years, BCP acknowledged that while DI has driven consistent gains in reading, math outcomes varied by school. BCP has been transitioning to a new mathematics program called Reveal Mathematics while keeping DI for reading.

For more information: baltimorecp.org

Arthur Academy Charter Schools, Portland Metro Area, Oregon

Arthur Academy operates six tuition-free public charter elementary schools in the greater Portland area. The school was founded on the principle that Direct Instruction, properly implemented, can serve every child, not just those falling behind. Bonnie Grossen, a long time DI researcher and advocate, chairs the board of directors. The schools serve students in kindergarten through fifth grade.

The six campuses are David Douglas Arthur Academy (Portland), Portland Arthur Academy, Gresham Arthur Academy (Gresham), Reynolds Arthur Academy (Troutdale), St. Helens Arthur Academy, and Woodburn Arthur Academy. Each uses DI programs as the core curriculum for all students across all academic levels.

The test score data across the network is mixed but generally positive, particularly when student demographics are taken into account. Portland Arthur Academy, serves a student body that is 49 percent economically disadvantaged. They reported a reading proficiency of 67 percent and math proficiency of 37 percent. Both are above Oregon state averages (44 percent reading, 33 percent math). In the 2024–2025 school year, Portland Arthur Academy’s 3rd, 4th, and 5th grade English Language Arts proficiency rates reached 77, 60, and 71 percent, compared to 55, 59, and 60 percent for the Portland public school district. Gresham Arthur Academy reported a 77 percent reading proficiency and a 44 percent math proficiency, above the Gresham-Barlow district averages of 39 percent reading and 22 percent math. Reynolds Arthur Academy, serving a population that is 71 percent economically disadvantaged, consistently outperforms its local district in both reading and math, and its 5th grade ELA proficiency rate of 54 percent compares to the district’s 27 percent.

Arthur Academy’s commitment to DI is not without criticism. Some parents and observers objected to the emphasis on repetition, choral responding, and to the discipline required to maintain the instructional pace. Others are emphatic in their support. “This school is amazing,” one David Douglas Arthur Academy parent wrote online. “It’s a small public charter school that excels in teaching kids to read, write, and do math all before leaving kindergarten.” Another Portland Arthur Academy parent wrote: “This school has changed my son’s life. The teachers and staff care deeply about every single student.” The contrast in perceptions reflects a genuine disagreement that surrounds Direct Instruction: the method produces measurable results, but the experience of being in a scripted, fast-paced classroom does not appeal to all teachers.

For more information: arthuracademy.org

Morningside Academy, Seattle, Washington

Morningside Academy occupies a unique place in the DI universe. Founded in 1980 by Kent R. Johnson, Ph.D., it is a small private school in Seattle’s Capitol Hill neighborhood serving students in grades 1 through 9. Tuition runs approximately $25,400 per year. The school is accredited and has never been replicated at scale, but it has influenced thousands of educators through its Morningside Teachers’ Academy, which has trained professionals in more than 150 schools and agencies across North America and Europe.

What makes Morningside distinctive is its use of Direct Instruction and the way it combines DI with two other systems: Precision Teaching, developed by Ogden Lindsley, and Think-Aloud Problem Solving (TAPS), based on the work of Arthur Whimbey. Together, these form what Johnson calls the Morningside Model of Generative Instruction.

Direct Instruction is used to build skills in reading, writing, and mathematics. Once students have mastered those skills to fluency, (a concept central to Precision Teaching) instruction shifts toward reasoning and problem solving using TAPS protocols. The goal is what Morningside calls generativity: the ability to apply learned skills in novel contexts that were never explicitly taught.

Most of Morningside’s students have normal intelligence yet arrive having failed in conventional school settings. Many carry diagnoses of dyslexia, dysgraphia, dyscalculia, or ADHD. Some simply arrived in the wrong classroom at the wrong time and fell behind. The school does not serve students with significant behavioral or developmental disabilities.

The outcomes are among the most thoroughly documented of any small private school in the country. An 11-year study of standardized achievement test gains found that reading averaged 2.5 grade levels of growth per school year, language arts approached 4 grade levels of growth per year, and mathematics exceeded 3 grade levels of growth per year. Over more than 40 years, less than one percent of tuition has ever been refunded under Morningside’s standing guarantee: any student who does not gain at least two grade levels in their area of greatest deficit within one school year receives a full tuition refund.

Parent responses to Morningside tend toward the emphatic. “After third grade in public schools with an IEP and during the pandemic, he was far behind and hated school,” one parent wrote. “He has been at Morningside for two years now and has progressed from one to two years behind grade level to grade level or better in all areas.” There is an intensive data tracking system. Students learn to plot their own data on Standard Celeration Charts, which provides a measure of their performance. Students and parents receive daily progress reports.

For more information: morningsideacademy.org

Mt. Helix Academy, La Mesa, California

Mt. Helix Academy is a private, non-sectarian school in the La Mesa area of San Diego County, operated by The Institute for Effective Education (TIEE). It was acquired by TIEE in 1994 and is located at 5955 Severin Drive, La Mesa, California. The school serves students from transitional kindergarten through 8th grade and is one of the very few schools in the country that uses full schoolwide Direct Instruction for both general education and special education students.

The curriculum includes Reading Mastery, Connecting Math Concepts, Spelling Mastery, and Language for Learning. The school also integrates Precision Teaching and Applied Behavior Analysis. TIEE also operates four other San Diego-area schools: Children’s Workshop, Mission Valley Academy, COOK Education Center, and Urban Skills Center. Those schools serve students with IEPs through contracts with local public school districts. Mt. Helix Academy is the only TIEE school that teaches general education students alongside its special education program.

Academic performance is as follows:

Reading Performance (Spring 2024) — Median National Percentile Rank by Grade Level

Transitional Kindergarten: 84th percentile | Kindergarten: 68th percentile | 1st Grade: 99th percentile | 2nd Grade: 98th percentile | 3rd Grade: 79th percentile | 4th Grade: 50th percentile | 5th Grade: 82nd percentile | 6th Grade: 62.5th percentile | 7th Grade: 60th percentile | 8th Grade: 78th percentile

Math Performance (Spring 2024) — Median National Percentile Rank by Grade Level

Transitional Kindergarten: 68th percentile | Kindergarten: 75th percentile | 1st Grade: 73rd percentile | 2nd Grade: 92nd percentile | 3rd Grade: 66th percentile | 4th Grade: 50th percentile | 5th Grade: 67th percentile | 6th Grade: 67th percentile | 7th Grade: 55th percentile | 8th Grade: 60th percentile

Middle School

The academy reports that for their junior high cohort, 63% of students scored in the top quartile (the top 25% nationally) on at least one SAT-10 subtest, and 32% scored in the top decile (the top 10% nationally).

This data also mirrors the “3rd Grade Cliff” phenomenon acknowledged by IDEA Public Schools. At Mt. Helix, the median national percentiles normalize in the middle grades (dropping closer to the 50th and 60th percentiles). This reflects the inherent shift in educational demands: early grades test foundational decoding and computation, where DI excels at building rapid fluency, while upper grades test broader comprehension, vocabulary, and abstract reasoning. Nevertheless, the median student at Mt. Helix still consistently performs at or above the national average in every grade and subject.

Emerson Elementary and Fickett Elementary, Nebraska and Georgia

Both schools are featured by NIFDI in their documentary Helping Kids Soar.

Emerson Elementary is a small public school in Alliance, a rural town in western Nebraska. Fickett Elementary is a public school in Atlanta, Georgia. Both implemented the full schoolwide DI model with NIFDI support and both documented significant academic gains across their student populations, including students who were already performing at or above grade level when the program began.

At Emerson, student performance improved substantially. By 2002, 100 percent of its fourth graders met or exceeded state standards in reading and 92 percent did so in math. Teachers and parents reportedly liked the program. However, DI is no longer used in the district. Emerson moved away from DI following changes in district leadership and the expiration of a grant. Based on the district’s recent Title I Schoolwide Plans, Emerson Elementary currently uses Amplify CKLA (Core Knowledge Language Arts) for its core reading curriculum across its K–2 classrooms. They also use mCLASS Burstand Amplify Reading for their reading interventions. For mathematics, they recently transitioned to a new series called Reveal Math.

Principal Cheryl Twyman transformed Fickett Elementary from one of the lowest-performing schools in Atlanta to a “School of Excellence.” Student proficiency in reading and math improved, and the school received national attention for closing the achievement gap. For example, 75 percent of all students met second grade reading mastery criteria.

The administration eventually transitioned Fickett away from Direct Instruction toward an International Baccalaureate (IB) Primary Years Programme, which heavily emphasizes inquiry-based learning and discovery methods, a departure from DI. for example, they use HMH Into Reading and the Primary Years Program (PYP) curricula. However, because of the new state laws forcing districts back toward the "Science of Reading," Fickett is now having to integrate structured, systematic phonics back into that inquiry based IB framework.

Today, the school’s academic performance is struggling. Based on the most recent state testing aggregations for the 2024–2025/2025–2026 reporting cycles, Fickett Elementary’s scores are: Reading/Language Arts Proficiency: 16 percent; Math Proficiency: 16 percent.

I note the 16 percent proficiency data to provide a quantifiable contrast between the outcomes of the school’s historical, structured DI program and its current inquiry based model.

IDEA Public Schools — Texas

IDEA Public Schools is a large, well regarded charter network serving more than 76,000 students across 125 campuses in Texas. Its population is predominantly Hispanic and economically disadvantaged. IDEA partnered with NIFDI to implement Direct Instruction in all Pre-K through 2nd grade classrooms across its elementary schools.

IDEA has integrated DI into a hybrid model that combines scripted instruction with computer-based instruction. Computer-based curricula include DreamBox Math, Imagine Math, The ARZone, Renaissance Accelerated Reader, and MyOn.

In grades K–2, DI is the primary core curriculum for reading, language, and math. All students in these grades participate in DI groups daily. The district specifically utilizes Reading Mastery Transformations to ensure students are fluent readers by the end of 2nd grade. IDEA also uses Direct Instruction Spoken English (DISE) for English Language Learners and Direct Instruction Decoding for students identified as dyslexic.

According to IDEA’s most recent District Improvement Plan (published for the 2024–2025 school year, reflecting Spring 2024 data): 85% of all 2nd graders ended the year reading on grade level. In specific high performing regions, like the Rio Grande Valley and Permian Basin, over 90% of 2nd graders ended the year reading on grade level. For students who are two or more grade levels behind, IDEA reported that 42% grew two full reading years or reached grade level reading by the end of the year.

Once students transition out of the K–2 DI block, they begin taking the State of Texas Assessments of Academic Readiness (STAAR). Based on Spring 2024 district-wide aggregations: approximately 73 to 75% of all tests reached grade level (passing); approximately 46 to 47% met the proficient standard; and approximately 18% mastered grade level. IDEA consistently sees stronger performance in English Language Arts and Reading than in mathematics. This pattern mirrors the math struggles noted at the Baltimore Curriculum Project.

An instructive data point emerged in IDEA’s internal review. In their Improvement Plan, district leaders formally noted a gap between their K–2 successes and upper-elementary testing: “There is not a correlation between the end of year 2nd grade reading proficiency and 3rd grade reading STAAR performance.” This illustrates a consequence of DI implementations. Even when a DI program successfully teaches 85 to 90 percent of children to decode and read fluently by age 8, maintaining those gains requires an equally rigorous, knowledge-building curriculum when state testing begins in 3rd grade.

For more information: ideapublicschools.org

What Teachers, Parents, and Administrators Say

The experience of Direct Instruction is polarizing in a way that other evidence-based reforms generally are not. The scripted lessons, the choral responding, the fast pace, and the regrouping of students based on data, all depart significantly from what most teachers were trained to do and what most parents expect.

