JMIR Publications

ABSTRACT

Background:

Introduction Behavior analytic services can have profound positive impacts on individuals and their families. Behavior analysts identify behaviors to target for acquisition, increase, or decrease by working closely with stakeholders and have used this information to develop numerous techniques and treatment approaches for analyzing and changing behavior, and ultimately, to create socially significant and meaningful change for individuals (BACB®, 2025a). Because social significance varies from person to person, environment to environment, and culture to culture, the specific behaviors targeted by behavior analysts can be as varied as vocational skills and dental care, shoe tying and aggressive behavior, or law enforcement personnel behaviors and kindergarten readiness skills (Heward, Critchfield, Reed, Dietrich, and Kimball, 2022). Due to efforts driven by families, grassroots organizations, service providers, and researchers across several decades, behavior analytic services are widely implemented to support autistic individuals and their families (Barry, et al. 2017). In recent years, the prevalence of Autism Spectrum Disorder (ASD) has become a topic of national interest within the United States, particularly following the release in April 2025 of updated prevalence data from 2022. This report indicates an increase in prevalence to 1 in 31 children from 1 in 36 children in 2020 (Shaw, et al. 2025). Additionally, according to the Centers for Disease Control and Prevention (CDC), as of April 2022, 47 of the 50 states in the United States recognize the diagnosis of ASD as a legitimate diagnosis and legally support some level of state or private insurance funding for services including applied behavior analysis (CDC, 2021). According to a review conducted by Gitimoghaddam, Chichkine, McArthur, Sangha, and Symington (2022), applied behavior analysis (ABA) has been recognized as one of the most empirically validated forms of support for persons diagnosed with ASD. In the field of ABA, services have largely been funded by private health insurance, state Medicaid plans, school districts, and private grants (Autism Speaks, 2025). The availability of funding has directly shaped behavior-analytic practice in several ways. For example, there has been a dramatic increase in the number of credentialed professionals trained to assist families and provide support. According to the Behavior Analysis Certification Board, Inc.® (BACB®), the demand for Board Certified Behavior Analysts (BCBAs®) in the United States increased 4,209% in the period from 2010 to 2020 (BACB®, 2025b). The entry level credential, the Registered Behavior Technician® (RBT®), was created in 2014 and is now the largest credentialing level by volume (BACB®, 2025). The BACB® consistently enhances the professional credentialing process and standards across all levels of credentialing. There is also proliferation of ABA centers across the United States. Today, the predominant form of ABA therapy in the United States is center- or home-based services delivered in a 2- or 3-tier model, in which the RBT® is the primary deliverer of care who is directly supervised by a Master’s- or Bachelor’s-level behavior analyst (often a BCBA®). This model is typically considered “comprehensive,” which, according to the Council for Autism Service Providers (CASP), is defined as a scope of treatment that “typically encompasses multiple simultaneous goals within and across multiple domains, such as language, behavior, activities of daily living, social skills, and cognition” in their practice guidelines (CASP, 2024). According to these guidelines, comprehensive ABA treatment generally requires 30-40 hours of therapy per week to ensure adequate time for teaching and practice opportunities. Although there is evidence of the efficacy of in-person, comprehensive ABA therapy delivered in a 2- or 3-tier model (Collins, et al, 2025), there are also many challenges related to this service delivery model. The first and most common challenge related to the traditional service delivery model is lack of access due to