Teachers who resist DI most often describe the scripted lessons as deprofessionalizing. They feel that the script takes away their judgment and reduces them to a delivery mechanism. Some teachers who have worked through that initial resistance describe a different experience: that once students have the foundational skills that DI instills, the classroom becomes richer, not poorer, because students can actually engage with content at a meaningful level. Teachers who participated in a multi-school DI implementation in Broward County, Florida, noted that while some initially felt the repetition was boring for faculty, over time they found ways to innovate within the structure and became drawn into the model.

Parents who choose DI schools tend to be persuaded by outcomes. The consistent testimony at Arthur Academy, Morningside, BCP schools, and Mt. Helix is that their children learned to read with speed and accuracy in ways that had not happened in previously attended schools. Parents and professionals who are unhappy with DI schools tend to look for project-based learning, creative discovery, and child-led inquiry.

Administrators who sustained DI implementations over years emphasize two things consistently. First, training matters. DI materials sold commercially without the accompanying coaching and data systems may produce mediocre results; DI with full NIFDI implementation support produces the results described in the research literature. Second, DI implementations may become upended when principals change, when district leadership shifts, or when budget pressure forces cuts to coaching. The quality of implementation may worsen. The schools that have sustained excellent DI outcomes over decades have tended to have stable leadership and an institutional culture that protects the model from the pressures that erode school reforms.

Costs: DI vs. CKLA

Several school districts have transitioned from Direct Instruction to newer curricula like Core Knowledge Language Arts (CKLA), a shift often driven by budget constraints. While both programs require purchasing physical materials, DI demands a much steeper upfront and ongoing financial investment. According to the Institute of Education Sciences, initial DI student materials run $100 to $150 per student, with teacher presentation sets costing $600 to $1,000 per grade level. However, the true financial differentiator is the cost of maintaining instructional fidelity. A full implementation contract with the National Institute for Direct Instruction (NIFDI)—which includes remote coaching at $121 per hour, data review at $62.50 per hour, and multi-day leadership academies at $2,750 per session—can easily cost a mid-sized school $50,000 to over $100,000 just for coaching and monitoring in its first year. In contrast, CKLA operates on a more standard pricing model. Student consumables typically cost $35 to $50 per student, and while teacher kits range from $1,500 to $2,500 per classroom, the associated training costs are significantly lower. Instead of year-round, weekly behavioral coaching, CKLA relies on standard professional development workshops priced between $2,500 and $5,000 per day for the entire staff, making it a considerably cheaper model to adopt and maintain.

Final Thought*

If Direct Instruction remains rare in American classrooms, the reason is unlikely to be lack of evidence. More often, it reflects the difficulty of changing institutional habits: how universities prepare teachers, how administrators are trained, and how schools provide instruction. Most Colleges of Education have a developmental, cognitive, constructionist orientation which is at odds with the learning principles behind Direct Instruction. Direct Instruction principles were developed through the scientific method and applied to produce DI curricula. That is, they developed and tested the lessons empirically, with curricula revised until student mastery data confirmed they worked Not all curricula are developed this way. Frequently curricula are determined by theory with the research to follow afterward.

The poster child for bad curricula is the whole language approach to reading. The nationwide abandonment of the whole-language approach and the acceptance of curriculum based on the ‘science of reading’ is, one hopes, a step in the right direction.

PS. Perhaps one way to increase DI popularity would be to recreate a Morningside School type of school with a large enough budget, to be in a beautiful building in an affluent neighborhood instead of a low income neighborhood. When the student’s academics become stellar, publish news about the school, bring some publicity. Charge a substantial tuition so costs can be covered and make a decent profit. Perhaps other schools would follow suit.

Share

Leave a comment

Subscribe now

I wrote the article. I obtained the data and references using Gemini and ChatGPT and crossed referenced for accuracy. Paul Coyne PhD

Notes

1. In 2009, a massive state investigation uncovered systemic, district-wide cheating on the state CRCT exams across Atlanta Public Schools between 2001 and 2009. While Principal Twyman was reported to have been cleared of intentional wrongdoing and subsequently returned to a principal role within APS, the broader scandal led many education researchers to view Atlanta’s “miracle” state test scores from that decade with deep skepticism. NIFDI’s internal, independent tracking of early reading fluency — which relies on direct student observation rather than the state’s written CRCT exam — remains a valid data point.

References

National Institute for Direct Instruction: nifdi.org

Association for Direct Instruction: adihome.org

Baltimore Curriculum Project: baltimorecp.org

Arthur Academy Charter Schools: arthuracademy.org

Morningside Academy: morningsideacademy.org

Bethel School District: bethelsd.org

IDEA Public Schools: ideapublicschools.org

Reading Rockets overview of Direct Instruction: readingrockets.org/topics/curriculum-and-instruction/articles/direct-instruction-di-reading-intervention-program

Hattie, J. (2009). Visible Learning: A Synthesis of Over 800 Meta-Analyses Relating to Achievement. Routledge.

Adams, G. and Engelmann, S. (1996). Research on Direct Instruction: 25 Years Beyond DISTAR. Educational Achievement Systems.

Johnson, K. R. and Layng, T. J. (1992). Breaking the structuralist barrier: Literacy and numeracy with fluency. American Psychologist, 47(11), 1475-1490.

Morningside Academy outcomes: aubreydaniels.com/media-center/organizational-solutions/articles/morningside-academy-building-foundations-for-student

Hampstead Hill Academy profile: usnews.com/education/k12/maryland/hampstead-hill-academy-202718

Arthur Academy school profiles: schooldigger.com/go/OR/schools/1004001667/school.aspx

Engelmann, S. Teaching Needy Kids In Our Backward System: 42 years of trying. ADI Press 2007

The Work of Judy Cameron and David Pierce

Permit me to summarize the work of Judy Cameron and David Pierce who offer strong evidence that providing extrinsic rewards under most circumstances will increase not decrease intrinsic motivation.

Judy Cameron and W. David Pierce examined the conditions under which the presentation of rewards either negatively affected or improved intrinsic motivation. They published the results in their book “Rewards and Intrinsic Motivation: Resolving the Controversy”(2002) and in several subsequent publications. They reviewed over 100 research studies that investigated the effects of reinforcement on intrinsic motivation, including the work of Edward Deci and his colleagues. They defined intrinsic motivation by measuring the time a child spent engaged in a target activity without extrinsic reinforcement given a choice of activities.

Their analysis indicated the following:

1) providing rewards for activities that are inherently reinforcing did not reduce the frequency a person will engage in the activities once the extrinsic rewards were no longer provided;

2) Rewards provided for low-interest tasks increased the chances the task became more inherently reinforcing once the artificial rewards were no longer forthcoming;

3) reinforcers (including praise) provided for high-interest tasks increased the chances the task remained inherently reinforcing once the extrinsic rewards were no longer forthcoming;

4) When rewards were presented unexpectedly, there was no effect on intrinsic motivation;

5) When rewards were presented independent of the activity (i.e., regardless of what the person did), there was no effect on intrinsic motivation;

6) When people were reinforced for doing a task with little regard to the quality of their work, or with little regard to ensure the reinforcer was considered a reinforcer by the person, the person tended to do the activity less often in a free-choice period provided thereafter;

7) When people were reinforced for surpassing a personal goal or surpassing the performance of others intrinsic motivation increased;

8) When people were reinforced for each unit completed (e.g., here are 10 puzzles, I’ll give you 10 cents for each puzzle completed), yet did not obtain the full reward available (e.g. only completed 6 of 10 puzzles) people tended to do the activity less in a free choice situation presented thereafter;

9) If people were told they could obtain a certain level of reward for doing an activity within a certain time limit but did not obtain that level of reward because they ran out of time, they tended to do the activity less in a subsequent free choice situation.

10) When researchers exposed children to a baseline period where no external rewards were available for engaging in an inherently enjoyable activity, then began an equal period where rewards were provided, then returned to baseline where no external rewards were available, children who were rewarded spent as much (or more) time on the enjoyable activity during the post reward phase as they did during the initial baseline period.

11) Negative effects of reward (e.g., #8 & 9) were transitory when task performance was reinforced over several occasions.

The Promise of a Reward

Extrinsic reinforcement itself did not result in a decrease in intrinsic motivation but rather it was the way the promise of the reward was initiated that mattered. Intrinsic motivation decreased when rewards were not tied to the quality of the work, or when the person received fewer rewards than those promised. The notion is to avoid promising a reward after the child misbehaves to entice her to behave well. The promise of a reward can be a reward itself so by doing so they may inadvertently reinforce the child for behaving badly. Another point: don’t promise rewards that require a level of performance beyond the child’s ability.

Cameron and Pierce’s discovery that rewards increase performance on low-interest tasks, and simultaneously increase intrinsic motivation is an important finding for parents and teachers. Parents and teachers can, without feeling guilty, use external rewards to motivate children to learn important academic skills when the child has little initial interest, and by doing so make the activity more enjoyable.

Pay Kids To Read?

Robert Eisenberger, author of Learned Industriousness encourages tying rewards to demanding standards as one way to increase motivation for challenging tasks, such as reading and math. Researchers Stephen Flora and David Flora urge parents to pay children to read. They examined the reading habits of college students who participated in the ‘Book It’ reading program. Over 22 million children in Australia set reading goals and received money or coupons for pizzas if their goals were met. Children who were rewarded for reading in- increased the amount they read as children and maintained their interest in reading as college students. Dr. Flora states there was no evidence for “the myth that extrinsic rewards inhibit internal motivation”.

Sources

Cameron, J., and Pierce, W. D., Rewards and Intrinsic Motivation: Resolving the Controversy Bergin & Garvey 2002.

Cameron, J., Pierce, W. D. (2005) Banko, M and Gear, A., Achievement Based Rewards and Intrinsic Motivation: A Test of Cognitive Mediators. Journal of Educational Psychology vol 97 No. 4, 641-655.

Eisenberger, R. Learned Industriousness Psychological Review, 99 248-267

Flora, S.R. & Flora D.B., (1999) The Effects of Extrinsic Reinforcement for Reading during Childhood on Reported Reading Habits of College Students. The Psychological Record 49, 3-14.

I made a post today about the song, “Good Day”. It is a feel good song that will put you in a good mood. I thought a good mood was a good idea for today. However, one of the lines of the song is ‘nobody’s gonna die today’. Well, we have a war on. Easy for me to say no one’s gonna die sitting in California. So, given the war in Iran, me presenting the song in a post seems in bad taste. I thought I appeared insensitive. So, I deleted the post. My apologies.

Paul Coyne

A few weeks ago my wife and I had dinner with our son, who works in advertising. We joined Ben, a car designer and his wife Yunyun an appliance engineer. The topic of conversation was the future of AI and its effects on jobs. Will any of them have a job in 5 years? Who will have a job? What careers are available once their daughters graduate from college?

Open any news feed in early 2026, and you will see blaring headlines warning that Large Language Models (LLMs) and new “AI Agents” are going to eliminate massive numbers of white-collar jobs in the US within the next 12 to 24 months. Writers predict a looming wave of unemployment.

But is this 24-month apocalypse realistic, or is it just a worst-case scenario?

How Humans Learn vs How Machines “Learn”

As educators and parents, we are intimately familiar with how human beings acquire knowledge.

As discussed in a previous post, when a child learns the concept of a “dog,” we show her set of examples and non-examples. Learning is governed by consequences. When the child points to a dog and says “Dog!”, a parent provides praise (a reinforcer). When the child points to a cow and says “Dog!”, the parent provides a gentle correction (a punisher).

Through this history of reinforcement and punishment, the child learns to discriminate the relevant features of a dog (snout, fur, barking) from the irrelevant features (size, collar color). Eventually, the presence of a dog establishes stimulus control over the child’s verbal behavior. Cognitive scientists and AI researchers, such as Gary Marcus, describe this end result as building a robust, flexible concept—a “world model.”

Do AI Engineers Use Behavior Analytic Terminology?