geography or a lack of providers within proximity. According to the U.S. Census Bureau census conducted in 2020, 80% of the population of the United States live in urban areas (defined as greater than 2500 people), and 20% live in rural areas (defined as 2500 people or less). Additionally, the availability of services in general tends to be concentrated in larger metropolitan areas (The Professional Society for Health Economics and Outcomes Research, 2025). Families who do have ABA services available within reasonable proximity of their homes experience other challenges. First, the available services do not always meet the needs of specific individuals with an autism diagnosis. In particular, school-age children with less global functional impairment may not be well-served by a comprehensive treatment program whose time commitment impacts school attendance or participation in enriching extracurricular activities. Recent studies examining the efficacy of lower intensity ABA services have demonstrated support for less intensive interventions (Levato, et al. 2025; Anderson, et al. 2024). Next, the comprehensive model relies on lesser-qualified service providers (e.g., a provider with a high school diploma and 40 hours of training, supervised by a Master’s- or PhD-level provider roughly 5% of the time) to deliver the bulk of client care. Turnover among these providers can be as high as 109% annually (CentralReach, 2025), leading to disruptions in care and concerns about quality. Another limitation of the traditional ABA service delivery model is a lack of caregiver involvement. According to CentralReach data aggregated across 305,000 clients whose service providers use CentralReach for practice management, families received on average 42 minutes per month of caregiver training. According to current literature exploring parent training and caregiver mediated interventions, increased caregiver participation in intervention can reduce parental stress levels, increase parental self-efficacy, and support positive child outcomes (Sneed & Samelson, 2022). Finally, the costs associated with healthcare for a child with autism, including traditional ABA services is significant. A recent report from the Agency for Healthcare Research and Quality (AHRQ) indicates that the average total health cost for a child being treated for autism was $20,122 compared with the average total health cost of $2,201 for a child not being treated for autism (Monnet & Zuvekas, 2025). Given the many challenges associated with the comprehensive ABA service delivery model, some providers are seeking alternative models that address the concerns detailed above and also deliver outcomes on behalf of the families they serve. One relatively new option is a BCBA® direct, technology-enabled, focused ABA therapy model. Services delivered via telehealth are less constrained by the geographic location(s) of the client or clinician, and it is possible for clinicians to reach clients in more rural areas, as well as those residing in areas with a lower ratio of providers to clients. Another benefit of this approach is that individuals who have more targeted clinical needs can receive more tailored intervention, focusing on the area of functional impairment that is most impactful at the time of treatment. These services can be delivered at a lower “dosage” than the recommended 30-40 hours per week for comprehensive services, making them more accessible to busy families and more cost effective for families and their funding sources. This approach is further bolstered by the more frequent and consistent contact with a BCBA®, who can make treatment decisions throughout each therapy session and evaluate procedures on an ongoing basis. Finally, the telehealth treatment modality can allow for increased caregiver involvement and therefore a potential increase in naturalistic learning opportunities by removing many of the barriers families receiving center-based services encounter, such as logistical difficulty with going to the center for caregiver training sessions and skills being targeted in an environment very different from their home.

Objective:

The purpose of this study was threefold: a) to examine the outcomes produced by the service model in an effort to determine if improvement was observed in accordance with industry standard progress monitoring tools the Vineland Adaptive Behavior Scales, 3rd Edition (Vineland-3) Comprehensive Parent/Caregiver version (Grzadizinski, Janvier, and Kim, 2020) and the Child and Family Quality of Life, Second Edition (CFQL-2), b) to examine the extent to which Minimum Clinically Important Differences (MCID) were created by therapy, and c) to further examine what variables may influence changes in scores over time including, age at the onset of therapy, severity level of diagnosis, gender, therapy funding source, type of services received, and hours of services rendered.

Methods:

Methods Service Model Behavior analytic services were delivered through a fully remote, focused treatment model provided by the same organization. Clinicians delivered services through proprietary, HIPPA-compliant, synchronous audio-video software. All services were rendered by a BCBA®, including assessment, direct therapy, therapy protocol modification, and caregiver training. At the outset of care, the assigned BCBA® conducted a comprehensive assessment with each participant that included the Vineland-3 the CFQL-2, indirect behavior assessments (e.g., Functional Assessment Screening Tool, Questions About Behavior Function), and direct assessments based on the participant’s clinical presentation (e.g., Verbal Behavior Milestones Assessment and Placement Program, Assessment of Functional Living Skills). The BCBA® completed the initial assessment by determining strengths and areas of need as well as identifying any behaviors that interfere with the participant being able to learn new skills or successfully contact less restrictive environments. Following the initial assessment and throughout the course of treatment, the BCBA® updated goals and adjusted therapy procedures in response to the participant’s progress. At 6-month intervals, the BCBA® conducted a re-assessment to further evaluate outcomes and adjust goals accordingly. This re-assessment also included re-administration of the Vineland-3, the CFQL-2, indirect behavior assessments, and direct assessments based on the participant’s clinical presentation. During ongoing treatment, services were delivered within three different treatment modalities: Direct Therapy plus caregiver training, Caregiver Mediated, and Caregiver Training Only. For families receiving Direct Therapy plus caregiver training, some sessions consisted of the child interacting directly with the BCBA® via synchronous audio-video while other sessions consisted of the BCBA® and caregiver(s) working directly together. Caregiver Mediated sessions consisted of the child, caregiver(s), and BCBA® being present via synchronous audio-video with the BCBA® providing coaching and guidance to the caregiver throughout the sessions, while the caregiver(s) interacted directly with the child. Finally, Caregiver Training Only services consisted of the BCBA® and caregiver(s) meeting via synchronous audio-video to discuss and practice strategies and to review changes in the child’s behavior. The child was present during some sessions. Participants The sample included de-identified records from individuals with the diagnosis of Autism Spectrum Disorder (ASD) who received technology enabled behavior analytic services delivered exclusively by Board Certified Behavior Analysts® (BCBA’s®) between 2021 and 2025. The total sample was 504 individuals including 388 (77%) males and 116 (23%) females and the average age at first Vineland-3 administration was 8.4 years of age. Figure 1 provides a breakdown of Medicaid recipients and non-Medicaid recipients according to the state where the participants resided at the time of the study. The majority of the participants (N=352; 70%) were Medicaid recipients and 30% (N=152) were non-Medicaid recipients who either utilized a commercial insurance product to fund services or did not use a third-party payer source for funding. The participants resided in one of the 5 following states: California, Georgia, New York, Texas, or Virginia. The largest number of participants resided in Virginia (N=241; 48%) followed by Georgia (N=205; 41%). The remaining states accounted for a small portion of the participants (12%) with only 1 participant from New York and fewer than 20 from California. Figure 1. Funding Source Distribution State Medicaid Recipients Non-Medicaid Total Percent California 18 1 19 3.8 Georgia 153 52 205 40.7 New York 1 0 1 0.2 Texas 25 13 38 7.5 Virginia 155 86 241 47.8 Total 352 152 504 100 ASD severity level was not indicated for all participants because that information was not provided by the diagnosing professional. The severity levels were available for 29.6% (N = 149) of the participants in the sample, however. Figure 2 displays the breakdown of severity levels by diagnostic domain with a slight majority of the participants being identified as Level 1. Figure 2. ASD Severity Levels Severity Level Diagnostic Domain 1 2 3 Social Communication 39.6% 34.9% 25.5% Restricted, Repetitive Behaviors 40.9% 34.9% 24.2% Note. N = 149 (29.6%) of the entire sample Procedures The analysis involved evaluating change over time in the results obtained from the comprehensive battery of assessments between each of the 6-month administrations beginning with the first administration until the last administration regardless of the number of administrations. Any participant with only one administration was not included in the sample and the participants ranged between having 2-7 comprehensive assessment batteries. The comprehensive assessment included the administration of the Vineland-3 including all domain scales (Communication, Socialization, Daily Living Skills, Adaptive Behavior Composite (ABC), Internalizing, and Externalizing), the Child and Family Quality of Life, Second Edition (CFQL-2) scales (Child, Family, Caregiver, Financial, Social Network, Partner Relationship, Coping, Overall), indirect behavior assessments (e.g., FAST, QABF), and direct assessments based on the participant’s clinical needs (e.g., VB-MAPP, AFLS). For the purposes of this study the domain scale scores from the domains in the Vineland-3 assessment tool and the CFQL-2 assessment tool were utilized to assess participant progress and outcomes achieved through the use of the technology enabled therapeutic model. Scale scores from each domain in the Vineland-3 and the CFQL-2 were captured from each administration and Minimal Clinically Important Differences (MCID) were calculated for Vineland-3 score changes. Additionally, various comparison analyses were conducted to determine the potential influence on score change from several variables including, gender, insurer, type of services rendered, hours of services, and ASD severity level. All participants consented to have their data utilized in this study and the data was de-identified prior to analysis. The data was analyzed using IBM SPSS Version 31.