Interestingly, yes and no. The computer scientists building LLMs rely heavily on a behavioral paradigm called Reinforcement Learning from Human Feedback (RLHF). During this phase, engineers explicitly use the terms reinforcement, reward (their term for a reinforcer), and penalty (their term for a punisher).

They essentially set up a digital operant conditioning chamber for the AI. The LLM generates a response, and a human grader provides a reward or a penalty. Over millions of trials, the AI learns which behaviors maximize its rewards.

However, computer scientists rarely use the term stimulus control. Instead, they talk about “context windows” and “attention mechanisms.” But functionally, when a specific text prompt reliably triggers a specific AI output due to a history of reinforcement, the AI is operating under a form of text-based stimulus control.

The contrast, as AI researcher Gary Marcus points out, is that the AI never actually abstracts a true “world model” from these consequences. A human experiences what a dog is. A human can see, hear, feel, smell a dog. An AI, acting as “autocomplete on steroids”, simply learns the statistical probability of the word “dog” appearing next to the word “bark.”

The Math Behind the AI

When you ask an LLM a question, it relies on three pillars of statistics to generate a response:

1. Factor Analysis (Word Embeddings) Computers cannot read text, so they must convert words into numbers. Using math very similar to factor analysis, the AI maps every word into a massive, multi-dimensional graph. Words with similar meanings cluster together. (Technically, embeddings arise from neural network training and not exactly factor analysis, but the math is quite similar)**

The AI also discovers “latent variables” (like plurality, tone, or geographical relation). This is how the AI understands the context of your sentence.

2. Logistic Regression (The Softmax Function) Once the AI understands the context, it has to pick the next word. It does this using the same math found in multinomial logistic regression. It assigns a raw score to every single word in its 50,000-word vocabulary, then runs those scores through an equation called a “Softmax function.” This converts the scores into percentages. The AI might calculate a 75% probability the next word is “apple,” a 15% chance it is “banana,” and a 0.001% chance it is “car.” (Softmax resembles multinomial logistic regression but is not literally the same thing).

3. Signal Detection Theory and ROC Curves How do we know if the AI is actually smart, or just blindly guessing? We test it exactly how we test human perception: using Signal Detection Theory (SDT).

When engineers test if an AI can successfully detect a toxic prompt without censoring a safe one, they are measuring its ability to detect a “signal” amidst the “noise.” They plot Receiver Operating Characteristic (ROC) curves to measure the AI’s true sensitivity independent of its programmed response bias. If the model confidently hallucinates a fake fact, that is a classic false alarm. By the way, I thought it was interesting to discover the use of SDT. Other methods are used, also.

Reality vs Hype

Given this incredible statistical power, will millions soon be out of work?

The “Doom” Camp Tech leaders like Mustafa Suleyman (CEO of Microsoft AI) have stated that most white-collar tasks will be fully automated within the next 12 to 18 months. Andrew Yang has similarly warned that millions of office jobs will evaporate."

Their argument hinges on the rise of “Agentic AI”, i.e, AI that doesn’t just chat, but independently executes multi-step operations. The fear is that as soon as one major company successfully replaces midlevel office workers (like junior coders or basic financial analysts) with an AI Agent, Wall Street will punish any competitor that doesn’t immediately do the same.

The “Transition” Camp. There are institutions looking at actual corporate adoption rates (job rates) that are not in a panic. Forrester’s latest 2026 AI Job Impact Forecast predicts that while AI will influence or augment 20% of jobs over the next five years, it will only permanently replace about 6% of U.S. jobs by 2030.

Forrester’s data reveals a fascinating trend called “AI Washing.” Many of the layoffs currently being blamed on AI are actually just standard, financially motivated corporate cuts. Executives use AI as an excuse to appease investors, even when they do not have mature AI applications ready to fill those roles. So says Forrester. Link to https://www.forrester.com/blogs/ai-and-automation-will-take-6-of-us-jobs-by-2030/

The Reality of Corporate Implementation

The consensus from the data, according to Daniel Guetta (professor at Columbia University), is that companies are finding it incredibly difficult to actually implement these AI systems at scale.

It is one thing for an AI to write a Python script in a vacuum; it is entirely different to safely integrate that Agentic AI into a complex corporate database. Because LLMs lack a world model, they hit a wall when faced with novelty. If a “rogue” AI agent hallucinates while accessing a bank’s legacy systems and automatically misroutes a massive financial transaction, it costs millions of dollars. “Before a company can actually turn on an advanced AI assistant, human workers have to spend months cleaning up the company’s old, messy computer systems. In fact, researchers have found that 80% of the work in an AI project isn’t about the AI at all—it is just human engineers acting as ‘data janitors’ to organize old files so the AI can read them.”

Robots.

There is the recent surge in humanoid robotics. Videos of robots walking factory floors or sorting packages make it seem as though physical labor is on the verge of disappearing overnight. In reality, most of these systems operate in tightly controlled pilot environments and require extensive human supervision, calibration, and maintenance. Moving from a staged demonstration to reliable, 24/7 deployment in messy real-world settings is enormously difficult and expensive. Just as with software-based AI agents, the problem is not capability in a lab but safe, accountable integration into complex human systems. In the near term, the rise of robots is more likely to reshape tasks and create new technical and supervisory roles than to trigger a sudden collapse of employment.

5-Years

So, how much unemployment will there actually be in the next five years due to AI?

I would be surprised if there was a sudden jobs apocalypse. Instead, I think we are entering a period of reskilling. I heard it said, that the transition will be similar to the typewriter to the computer. I can’t say that will be true. The roles at highest risk are those dedicated entirely to routine, high-volume digital production. The humans who thrive will have expertise in AI. They will move into roles focused on compliance, governance, and strategy. Perhaps, we will shift to an economy of managers and editors.

Job Loss vs Job Creation

When people hear about AI, they often picture a factory where a robot unplugs a human and takes their seat. The reality may not be that way.

Recent projections from the World Economic Forum estimate that while AI and automation will eliminate roughly 83 million jobs globally by 2027, the exact same technological shift will create 69 million new roles.

Hopefully, we are not looking at a net loss of jobs; we are looking at a shift in employment. The transition of the next five years will be driven by a mismatch of skills.

Roles at Risk

The roles facing the highest risk of elimination over the next 18 to 60 months are those built entirely around routine, high-volume digital production.

• Junior Coders: Writing basic, boilerplate code to build standard web pages.

• Entry-Level Analysts: Pulling numbers from one database, sorting them, and putting them into an Excel summary.

• Basic Content Creators: Writing standard product descriptions or routine corporate updates.

These roles are highly vulnerable because they rely on exactly what Large Language Models do best: statistical pattern matching. If a task does not require a conceptual understanding of the world, a machine can do it instantly.

Editor and Manager

If the machine is now the ultimate, tireless producer of raw material, the human workforce may pivot to becoming orchestrators, editors, and managers.

Because an AI does not have a “world model,” i.e., because it does not actually understand what a contract is, what a brand’s reputation means, or what it feels like to be a frustrated customer, it cannot be trusted to act without supervision. Humans will thrive in areas that require conceptual understanding and accountability:

• Strategy: An AI can generate fifty different marketing plans based on historical data. But only a human strategist understands the current cultural climate, human emotion, and the nuanced “vibe” of a target audience to select the one plan that will actually work.

• Governance and Compliance: As companies begin giving AI “hands” (the ability to process refunds, send emails, or draft legal briefs), the risk of a hallucination causing catastrophic financial or legal damage skyrockets. A massive new sector of jobs is emerging for humans whose sole responsibility is to audit AI outputs, ensure ethical compliance, and verify that the machine isn’t breaking the law.

• Managing Consequences: AI is a prediction engine; it does not understand consequences. If an AI approves a faulty mortgage application, the AI does not go to jail or face a PR crisis. Humans will always be required to hold the ultimate decision-making power and bear the responsibility for the outcomes.

Conclusion

AI can generate an endless stream of statistics and text, but it requires humans to provide supervision. We are the ones who must provide the constant feedback by reinforcing highly accurate outputs, punishing hallucinations, and shaping AI’s behavior so it aligns with reality.

I could be wrong, but my guess is AI will not lead to an unemployed citizenry living on an allowance provided by the government. ……At least not before 2030.

Paul Coyne PhD

Note:

**Both factor analysis and neural network training are trying to do the same fundamental thing: take a large, messy dataset with many variables and discover a smaller set of hidden underlying dimensions that explain the patterns in the data. In factor analysis, you might take 50 survey items and discover they reduce to 5 underlying factors. In a neural network learning word embeddings, you take a vocabulary of 50,000 words and the network discovers that words cluster along hidden dimensions, things like gender, tense, geography, or sentiment, without being told to look for them.

The key difference is in how each method finds those dimensions. Factor analysis uses a direct mathematical decomposition (eigenvalues, covariance matrices) in one elegant step. A neural network finds its hidden structure iteratively, through millions of small adjustments via backpropagation and gradient descent . It is less a calculation than a long process of trial and error guided by prediction error.

I used Gemini to research and find information on this topic. Reference links were inserted by Gemini.

Fun fact: My Dissertation was titled: A Signal Detection Analysis of Awareness.

Leave a comment

Subscribe now

Share

<iframe width=”560” height=”315” src=/__u/paulcoynephd.substack.com/&#8221;%3C/p>

title=”YouTube video player” frameborder=”0” allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share” referrerpolicy=”strict-origin-when-cross-origin” allowfullscreen></iframe>

<iframe width=”560” height=”315” src=/__u/paulcoynephd.substack.com/&#8221;%3C/p>

title=”YouTube video player” frameborder=”0” allow=”accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share” referrerpolicy=”strict-origin-when-cross-origin” allowfullscreen></iframe>

The NYTimes published a terrific article, “Childhood Trauma Doesn’t Have To Be a Lifelong Curse”, by Rochelle Sharpe, last Saturday January 31. The premise was straightforward. Adverse Childhood Experiences (ACE) cause trauma. Trauma can lead to cancer, heart disease, stroke, alcoholism and suicide. To mitigate the effects of trauma arrange Healthy Outcomes from Positive Experiences (HOPE), so says Robert Sege MD a pediatric researcher from Tufts Medical Center.

Vincent Felitti and Robert Anda studied Adverse Childhood Experiences (ACE) in 1998. The study changed how researchers understand the link between childhood trauma and adult health. What follows are the details on the original study, the HOPE intervention framework, and the involvement of behavior analysts in this field.

The 1998 ACE Study

The original study was a collaboration between the Centers for Disease Control and Prevention (CDC) and Kaiser Permanente‘s Health Appraisal Clinic in San Diego.

• The Researchers: The principal investigators were Dr. Vincent Felitti (Kaiser Permanente) and Dr. Robert Anda (CDC).

• The Research: From 1995 to 1997, they surveyed over 17,000 adult patients. The patients were asked about their current health and their childhood history of abuse, neglect, and household dysfunction (e.g., divorce, domestic violence, substance abuse).

• Key Findings: The study had two major points:

  1. Adverse Childhood Experiences are common: Nearly two-thirds of participants reported at least one ACE.

  2. The Dose-Response Relationship: There was a direct correlation between the number of ACEs a person experienced and their risk for negative adult outcomes. A higher “ACE Score” predicted heart disease, cancer, lung disease, depression, substance abuse, and attempted suicide decades later.

HOPE (Healthy Outcomes from Positive Experiences)

Dr. Robert Sege (Tufts Medical Center) and his colleagues recognized that the ACE study focused primarily on deficits and trauma. Sege asked: Why do some children with high ACE scores still thrive?

This led to the research on Positive Childhood Experiences (PCEs). Research published in JAMA Pediatrics (Bethell, Jones, Gombojav, Linkbach, & Sege, 2019) showed that positive experiences could mitigate the stress of ACEs. Positive experiences are not just “nice to have”, they are biologically necessary for resilience. HOPE identifies the protective factors that can mitigate the damage, say the researchers.

Researchers found that adults with high numbers of Adverse Childhood Experiences (high ACE scores) were significantly less likely to develop depression or mental health issues if they also had high levels of Positive Childhood Experiences (PCEs).

Positive experiences can buffer the effects of adversity. A child with high trauma and high positive support (high density of reinforcement) often has better outcomes than a child with low trauma but low support (neglect).

The Four Building Blocks of HOPE

Dr. Sege crystallized these positive experiences into four categories, known as the “Building Blocks.”

Relationships. Safe and supportive relationships with other adults and peers is a strong predictor of resilience. It includes having a parent who listens, but crucially, it also includes “other adults” (coaches, teachers, neighbors) who take a genuine interest in the child. According to Dr. Sege, these relationships stimulate the production of oxytocin, which directly counteracts the cortisol (stress hormone) spikes caused by ACEs.

Environment. Safe, stable, and equitable environments to live, learn, and play. This refers to the physical and emotional safety of a child’s surroundings—schools free from bullying, parks that are safe to play in, and a home with food security. When a child feels physically safe, they can move from “survival mode” (brainstem activity) to “learning mode” (prefrontal cortex activity).

Engagement. A sense of belonging and connectedness. This involves obtaining social reinforcement from family, friends, and acquaintances. It includes participation in sports teams, music groups, religious communities, or other groups. Engagement combats the isolation that often accompanies trauma. It provides a social identity that is positive (”I am a drummer,” “I am a scout”) rather than negative (”I am a victim”).

Emotional Growth. Opportunities to feel capable and confident. This involves “emotional intelligence.” It also includes the experience of having one’s views accepted rather than dismissed by adults. Children learn that difficult emotions are temporary and manageable, rather than catastrophic.

Other Researchers Treating Trauma based on ACEs

The original 1998 ACE study was the catalyst for the Trauma-Informed Care (TIC) movement. Other key researchers and clinicians who developed procedures to translate ACE findings into practice include:

• Dr. Nadine Burke Harris: A pediatrician who developed the PEARLS (Pediatric ACEs and Related Life-events Screener) tool. She pioneered the procedure of screening every child for ACEs during regular check-ups to intervene early.

• Cognitive Behavior Therapies: Therapies like Trauma-Focused Cognitive Behavioral Therapy (TF-CBT) and Parent-Child Interaction Therapy (PCIT) are used for treating children identified with high ACEs.

Are Behavior Analysts Involved In Trauma Research?

In the original 1998 study there were no behavior analysts involved. The original research team consisted primarily of physicians (preventive medicine, pediatrics) and epidemiologists.

In current research, Yes. For many years, there was a “silo” between trauma researchers (psychology/psychiatry) and behavior analysts. However, a sub-field called Trauma-Informed Behavior Analysis (TIBA) has emerged. Behavior analysts are now researching how ACEs affect behavior and how ABA (Applied Behavior Analysis) can be used to counter the effects.

Key Behavior Analysts & Researchers in this space:

• Dr. Camille Kolu: A Board Certified Behavior Analyst (BCBA-D) who leads Cusp Emergence. She is a primary figure in integrating the ACE study into behavioral engineering. She researches how to modify Functional Behavior Assessments (FBAs) to account for a child’s trauma history so that behavioral interventions don’t accidentally trigger “fight or flight” responses. Kolu treats trauma as a history of conditioning. She says that if we ignore the child’s “ACEs history,” we are ignoring the variables that turned, for example, a classroom into a threat. She moves the field from asking “What is the child getting out of this?” (Function) to “What happened to this child to make this situation feel dangerous?” (Context).

• Dr. Greg Hanley: While not exclusively an “ACEs researcher,” his work on “compassionate care” and “Today’s ABA” is cited in trauma-informed behavior analysis. His “Skills-Based Treatment” (SBT) prioritizes safety and trust before placing demands, which aligns directly with treating children with high ACE scores. He promotes enriched environments. Instead of a sterile “work” environment, the starting point should be a highly reinforcing context where the child is happy, relaxed, and engaged (HRE) before any demands are placed.

• Rajaraman et al. (2022): Published a seminal paper in the Journal of Applied Behavior Analysis titled “Trauma-informed care in applied behavior analysis,” formally introducing procedures for behavior analysts to use ACEs information in their practice. Rajaraman et al. successfully argued that trauma is Motivational Operation that fundamentally alters the value of reinforcers and punishers.

According to Rajaraman et al. (2022), treating a child traumatized, for example, by parental violence requires a shift in the function of the intervention. In traditional ABA, the goal might be “compliance to instructions.” In Rajaraman’s Trauma-Informed ABA, the primary goal is restoring safety and agency.

Here is how Dr. Rajaraman would likely construct a treatment program for a child exposed to physical trauma.

ABA Assessment: “Do No Harm” A traditional Functional Analysis (FA) might intentionally trigger the child (e.g., place the child in an environment where the child’s aggressive behavior is likely to occur) to observe the antecedents and consequences of the behavior. Rajaraman argues this is unethical for a trauma survivor because it may replicate the abuse. Instead, he would use an Interview-Informed Synthesized Contingency Analysis (IISCA).

• Rely heavily on interviews with safe caregivers to identify triggers.

• If they do run a trial, they do strictly open-door analyses. The child is never physically blocked from leaving. If the child shows precursor behaviors (like looking anxious or tensing up), the therapist immediately stops. They do not necessarily wait for hitting to occur.

The Context: “Happy, Relaxed, and Engaged” (HRE) Before any teaching happens, the behavior analyst must become a Conditioned Safety Signal. For a child abused by a parent, any adult figure can be a “Conditioned Aversive Stimulus.” The mere presence of an adult holding a clipboard or giving orders can trigger a fight-or-flight response. The sessions begin with Non-Contingent Reinforcement. The child is given free access to all their favorite items, and the therapist makes zero demands. The therapist’s only job is to show the child: “I am not a threat. I am a resource.”

The Core Intervention: Functional Communication Training (FCT) Rajaraman views “misbehavior” (aggression, screaming) as the child’s attempt to establish safety. The treatment replaces this dangerous behavior with a safe communication response, he calls the “Omnipotent Mand.”

• The Teaching Step: The child is taught a simple, powerful phrase, such as “My way, please,” “I need space,” or simply “Stop.”

• The Consequence: Every single time the child uses this phrase, the therapist immediately reinforces it by removing all demands and backing away.

• Why this works for Trauma: It teaches the child: “I don’t have to hit you to get what I want. I can speak, and you will listen.”

What is Strictly Forbidden For a child with a history of physical violence, Rajaraman’s framework explicitly advises avoiding procedures that mimic the trauma:

• No Escape Extinction: Never physically block the child from leaving the room.

• No Physical Management: Restraint is used only as a last resort for imminent life-safety, never as a “consequence” for behavior.

• No “Wait it Out”: Ignoring a child in distress (planned ignoring) is avoided, as it can mimic neglect.

Gradual Tolerance Training Only after the child trusts that they can turn off the pressure with their words (the “Omnipotent Mand”) does the therapist introduce small disappointments.

• Example: “I know you want to play outside, but let’s put these toys away first.” This teaches the child to tolerate mild stress in a safe, controlled environment, slowly building resilience without triggering the trauma response.

I don’t necessarily agree completely with Rajaraman. However, he is one of the few behavior analysts working in this field.

Conclusion

The integration of the ACE study with modern Applied Behavior Analysis represents an evolution in the field. It’s a straightforward premise. Mitigate adverse childhood experiences that cause trauma with positive childhood experiences. Provide love, provide HOPE.

This post was written by Paul Coyne PhD. Thanks for reading.

Leave a comment

Share

References & Links

• The Original Study: The CDC ACE Study Page

• The HOPE Framework: PositiveExperience.org (Dr. Sege’s Hub)

• Behavior Analysis & Trauma: Cusp Emergence (Dr. Camille Kolu’s Research)

Bethell, C., Jones, J., Gombojav, N., Linkenbach, J., & Sege, R. (2019). Positive childhood experiences and adult mental and relational health in a statewide sample: Associations across adverse childhood experiences levels. JAMA Pediatrics, 173(11), e193007. https://doi.org/10.1001/jamapediatrics.2019.3007

Felitti, V. J., Anda, R. F., Nordenberg, D., Williamson, D. F., Spitz, A. M., Edwards, V., Koss, M. P., & Marks, J. S. (1998). Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults: The Adverse Childhood Experiences (ACE) Study. American Journal of Preventive Medicine, 14(4), 245–258. https://doi.org/10.1016/S0749-3797(98)00017-8

Rajaraman, A., Austin, J. L., Gover, H. C., Cammilleri, A. P., Donnelly, D. R., & Hanley, G. P. (2022). Toward trauma-informed applications of behavior analysis. Journal of Applied Behavior Analysis, 55(1), 40–61. https://doi.org/10.1002/jaba.881

Sharpe, R. (2026, January 31). Childhood Trauma Doesn’t Have To Be a Lifelong Curse. The New York Times. https://www.nytimes.com/2026/01/31/headway/childhood-trauma-recovery-healing-research.html?searchResultPosition=1

Substance Abuse and Mental Health Services Administration. (2014). SAMHSA’s concept of trauma and guidance for a trauma-informed approach (HHS Publication No. SMA 14-4884). Substance Abuse and Mental Health Services Administration. https://store.samhsa.gov/sites/default/files/d7/priv/sma14-4884.pdf

Please note, I did not design their website. I did not write the text adjacent to the videos. I have not been involved with C&A since 2013, however, they still post my videos. I do not have permission to re-post the videos, or the related photos on my Substack page. So, to view the videos you must go to the Coyneandassociates.com website and look at the ‘Resources for ABA Professionals’ and ‘Resources for Parents’ webpages. Links are below.

There are 2 videos on this page: https://coyneandassociates.com/resources/for-parents/

One is on Natural Environment Teaching/Pivotal Response Teaching. Here the behavior therapist is teaching language skills while playing Caribou, a children’s board game.

The other video“Becoming a Listener” is about how to teach a difficult auditory visual discrimination using discrete trial teaching. This is one of my favorite videos. You can see the child learn the task right before your very eyes. I did not write the words, “Becoming a Listener” and related text. My text is incorporated within the video.

On this page: https://coyneandassociates.com/resources/for-aba-professionals/ there are 4 videos.

One is a video about a day in the life of a behavior therapist - that is not my video.

I created the other 3 videos on the page: Teaching Eye Contact, Matching to Sample, and Non-verbal imitation.

Generally speaking I prefer Pivotal Response Teaching (PRT) over Discrete Trial Teaching (DTT), but there is a place for both techniques.

I’m sorry I can’t show the videos here. However, if the subject is of interest to you the videos are worth the clicks to view them. If you see any errors, let me know.

Paul Coyne PhD

Leave a comment

In his course on Verbal Behavior**, Jack Michael related concept formation to Generic Tact Extension, that is the child says, ‘chair’, when he sees a novel chair, but not when he sees a stool or a ball. He explained that “learning a concept” is technically the process of bringing a verbal response under the control of specific properties of a stimulus (leg, seat, back) which are the critical features while extinguishing control by other properties (color, size, shape) which are irrelevant features). Concepts are often under multiple sources of control.

When Jack Michael discussed how to teach concepts in his course on Verbal Behavior he referenced Susan Markle and Philip Tiemann’s ‘Really Understanding Concepts” (1990) and their earlier work “Analyzing Instructional Content” (1976).

Let’s say you wanted to teach the abstract concept, “Red”. You would reinforce the verbal response, ‘red’ in the presence of red objects. Then you would vary all other properties, that is the irrelevant features, by presenting many different objects: a red apple, a red car, a red book, a red ball and so on and interspersing other objects that have no or some, but not all, of the relevant features (a multicolored ball, a white shirt with pink stripes). Consequently, the control exerted by the irrelevant features is extinguished because they are not present when reinforcement occurs. The only constant feature across all examples is the critical feature (redness) which eventually gains control over the verbal response, red.

The Question

So, the question arises should you tell the student what the concept’s critical features are before you present relevant and irrelevant examples? Will he master the concept better if he has a chance to figure out the defining features on his own or is it better to tell the student the essential features upfront, and then present the examples?

Discovery Learning

In the 1960s, Jerome Bruner was a proponent of Discovery Learning. He argued that if you give a student the rule immediately, they become passive recipients of information. He suggested that the teacher provide a set of examples and non examples but withhold the rule. Burner felt that the student must act like a scientist, hypothesizing and testing to ‘discover’ the defining features of a concept. He argued the students would have better retention if they figured it out themselves. They would acquire better problem solving skills.

Hilda Taba developed a specific teaching protocol based on presenting examples first. The teacher presents labeled examples (Yes) and non examples (No). The teacher does not define the concept. Students compare the Yes and No columns and generate their own hypotheses. The class discusses and agrees on a definition of the rule at the end. This procedure is used to teach the concept and to teach inductive reasoning skills.

What if there is no rule?

Eleanor Rosch argued that for many concepts, there is no rule. Rules work great for squares and electrons where clear rules exist. She says it fails for concepts such as Bird, Game, or Furniture. Rosch argued to not check a list of critical attributes. Instead, the student should compare the new object to a mental prototype, the average best example. For example, a robin is the prototype bird. Don’t bother with a strict definition, just show the best example first, then show variations that get further away. The student absorbs the category ‘fuzzy boundaries” without ever needing a hard rule, so says Rosch.

Skinner on the other hand argues that you can teach pigeons and people the concept of human, for example, purely through differential reinforcement, without telling them what the rule is. However, for humans with language, Skinner agrees that telling the student the rule vastly speeds up the process.

Markle & Tiemann

Markle and Tiemann favor giving the rule first. In Really Understanding Concepts they classify instruction sequences into two main types: 1) state the definition then show the examples; 2) Show examples and have the student guess the definition. Their reasoning is: the rule acts as a verbal prompt that orients the learners attention to the critical attributes. Without the rule the student has to look at a chair, a stool, and a sofa and blind guess which features matter. They might guess Legs are critical. If the teacher says the rule is, ‘it must have a back, a seat, and at least 1 leg.’ Now, when the student looks at the stool they immediately look for the back, leg and seat. Learning is faster. If you don’t provide a rule, students may form mistaken rules. For example, if you show a student 3 wooden chairs and no rule is given the student might learn an incorrect rule, chairs are made of wood. Giving the rule, material is irrelevant, may prevent this error before it occurs.

An Exception

There is one specific exception where Markle and Tiemann might suggest not giving a rule: when the concept cannot be defined by words. For example, the color red or the sound of a clarinet. You cannot easily write a rule for red that a child would understand. In such cases they suggest you present examples first because the verbal rule is inaccessible to the student.

Williams & Roop

Williams and Roop (2025) investigated whether is was better to provide the rules to adult learners first when teaching concepts or whether to provide a Discovery Learning opportunity by presenting examples and non examples of the concept without offering the rule that described the essential defining features. They looked for generalization, that is responding correctly to new examples of the concept that were not explicitly taught. They looked, also, to see if adults withheld the verbal response or responded differently to new non examples.

The researchers used an artificial concept created specifically for the study to ensure no participant had seen it before. The rule provided described the concept’s boundaries using two parts:

1. It listed the Must-Have Features (e.g., ‘It must have a specific shape in the center, a dashed border, and three distinct lines’).

2. It listed the Can-Have Features (e.g., ‘The color, overall size, and background do not matter’).

Participants were then shown examples that matched this rule and nonexamples that violated exactly one of the must-have features.

They ran two experiments. The first was designed to see if receiving the defining characteristics of the concept before starting practice eliminated errors and taught the concept. Participants received the instructional package (rule+examples+non-examples) to assess learning concepts.

The second experiment was designed to see if the adult participants could learn the concept through practice trials alone. If not, would adding the rule later result in learning the concept. Participants started first with the practice phase, if they failed to reach mastery (which they failed to do), they were then given the whole instructional package rules included, followed by more practice and testing.

Results were as follows: Practice alone (differential reinforcement with no rules) was insufficient in teaching adult learners the concept. Participants failed to acquire the concept. They could not reliably determine the essential features that defined the concept. On the other hand, the instructional package (rule+examples+non examples) produced rapid and reliable learning of the concept.

In experiment 1 participants who received the rules first performed with near perfect accuracy almost immediately. They passed the concept test with high scores. In experiment 2 participants struggled during the initial practice phase, however as soon as the instructional package was introduced, their performance improved dramatically.

Also noted was the finding that the order didn’t matter. Whether you have the learner practice first with no rule, then provide the rule later with additional practice or provide the rule first and then practice the student will learn the concept. However, considerable time is saved by teaching the essential ingredients rule first.

Conclusion

If you are teaching a concept provide a description of the essential characteristics before you provide correct and incorrect examples of the concept.

**Actually I double checked my college notes and a more recent copy of Michael’s materials for his Verbal behavior class and Susan Markle and Philip Tiemann’s ‘Really Understanding Concepts” (1990) and their earlier work “Analyzing Instructional Content” (1976) were a part of his Technology of Teaching course, although he probably talked about this same issue in both classes.

Thanks for reading. I appreciate your comments.

This post was written by Paul Coyne

References

1. Michael, J. L. (2004). Concepts and Principles of Behavior Analysis. Kalamazoo, MI: Society for the Advancement of Behavior Analysis. Link to Book

2. Markle, S. M., & Tiemann, P. W. (1990). Really Understanding Concepts. Champaign, IL: Stipes Publishing. Link to Google Books

3. Bruner, J. S. (1961). “The act of discovery”. Harvard Educational Review, 31, 21–32. Link to Article Record

4. Rosch, E. (1973). “Natural categories”. Cognitive Psychology, 4(3), 328-350. Link to Article

5. Williams, C. L., & Roop, J. C. (2025). “Instruction consisting of a rule and set of examples and nonexamples reliably teaches concepts”. Journal of the Experimental Analysis of Behavior, 124(3). Link to JEAB Journal

Paul Coyne

Critics of fluency building say it’s drill and kill. It ruins motivation. It’s boring. It’s too hard. It’s too competitive. Kids hate it.

The reality is another story.

Building fluency is NOT about racing or “going fast.” It’s about going as quickly as needed. As we say at Morningside, “it’s a pace, not a race.”

Good fluency practice builds motivation and confidence as students find daily success. They learn to make dramatic progress through small, daily improvements.

High quality deliberate practice teaches self-monitoring, as students record and analyze their own learning data. It teaches self-advocacy skills as students learn when, why, and how to get support if they’re not making progress. It also teaches peer coaching, as students encourage each other to make progress.

And it doesn’t take long. With just 10-15 minutes a day, students can make remarkable outcomes like these:

Sign up for The Morningside Professional Development Series: Building Math Fluency to learn the curriculum and techniques our classroom teachers use to produce these life-changing outcomes!

• February 2-4, 2026 at our laboratory school in Seattle, WA!

• $800 l 12 BCAB CEUS available for $150

Seats are limited! Sign up today!

Copyright © 2025 Morningside Academy, All rights reserved.
Thanks for being a friend of Morningside

Our mailing address is:

Morningside Academy

901 Lenora St.

Seattle, Wa 98121

Add us to your address book

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.

Lindsay Unified School District (LUSD) integrates students with special needs into its Performance-Based System (PBS) through a model of full inclusion.

Because the entire district already operates on a learner-centered Strategic Design, the gap between a “General Education” student and a “Special Education” student is structural. In a traditional school, a 5th grader reading at a 2nd-grade level stands out or is removed from class. In Lindsay, that student is simply working on “Level 2” content, just like other learners who might be younger or older, without the stigma of failing.

Here is how the system functions for children with special needs:

Age vs. Ability

In the PBS model, students are grouped by performance level, not strictly by age. This is the single biggest advantage for special education.

• Traditional Model: A student with a learning disability is forced into a 6th-grade math class because they are 11 years old, even if they don’t understand multiplication. They require an aide or a “pull-out” session.

• Lindsay Model: That same student is placed in a math group working on multiplication (Level 4/5). They are surrounded by peers working on the same content. The instruction matches their ‘Zone of Proximal Development’.

Lindsay Unified explicitly builds its instructional model on Vygotsky’s Zone of Proximal Development (ZPD). While they refer to this as the ‘just-right level’ when speaking with students, their internal teacher training documents identify ZPD as the core mechanism for ensuring appropriate academic challenge.

IEPs and Playlists

LUSD uses a Learning Management System (LMS) called Empower. This software allows teachers (Learning Facilitators) to “hide” the differences in assignments between students.

• Customized Playlists: Every student logs into a laptop to see their “playlist” of work. A Special Education student’s playlist is modified according to their Individualized Education Program (IEP).

• Modifications: If an IEP calls for a modification, such as: shortened assignments or visual aids, the Facilitator builds those directly into the student’s digital playlist. To an outside observer, the student looks exactly like their neighbor, quietly working on a laptop, even if the content is vastly different.

• Accommodations: Since the system is self-paced, the common accommodation of extended time is automatic. A student does not need to ask for extra time; the system grants it by default until mastery is shown.

Inclusion

Lindsay pushes for full inclusion, meaning special needs students spend the vast majority of their time in general learning environments (classrooms).

• Targeted Small Groups: Instead of lecturing to 30 kids, a Facilitator pulls small groups of 4–6 learners to a table for a mini-lesson. A Special Education teacher (or Learning Facilitator for Specialized Academic Instruction) goes into the room to run these small groups or co-teach, rather than pulling students out to a separate trailer or room.

• Social Emotional Support: Because “being behind” is normalized (everyone is “behind” or “ahead” in something), the social stigma often associated with Special Education is reduced.

Special Day Classes

For students with severe cognitive or physical disabilities who cannot benefit from a general education classroom even with modifications, Lindsay utilizes Special Day Classes (SDC).

• Location: These classrooms are located on regular school campuses (such as Washington Elementary) rather than in isolated separate schools. This allows students to remain part of the “Learning Community” and participate in school-wide events, lunch, and PE with general education peers (mainstreaming).

• Partnership: Lindsay partners with the Tulare County Office of Education (TCOE) to operate some of these programs. TCOE provides specialized staff, equipment, and curriculum for “low incidence” disabilities (e.g., severe orthopedic impairment, deafness/blindness, or severe cognitive delays) that a small rural district cannot support alone.

Functional Academics

In the Performance-Based System (PBS), “mastery” is redefined for these learners. Instead of mastering Algebra, their “Empower” playlist is built around Functional Academics and Life Skills.

Adapted Goals: A student’s “Standards” might include:

• Independent Living: Toileting, feeding, safety signs, and money management.

• Communication: Using assistive technology (like an iPad with Proloquo2Go) to express needs.

• Vocational Skills: Sorting, organizing, or simple tasks to prepare for employment.

Tracking: These goals are tracked in the system just like academic standards. A student turns “Green” (Proficient) on “Tying Shoes” or “Identifying Exit Signs,” allowing them to celebrate mastery just like their peers mastering History or Math.

The Adult Transition Program (Ages 18–22)

For severely handicapped students, education does not end at grade 12. LUSD (in coordination with the County) offers an Adult Transition Program (ATP).

• Focus: The goal shifts entirely from “School” to “Life.” Students learn how to take public transit, shop for groceries, cook, and perform job tasks.

• Placement: This program bridges the gap between high school and adult regional center services, ensuring students don’t “fall off a cliff” after graduation.

Summary of Service Levels

• Mild/Moderate: Full Inclusion (General Ed Class). Standard Curriculum (State Standards) with accommodations.

• Moderate/Severe: Special Day Class (SDC). Functional Academics (Life Skills, modified standards) focusing on independence.

• Low Incidence (e.g., Medically Fragile): County Program (TCOE). Highly specialized care (medical/physical) provided by County staff on or near district sites.

Thank you for reading. Comments are appreciated.

For more background on the district’s transformation, see the Aurora Institute’s case study on LUSD.

This is the story of Tom Rooney and the Lindsay Unified School District (LUSD), a rural district in California’s Central Valley serving students from pre-kindergarten through high school. Mr. Rooney transformed Lindsay, a traditional school district, into a self-paced program similar to Fred Keller’s “Personalized System of Instruction” (PSI) aka “Competency-Based Education” (CBE) in other circles. Unlike Keller, who used PSI to organize his college classroom, Tom Rooney used a similar Performance-Based System (PBS) to organize an entire district, pre-K through high school. Subsequently, student academic performance improved. Teachers became better teachers. Flowers bloomed. Here is how he did it.

The Challenge

In the early 2000s, Lindsay Unified faced challenges that seemed insurmountable. Located in an agricultural community, the district served a population that was 100% Title I (low-income) and roughly 93–96% Hispanic/Latino. Many students were children of transient migrant field workers. HubSpot+1

By 2006, the district was in crisis. Academic performance was low, with only about 26% of students proficient in state standards. The graduation rate hovered around 69%, meaning nearly one in three students dropped out. Perhaps most alarming were the social indicators: gang affiliation among students was estimated at about 14%, teen pregnancy rates were high, and staff turnover was rampant. The prevailing view was that the students could not achieve high academic performance because of their difficult backgrounds and circumstances. U.S. Department of Education+1

The Denver Trip

In 2007, Tom Rooney took the Lindsay Unified leadership team to Colorado. It was not a formal conference hosted by a specific organization, but rather a self-initiated investigative tour to visit high-performing schools in the Denver area that were succeeding with high-poverty populations.

The “eureka moment” occurred during a debriefing session held after visiting several schools. The team realized that while they saw excellent teaching, the schools were still operating under a traditional time-based system. They concluded that this structure inherently ensured a percentage of their own high-needs learners to fail because it moved them along before they had mastered the material.

While the problem was identified in Denver, the solution came from looking for districts that had successfully used a self-paced system. This search led them to the Chugach School District in Alaska and the Reinventing Schools Coalition (RISC).

Chugach School District (Alaska). This is the true pioneer. In 1994, under Superintendent Richard DeLorenzo, Chugach became the first K-12 district in the U.S. to dismantle grade levels in favor of a self-paced, mastery based system. Aurora Institute+2Education Reimagined+2

RISC. Richard DeLorenzo founded the Reinventing Schools Coalition (RISC) to teach this model to others. LUSD partnered directly with RISC to design their Performance-Based System (PBS). The framework Rooney used was largely adapted from the RISC model. Amazon+1

As you can see, LUSD is not the only district to use a self-paced system. Other notable districts that have used similar systems include:

• Chugach School District (Alaska): The originator of the model. Aurora Institute+1

• Westminster Public Schools (formerly Adams County School District 50) in Colorado: Often cited alongside Lindsay. They implemented a similar Competency-Based Education (CBE) model across their district around 2009 and also faced the challenge of translating a mastery-based system to state standardized testing. ERIC

• Kennebec Intra-District Schools RSU 2 (Maine): A district that has been a long-time leader in learner-centered, mastery-based education. Aurora Institute

• Building 21 (Philadelphia/Allentown): A network of public high schools operating entirely on a competency-based model. Aurora Institute

Lindsay serves as a demonstration site for other school districts. Through their Lindsay Leads program, they actively host visitors (including superintendents and teachers) to show them how to implement this system in a standard public school district. Lindsay Unified School District+2Amazon Web Services, Inc.+2

The Transformation

Tom Rooney and his team spent two years gathering input from parents, students, community leaders, and staff before they began their Performance-Based System. Lindsay Unified School District+1

The result was the Lindsay Strategic Design, a community-authored plan for a new kind of school district. It established a set of ideals, primarily that the district would be “uncompromisingly learner-centered.”

To signal this cultural shift, they changed their lexicon:

• Students became “Learners” (emphasizing active agency).

• Teachers became “Learning Facilitators” (emphasizing guidance).

• Schools became “Learning Communities.”

• Grades (A–F) were replaced by “Performance Levels” (1–4).

The Implementation

In their Performance Based System (PBS) system, learners progress only when they demonstrate mastery. If a learner needs three weeks to master a concept that takes another student three days, they get that time. If they can master it in one day, they move on immediately. Lindsay Unified School District+1

This required a massive structural overhaul:

Guaranteed curriculum. The district determined what mastery looked like for every standard. Learners knew exactly what they had to prove to advance. The curriculum was online and available to students 24 hours a day. Amazon Web Services, Inc.+1

Computer equipment. The district implemented community-wide Wi-Fi and provided laptop computers (and other equipment) to all students, ensuring that the poorest families had the same access to learning tools as the wealthiest.

Data collection. As computer equipment was rolled out, learners began tracking their own data. Even elementary-aged children could look at a dashboard, see where they were “red” (not yet proficient) or “green” (proficient), and set their own goals for the week.

Curriculum

All curricula are online.

LUSD uses Lexia Core5 Reading to provide diagnostic placement and phonics-based instruction. Lexia Core5 is a structured, adaptive curriculum that delivers explicit instruction across the core components of reading as identified by the National Reading Panel (2000): Lexia+2Lexia Learning Resources+2

• Phonological awareness. The program explicitly teaches learners to identify and manipulate the sounds (phonemes) in spoken words.

• Phonics. It provides systematic and explicit instruction on the relationship between sounds and letters to develop decoding skills.

• Fluency. It offers timed reading and repeated practice to help students read accurately and quickly, a key component of fluency.

• Vocabulary. It embeds direct vocabulary instruction and strategies to infer meaning from context.

• Comprehension. It teaches specific comprehension strategies, such as inferring, monitoring, and summarizing, especially at later levels.

Conversely, resources like Reading Plus are deployed once decoding is secure, focusing on improving fluency, vocabulary, and comprehension. Digital Promise+2ESSA Evidence+2

DreamBox Reading Plus

Reading Plus is a digital reading intervention program designed primarily to build reading fluency, and comprehension. Unlike programs focused on beginning reading skills, Reading Plus targets intermediate and advanced learners who have already mastered basic decoding but struggle with reading rate, stamina, or understanding complex text. ESSA Evidence+2Reading Plus+2

Reading Plus offers adaptive, personalized instruction based on an initial assessment (InSight) that places the learner in the curriculum and then adjusts reading materials and pacing as students progress. ESSA Evidence+1 It has been rated “Strong” under ESSA by Evidence for ESSA. ESSA Evidence+2Reading Plus+2

In 2021, DreamBox Learning acquired Reading Plus (along with Squiggle Park), and the program is now marketed as part of the DreamBox Reading suite—often referred to as DreamBox Reading Plus. Reading Plus+4Marketbrief+4Business Wire+4

Math

Although LUSD’s overall model is competency-based and individualized, they do use (or have used) a number of software-based and adoption-approved math curricula. On their own and in external profiles, the district has named several math-related programs, including: Lindsay Unified School District+2U.S. Department of Education+2

• enVision Math

• i-Ready Math

• Math 180

• Fraction Nation

They also list other supplemental or intervention resources (some not strictly math), but the above are explicitly math-oriented.

DreamBox Math

DreamBox Math is a digital, adaptive math program used by many elementary and middle schools to provide personalized instruction. It is primarily known for its ability to individualize the program of instruction for each student in real time. DreamBox Math is an adaptive K–8, standards-aligned program that combines curriculum and continuous formative assessment. the Learning Counsel+3DreamBox by Discovery Education+3DreamBox by Discovery Education+3

LUSD has used DreamBox Math; it has been described in case studies and district profiles as part of their blended-learning toolkit. However, it is not entirely clear from public documents whether DreamBox Math is still in active use in every school or whether other programs have replaced or supplemented it in some grade bands. Because the district currently uses DreamBox Reading (Reading Plus) in its literacy work, and recent case studies explicitly reference “DreamBox for math,” for present purposes I will assume they still make use of DreamBox Math as one of several available math tools. Digital Promise+2Discovery Education+2

DreamBox uses software to mimic a highly skilled human tutor. Its core function is to assess a student’s actions and adjust the curriculum instantly:

• Continuous assessment. As a student interacts with lessons, DreamBox tracks every click, answer, and time spent on a problem. It gathers information on the final answer and the strategy the student used. the Learning Counsel+3DreamBox by Discovery Education+3DreamBox Learning+3

• Real-time adaptation. Based on this data, the program adapts. If a student solves a problem quickly but shows they do not yet grasp the concept completely, DreamBox will immediately introduce a different, more remedial lesson. If a student masters a skill, the program bypasses remediation and moves them to more challenging content.

• Visual modeling. Lessons often use virtual manipulatives and interactive tools to help students explore mathematical concepts visually.

Results

The transition was difficult. It required retraining staff, educating parents, and overcoming the fear of the unknown. But the PBS approach paid off. Public Policy Institute of California+1

During the first 10 years:

• Academic gains. Proficiency rates nearly doubled, moving from roughly 26% to the mid-40% range in English Language Arts by around 2019, outpacing peer districts serving comparable populations. Aurora Institute+1

• Graduation rates. The graduation rate soared from about 69% to well over 90%, with district-level rates now in the low- to mid-90s and the comprehensive high school often in the high 90s. CDE California+2California School Dashboard+2

• College success. College-going rates increased, with more graduates attending and persisting in four-year universities; some reports note that four-year college entrance rates are about twice the national average for low-income communities. Reading Plus+1

• Culture. School climate improved markedly. Recent needs-assessment data show very low rates of bullying relative to state norms and dramatic reductions in gang affiliation (from about 14% in the early 2000s to roughly 3% today when combining “gang member” and “gang affiliated” categories). Suspension rates for misbehavior have also fallen. CDE Unified Storage West+2Aurora Institute+2

When the COVID-19 pandemic hit in 2020, while other districts scrambled to solve remote learning, Lindsay learners barely missed a beat. They already had laptops, connectivity to the internet, and access to the curriculum. The self-pacing feature of their PBS system was essential in keeping the schools operational. U.S. Department of Education+1

Performance Data (2024–2025)

The most recent data available come from the 2024 California School Dashboard and the 2025 CAASPP (state testing) results, supplemented by public summaries. The picture is mixed, showing strong graduation and climate outcomes but ongoing challenges in standardized test proficiency, typical of post-pandemic trends in high-poverty districts. CDE California+2Lost Coast Outpost+2

Graduation Rate

• 2024 Dashboard (district). The Graduation Rate indicator for Lindsay Unified is rated Green (High, Increased), with a district-wide graduation rate of 93.5%, up from 90.5% the prior year. CDE California

• Lindsay Senior High School. At the high school level, the graduation indicator is Blue (Very High), with rates around 97–98%, well above the state average. CDE California+1

• Dropout rate. Recent needs-assessment data indicate an overall dropout rate of about 1.6%, and approximately 1.3% for socioeconomically disadvantaged learners, placing Lindsay among the top performers in the state for student retention. CDE Unified Storage West+1

Academic Proficiency

• English Language Arts (ELA).

  • 2025 results. Approximately 41.8% of students met or exceeded standards on CAASPP ELA, an increase of nearly 3 percentage points from 2024. Lost Coast Outpost

  • Dashboard status. Depending on the specific school, ELA is generally rated Yellow (medium) or Orange (low), but with upward movement in many cases (e.g., one elementary school gaining more than 10 points). California School Dashboard

• Mathematics.

  • 2025 results. Approximately 31.7% of students met or exceeded standards in mathematics (reported as 31.74% in EdSource’s summary), an increase of about 4–5 percentage points from 2024. Lost Coast Outpost

  • Dashboard status. Math is generally rated Yellow or Orange, indicating performance below the state standard but with meaningful year-over-year gains. California School Dashboard+1

• English learner progress. Many schools are rated Blue (Very High) or Green (High) on English Learner Progress, indicating strong growth for students learning English. California School Dashboard+1

College & Career Readiness

On the 2024 Dashboard, the College/Career indicator for Lindsay Senior High is rated Green (High), with roughly 60–65% of graduates deemed “prepared,” representing gains from prior years. California School Dashboard+2Aurora Institute+2

Consumer Feedback

Feedback from students, teachers, and parents regarding the PBS model is largely positive, particularly concerning the internal culture. Learners often report appreciation for the self-pacing and the removal of traditional letter grades, perceiving the system as motivating.

Learning Facilitators (teachers), who initially faced significant change, generally support the model, finding their roles more focused on individual teaching and coaching. LUSD also offers individual tutoring.

Parents express confidence in the straightforward approach and the resulting high learner retention, which is reflected in the high graduation rates. Aurora Institute+3U.S. Department of Education+3Digital Promise+3

With All This Reform, Why Are Lindsay’s Scores Still Below the State Average?

There are three main reasons why Lindsay Unified’s overall scores remain below the state average, despite the success of their mastery-based system:

1. Demographic & Socioeconomic Context

Poverty and migrancy. LUSD is a high-needs district, historically serving a population that is 100% socioeconomically disadvantaged and over 90% Hispanic/Latino. Many learners are from migrant families working in the surrounding fields. These challenges lead to economic hardship that often create significant academic gaps before a child even enters school. HubSpot+2Aurora Institute+2

English learners (EL). The district has a very high percentage of Spanish speaking students who are learning English (around one-third to 40%). State data consistently show that the test scores of EL students are substantially lower than the scores of native English speakers across the state, regardless of the teaching model. Aurora Institute+2CDE California+2

2. The Mismatch Between System and Test

California law requires students to take the state test (CAASPP) based on their age-appropriate grade level, not the content level they are currently working on in the mastery system.

Example: A 13-year-old student might have entered middle school several years behind their peers and is currently achieving mastery at a sixth-grade content level. On the state test, that learner is forced to take the eighth-grade exam. Since they have not been taught the eighth-grade material, they score non-proficient, pulling the district average down. Lost Coast Outpost+2California Department of Education+2

3. The Implementation Timeline

Low baseline. When LUSD began its transformation in the late 2000s, only about 25–26% of students were proficient in reading. The gains they have made (for example, nearly doubling proficiency rates in some subjects) are significant within the context of the district’s history and compared to similar high-poverty districts. Aurora Institute+1

Time lag. For a child who is four years behind in reading, it might take the mastery system three or four years to get them working on grade-level material. The standardized test only captures a snapshot, and scores will remain low until a critical mass of students has closed those prior gaps and is working proficiently at their age-appropriate content level.

Aurora Institute+3CDE California+3California School Dashboard+3

Conclusion

Tom Rooney used a self-paced competency based structure and solid curriculum to transform a poor, disadvantaged school district into one that serves students, teachers, and the community.

Imagine what would happen if your local school district, perhaps a district that is not poor, rural and not primarily serving children of migrant workers, adopted a self-paced, competency-based system of instruction. Would that please you?

Leave a comment

Share

Subscribe now

________________________

Thanks for reading. I used ChatGPT & Gemini to search for information. I sorted the information, compiled it into an article, wrote, edited and re-wrote the article several times. ChatGPT checked the content for accuracy, grammar and added the references. With references the post is too long for email. So, I removed the references in the emailed version, however the version with the references is posted on Paul’s Substack of Useful Information website.

References

Aurora Institute. (2016). Chugach School District: A personalized, performance-based system—Insights from the field. Aurora Institute.

Aurora Institute. (2019). Strong evidence of competency-based education’s effectiveness from Lindsay Unified School District. Aurora Institute.

Aurora Institute. (2025, March 11). The latest milestones and lessons from Lindsay Unified. Aurora Institute. Aurora Institute

Barbara Aved Associates. (2023). Lindsay Unified School District community needs assessment. Lindsay Unified School District. CDE Unified Storage West

DeLorenzo, R. A., Battino, W. J., Schreiber, F. J., & Gaddy, B. B. (2009). Delivering on the promise: The education revolution. Solution Tree. Amazon+1

Digital Promise. (2024, November 13). Behind the scenes: A look inside Lindsay Unified’s learner-led learning communities. Digital Promise. Digital Promise

Discovery Education. (2022). DreamBox Reading rated strong by Evidence for ESSA [Research brief]. Discovery Education. Discovery Education+1

Discovery Education. (2022). Lindsay Unified School District: Empowering and engaging students with DreamBox Reading [Success story]. Discovery Education. Discovery Education

Evidence for ESSA. (n.d.). Reading Plus. Center for Research and Reform in Education, Johns Hopkins University. ESSA Evidence+2Reading Plus+2

Evidence for ESSA. (n.d.). DreamBox Learning Math. Center for Research and Reform in Education, Johns Hopkins University. ESSA Evidence

Fullan, M., & Youth Thriving Through Learning Fund. (2025). Lindsay Unified School District: A new vision for adolescent learning and well-being [Report]. Motion Leadership

Lexia Learning. (2023). Core5 and the science of reading [White paper]. Lexia Learning. Lexia+2Lexia+2

National Institute of Child Health and Human Development. (2000). Report of the National Reading Panel: Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction (NIH Publication No. 00-4769). U.S. Government Printing Office. Lexia+2Lexia Learning Resources+2

Virginia Department of Education. (2023). Intervention instructional program guide & supplemental instructional program guide. Virginia Board of Education. Virginia Department of Education+1

Virginia Department of Education. (2024, January 29). Virginia adds to the resources for its ALL In tutoring reading initiative to include Lexia Core5 Reading and Lexia PowerUp Literacy [News release]. eSchool News

California Department of Education. (2024). California School Dashboard: Graduation rate indicator—2024 detailed data, Lindsay Unified [Data set]. CDE California+2California School Dashboard+2

EdSource / Lost Coast Outpost. (2025, October 10). Eureka City Schools is about to do away with grades and embrace competency-based learning [Article including CAASPP graphic for Lindsay Unified]. Lost Coast Outpost

Lindsay Unified School District. (2007). Lindsay Unified School District strategic design [Slide deck]. Amazon Web Services, Inc.+2Learning Accelerator+2

Lindsay Unified School District. (2019). Lindsay Unified School District – Narrative [Federal application narrative]. U.S. Department of Education. U.S. Department of Education

Reading Plus. (n.d.). Research & results. Reading Plus. Reading Plus+1

Reading Plus. (n.d.). Effective reading intervention program [Program description]. Reading Plus. Reading Plus+1

Sterling Partners / Avathon Capital. (2021, August 18). Education Fund sells developer of popular adaptive literacy intervention [Press release on DreamBox acquisition of Reading Plus]. Sterling Partners+2PE Hub+2

DreamBox Learning. (2021, July 19). DreamBox Learning enters ELA market with acquisitions of Reading Plus and Squiggle Park [Press release]. BusinessWire. Business Wire+2Entrevestor+2

DreamBox Learning. (2025). How does DreamBox Math work? DreamBox Learning Help Center. the Learning Counsel+3DreamBox Learning+3DreamBox by Discovery Education+3

Youth Thriving Through Learning Fund. (2022). Driving student outcomes through competency-based learning: Lindsay Unified case study [PDF case study]. HubSpot

The term “fluency” is buzzing in Math circles. But what does it mean? And more importantly, how do we develop it with learners?

Join us at Morningside Academy’s lab school to find out how we’ve been building fluency for 45 years!

February 2-4, 2026, in-person at our laboratory school in Seattle!

Seats are limited! Sign up today!

Share

As you may know, before I retired my wife, Mary Coyne BCBA, and I owned Coyne & Associates. Coyne provided in-home ABA services to children with autism and their families throughout Southern California. One day a private-equity firm (PE) offered to buy the company. At the time, I didn’t know what private equity was or how it was different from venture capital (VC). I wondered, why they were interested in my mom-and-pop ABA program? Since I was already negotiating to sell the company to several employees, the associates, I did not pursue the PE offer. But I always wondered what would have happened if I sold the company to a PE firm instead. Permit me to tell you what I learned.

VC vs PE—what’s the difference? Venture capital typically buys minority stakes in young companies (often pre-profit), accepts high failure risk, and doesn’t control day-to-day operations. Private equity firms typically acquire controlling stakes in mature businesses (often using debt), reshape operations and strategy, and then exit after several years. Growth equity is in the middle: minority stakes in more established, fast-growing companies with little or no debt. ABA companies like BlueSprig/Trumpet are Private Equity firms.

In health care broadly, there’s empirical and case-study evidence that some PE acquisitions are associated with worse clinical outcomes and financial distress. A recent review in Health Affairs summarizes the research results as mixed overall. healthaffairs.org+2healthaffairs.org+2 They flag multiple studies linking PE ownership to poor quality in certain sectors. The AMA’s Journal of Ethics reached a similar “it depends, but there are real risks” conclusion.

Here are some concrete examples where cost-cutting and asset moves were followed by decline or bankruptcy. Three widely discussed cases:

HCR ManorCare (nursing homes). Carlyle bought the chain in 2007, then executed a leaseback of its real estate to a REIT, turning the operator into a rent-paying tenant. Over time, violations rose, finances deteriorated, and ManorCare ultimately entered bankruptcy in 2018. Nonprofit ProMedica acquired the operations out of court. Critics argue the real-estate deal saddled the company with unsustainable fixed costs that made quality cuts likely.

Hahnemann University Hospital (Philadelphia). A PE owner separated the hospital’s real estate from operations. When losses mounted, the operator filed for bankruptcy and shut Hahnemann in 2019. Reporting and commentary at the time emphasized that the land under the hospital was the real prize, making the closure look like a maximise profits first outcome rather than a turnaround attempt.

Envision Healthcare. KKR took Envision private in 2018. After regulatory changes curtailed “surprise billing,” high debt and operational strain culminated in a 2023 Chapter 11 filing. While not a real-estate play, it’s a strong example of a heavily-levered PE firm facing policy difficulties, increased costs, and eventual bankruptcy.

According to a recent study published in Annals of Internal Medicine, private equity acquisition of hospitals led to an increase in deaths among emergency department patients covered by Medicare. There were about 7.0 additional deaths per 10,000 visits (≈13% relative increase), coinciding with roughly 18% cuts in ED salary expenditures and 11.6% reductions in full-time employees. acpjournals.org+2hms.harvard.edu+2 “Staffing cuts are one of the common strategies used to generate financial returns for the firm and its investors,” said senior author Zirui Song of Harvard Medical School. hms.harvard.edu+1 “Each of them sort of comes up with the same result,” said Martin Kenney, distinguished professor in the Department of Human Ecology at UC Davis and author of Private Equity and the Demise of the Local. “Private equity takes over things in the medical field, quality goes down, prices go up.” The Guardian Sector-watchers also noted a wave of PE owned health care defaults in 2024–2025. Healthcare Dive

ABA Providers

Roll-Ups. A roll-up has become a central feature of how private equity operates in fields like health care and autism therapy.

In plain English, a roll-up means buying many small, separate businesses in the same industry and combining them into one larger company. The idea is to “roll up” a fragmented market—dozens or hundreds of mom-and-pop operations—into a single, standardized platform. The acquiring firm (often a private-equity firm aka sponsor) starts with one “platform company,” usually the first acquisition, and then bolts on smaller companies, which are called “add-ons” or “tuck-ins.”

The goal is scale. A larger organization can negotiate better reimbursement rates from insurers. It can spread administrative costs over more clinics, and attract new financing or buyers later at a higher valuation. Private-equity funds call this multiple expansion. A business valued at, say, six times EBITDA* might sell for nine or ten times EBITDA* once it’s big enough to draw institutional buyers.

In theory, a roll-up can bring efficiencies, shared billing systems, centralized HR, better staff training. But there are risks. Integrating dozens of local clinics often means cultural friction, turnover, and inconsistent clinical standards. If the owners grow too fast or use too much debt, quality can slip while financial problems increase. CARD’s bankruptcy in 2023 and other distressed cases illustrate how roll-ups can go wrong when growth outpaces management capacity or when labor shortages hit.

So when I use the term “roll-up,” I’m describing a private-equity-backed effort to buy, merge, and standardize smaller providers under one brand or corporate structure, typically to make the combined company more valuable for a future sale.

How does PE affect ABA providers?

What about ABA (autism) providers specifically? Since the late 2010s, ABA has been one of PE’s busiest roll-up arenas. Big sponsors (KKR among them) created platforms and bought dozens of clinics, aiming to standardize operations, negotiate better rates, and expand quickly. Coverage in STAT documents show concerns from parents and clinicians that profit pressure was shortening sessions, raising caseloads, and increasing staff turnover. At the same time, it noted that investment also expanded access. In short: access up, but quality risks and workforce strain in some chains. academic.oup.com

Detailed ABA examples

Center for Autism and Related Disorders (CARD). Blackstone acquired CARD in 2018 during a rapid growth phase. By mid-2023, CARD had amassed heavy debt and large losses; it filed for bankruptcy and was sold back to founder Doreen Granpeesheh for about $48.5 million, preserving many clinics but at a fraction of its prior valuation. Reporting and court filings attribute the fall to a mix of reimbursement pressure, labor shortages, and leverage, classic PE headwinds when growth stalls. proto.life+3Reuters+3Forbes+3

Trumpet Behavioral Health (TBH) started life in 2009, founded by Chris Miller and Lani Fritts, and expanded as a multi-state ABA provider based in Colorado. In 2014, TBH sold its Hawaii operations to BAYADA Home Health Care, but kept growing on the mainland. In March 2017, New York based private equity firm WindRose Health Investors acquired Trumpet. WindRose remained the sponsor until the business was sold in October 2023. Profile+2BAYADA Blog+2

BlueSprig Pediatrics is a KKR-backed ABA platform created in January 2018, led by CEO Keith Jones (ex-PSA Healthcare). BlueSprig grew by opening new sites and a string of acquisitions: The Shape of Behavior (2018), Aptitude Habilitation Services (2020), and Florida Autism Center (2020). On Oct. 18, 2023, BlueSprig bought Trumpet from WindRose, adding roughly three dozen locations and also bringing in Trumpet’s affiliates Therapeutic Pathways and The Behavior Center. Deal terms weren’t disclosed, which is typical. OPEN MINDS+7MarketScreener+7PR Newswire+7

Why BlueSprig bought Trumpet (and why WindRose sold)?: scale and geography. BlueSprig’s strategy since 2018 has been to build a national footprint via add-ons. Trumpet delivered density and reputable clinical programs (e.g., Therapeutic Pathways’ Kendall Centers). For WindRose, a sale to a larger platform fit the usual private equity strategy of exiting after a multi-year hold. Scope Research+4FOCUS+4OPEN MINDS+4

What happened to the businesses after these roll-ups? There’s no public filing that Trumpet itself entered distress; rather, it was absorbed into BlueSprig’s platform. BlueSprig’s earlier acquisitions (Shape of Behavior, Florida Autism Center, Aptitude) were strategies meant to extend state coverage. Public announcements emphasize growth, but independent, audited clinical-outcome reporting is still rare, so it’s hard to prove they prospered beyond geography and deal momentum. In the autism sector more broadly, you do see both growth success stories and notable stumbles (CARD), although that was under a different PE firm. OPEN MINDS+4PR Newswire+4Bluesprig Autism+4

Are mergers good for autism services? The evidence is mixed. Investigative reporting and policy papers document parent/clinician concerns (shorter sessions, cookie-cutter programmes, workforce turnover) as PE money purchased ABA networks. At the same time, consolidation has undeniably expanded access in many regions. The bottom line: mergers can improve reliability and payer contracting, but quality depends on staffing ratios, BCBA supervision capacity, and the operator’s clinical expertise. academic.oup.com

There are still ongoing deals, indicating the strategy hasn’t vanished: for instance, Behavioral Innovations changed PE hands in 2024 (estimated ~$300 M), and Behavior Frontiers sold to NexPhase in 2025. These sales underscores that many ABA platforms remain financial assets first, clinical organizations second.

What portends for ABA going forward?

More consolidation and standardization. Larger platforms will keep absorbing single site clinics. That can spread best practices and data infrastructure, but also homogenize care. STAT

Insurers push down rates. Insurers continue to push down rates and documentation tolerance. Highly levered platforms are more fragile when labour markets tighten or authorisations shrink. CARD’s bankruptcy showed those risks starkly. Reuters

Workforce strain. RBT recruitment, BCBA supervision capacity, and retention will make or break quality. Reports of high turnover under PE ownership are common; research reviews caution that staffing cuts are a recurrent risk factor in worse outcomes across health care. STAT+1

Policy and oversight. Transparency around ownership, staffing ratios, clinical outcomes, and use of sale-leasebacks would help separate responsible owners from irresponsible ones. Several policy groups are already tracking PE health-care acquisitions and bankruptcies. Private Equity Stakeholder Project PESP+1

Audits. Federal and state probes and audits (e.g., FBI activity in Minnesota; HHS-OIG findings in Wisconsin/Indiana) increase billing risk and documentation burden. PE companies with high debt are more exposed. Minnesota Reformer+2Office of Inspector General+2

Sensitivity to reimbursement + licensure. Platforms will expand in states where rates and rules support profits, and exit where they don’t (in one case 2025 one chain shuttered Arkansas centres after licensing + staffing laws tightened). 5 News Online

Growing compliance overhead. OIG and state audits mean more documentation rigour; good operators can meet this, but “growth at all costs” models struggle. Office of Inspector General+1

Quality varies by operator. Some PE companies lean into clinical infrastructure and supervision (what you want to see), others emphasise speed of openings and utilization, a red-flag.

Why do PE firms buy ABA programs?

Private-equity firms began targeting ABA programs between 2015 and 2020, drawn by a combination of growing demand, market fragmentation, and predictable reimbursement. Autism prevalence rates were climbing, and all fifty states had mandated insurance coverage for ABA, making the revenue base both large and relatively secure. The industry was also dominated by small, independent clinics—an ideal setup for private equity: acquire a single “platform” company and then bolt on dozens of smaller practices. Reimbursement rates, while not high, were steady and tied to commercial insurers and Medicaid, which made them easier to finance with debt. The investment thesis rested on the “exit story.” After rolling up enough clinics into a large, standardized network, private equity could sell to another fund, a strategic health-care company, or, in rarer cases, take the enterprise public.

Once a firm such as KKR, Arsenal, or Charlesbank acquires an ABA platform, it layers on debt. The operating company, not the fund, carries that debt, and so must generate enough cash flow to cover interest. The owners then reshape operations, centralising billing, HR, and scheduling, and negotiating payer contracts with the bargaining power of scale. Efficiency is often gained, but it comes at the cost of stricter productivity targets. That might mean fewer BCBAs supervising more cases, heavier reliance on RBTs, or pressure to shorten sessions and increase billable hours. The effect on quality depends on how carefully such changes are managed.

Because private-equity funds exist to deliver returns within three to seven years, they are never permanent owners. They sell when they believe the investment has peaked in value—sometimes to a larger private-equity firm, sometimes to a health-care chain, occasionally through an IPO. The imperative is to return capital to investors such as pension funds and endowments. What happens to the company after the sale varies widely. Some platforms continue to grow, as BlueSprig has under KKR, buying Florida Autism Center in 2020 and Trumpet Behavioral Health in 2023, thereby expanding access to clinics. Others stumble (CARD). Many providers fall somewhere in between, changing hands repeatedly as each successive owner adds leverage or extracts dividends, leaving the business more vulnerable to reimbursement cuts or rising labour costs.

For children and families, the results are mixed. On the positive side, consolidation has expanded access in underserved areas, created more consistent billing systems, and in some cases offered better staff training and career paths. On the negative side, pressure to hit productivity targets can yield rushed assessments, formulaic programs, and staff stretched too thin. High turnover among both RBTs and BCBAs undermines continuity of care, and financial distress can trigger sudden clinic closures. In short, private-equity firms do not buy ABA providers for their current profitability. They buy so a large, standardized platform will be worth more when sold. Debt and scale are the tools, and the outcomes range from successful expansions to cautionary tales. Families obtain improved access in some communities, but also heightened risks to stability and quality of care.

How To Evaluate an ABA Program

When evaluating an ABA provider, look closely at how the organisation actually operates. One question is about supervision and caseload. Will the supervisor come once a week, once every other week, or once a month? Each client is different; however, the more often the supervisor observes the therapists in a teaching session the better.

Staff training and tenure offer another window into quality. Strong providers invest in extensive staff training, and ensure weekly supervision continues thereafter. They also keep turnover low, ideally under thirty percent a year. If too many therapists leave within a year and training is minimal, you can expect disruption in care.

Continuity of care is just as important. Ask what happens when your child’s assigned therapist departs. A reliable provider has a plan: smooth hand-offs, overlapping staff during transitions, and early communication with families. In contrast, constant cancellations or unexplained service gaps signal instability.

The intensity of services delivered often reveals whether the clinic can meet its promises. Families should receive at least eighty-five percent of the hours that insurance has authorised. If the gap is wide between what is approved and what is delivered, it usually points to staffing shortages or poor management.

Finally, examine how programmes are designed. High-quality clinics avoid cookie-cutter goals and objectives. Instead, they use data, observe sessions directly, and adapt programs to a child’s unique needs and family priorities. Beware of centers that rely on preset curricula with little room for individualisation.

You can also listen for red flags in how providers talk about themselves. Heavy emphasis on growth or size without any mention of outcomes is a danger sign. So too is defensiveness when asked about caseloads or turnover. Frequent rebranding or sudden closures often hint at financial strain. And while telehealth can be useful, any clinic that leans on remote assessments without ensuring adequate in person follow up risks short changing children.

A related thought: Is autism best treated under the medical model (behavior therapy) funded by medical insurance or via an educational model that views the solution through teaching rather than therapy. After all ABA is based on principles of learning. Two of our main techniques are called discrete trial teaching and pivotal response teaching. Will the medical model make ABA services for autism ineffective? A discussion for another day perhaps.

Thank you for reading today’s post. Pleases subscribe, share and comment.

Leave a comment

_________________

*EBITDA stands for Earnings Before Interest, Taxes, Depreciation, and Amortization. It’s especially common in private-equity analysis and acquisitions (like the ABA roll-ups described) because it provides a standardized way to compare profitability across companies with different capital structures or tax situations.

References

Kannan S., Bruch J.D., Zubizarreta J.R., Stevens J., Song Z. “Hospital Staffing and Patient Outcomes After Private Equity Acquisition.” Annals of Internal Medicine. 2025. DOI: 10.7326/ANNALS-24-03471. acpjournals.org+1

Offodile A.C. II, et al. “Private Equity Investments in Healthcare: An Overview of the Scope of Private Equity–Backed Hospital Acquisitions (2003-17).” Health Affairs. 2021. DOI: 10.1377/hlthaff.2020.01535. healthaffairs.org

Singh Y., Reddy M., Zhu J.M. “Life Cycle of Private Equity Investments in Physician Practices: An Overview of Private Equity Exits.” Health Affairs Scholar. 2024;2(4):qxae047. DOI:10.1093/haschl/qxae047.