Results:

Results Service Hours and Treatment Modality Fifty-eight percent of participants received Direct services, while 33% received Caregiver Mediated services, and 9% received Caregiver Training Only. As such, the majority of participants received some level of direct services and/or caregiver mediated services at adjusted levels. The breakdown of service hours rendered by funding source (Figure 3) was evenly distributed across funding sources and service types with the exception of non-Medicaid funded Caregiver Training Only which was lower in both average hours/week and average of total hours rendered from the first assessment until the most recent assessment. Figure 3. Service Hours Summary Average Hours/Week of Services Rendered Average Total Hours of Services Rendered Funding Source Treatment Modality Non-Medicaid Caregiver Mediated 2.54 121.36 Caregiver Training Only 1.75 91.40 Direct 2.52 119.12 Medicaid Caregiver Mediated 2.64 110.14 Caregiver Training Only 2.37 141.11 Direct 2.71 117.97 Vineland-3 Descriptives The following section is devoted to providing the descriptive statistics associated with the Vineland-3 administrations. Figures 4, 5, 6, and 7 display various types of score summaries of the scores between the first assessment administration and the last assessment (most recent) administration for all participants broken down by funding source and gender including the Maladaptive Behavior domains Internalizing and Externalizing, the Adaptive Behavior domains Communication, Socialization, Daily Living Skills (DLS), and the Adaptive Behavior Composite (ABC). Based on the recommended interpretation guidelines where High = 130-140, Moderately High = 115-129, Adequate = 86-114, Moderately Low = 71-85, and Low = 20-70 for scoring the Adaptive Behavior domains, and where Average = 1-17, Elevated = 18-20, and Clinically Significant = 21-24 on the Maladaptive Behavior domains. Figure 4 displays the average standard score at the time of first administration for all domains. Non-Medicaid funded participants tended to have the highest adaptive behavior domain scores on average at the first administration and all groups scored fairly consistently when comparing the maladaptive behavior domains. Figure 4. Average Standard Score at First Administration Funding Source Gender Communication Socialization DLS ABC Internalizing Externalizing Non-Medicaid Female 81.42 74.87 80.13 77.29 19.97 18.77 Male 76.41 71.72 78.84 74.99 19.40 18.68 Medicaid Female 66.01 68.79 71.92 69.09 19.84 18.84 Male 67.35 67.32 72.95 79.27 19.82 18.96 Figure 5 displays the average standard score at the time of the most recent administration. The ideal scenario is to see the scores from the Adaptive Behavior domains increase and the scores from the Maladaptive Behavior domains decrease. Across both funding groups and genders, there was improvement in all domains on average meaning that on average, participants displayed more adaptive behavior and less maladaptive behavior as a result of intervention at the time of the most recent administration. Figure 5. Average Standard Score at Most Recent Administration Funding Source Gender Communication Socialization DLS ABC Internalizing Externalizing Non-Medicaid Female 83.29 77.03 82.77 79.68 19.84 17.68 Male 80.01 77.36 82.26 78.90 18.77 17.85 Medicaid Female 70.34 71.68 74.53 72.04 19.48 18.22 Male 70.85 69.89 75.13 71.79 19.34 18.58 To further illustrate the change from first administration to most recent administration, Figure 6 displays the average standard score change across domains according to funding source and gender. While there was variability across the domains based on funding source and gender, on average there were increases in all Adaptive Behavior domains across gender and funding source and decreases in the Maladaptive Behavior domains across gender and funding source with the least amount of average change being in the Maladaptive Behavior domains. Figure 6. Average Domain Standard Score Change Between First and Most Recent Administration Funding Source Gender Communication N=504 Socialization N=504 DLS N=504 ABC N=504 Internalizing N=473 Externalizing N=473 Non-Medicaid Female 1.87 2.16 2.65 2.39 -0.13 -1.03 Male 3.60 5.64 3.42 3.91 -0.58 -0.75 Medicaid Female 4.33 2.89 2.61 2.95 -0.35 -0.65 Male 3.50 2.57 2.18 2.52 -0.46 -0.37 Figure 7 shows the aggregate standard score change data across all Adaptive and Maladaptive Behavior domains. The average aggregate data also indicates improvement in both the Adaptive and Maladaptive domains; however, it is worth noting that the standard deviations for the Socialization and Daily Living Skills domains were higher indicating greater variability in those domains. Figure 7. Aggregate Average Domain Standard Score Change Domain Average Change Standard Deviation Communication (N=504) 3.56 9.77 Socialization (N=504) 3.34 11.62 Daily Living Skills (N=504) 2.58 10.15 ABC (N=504) 2.92 7.89 Internalizing (N=473) -0.45 1.81 Externalizing (N=473) -0.55 2.05 To further evaluate changes in standard domain scores, the percentage of participants showing a minimum of a 1-point improvement was calculated for the full sample. Figure 8 illustrates this distribution with 82% of the participants showing a standard score change of +1 point in at least one domain with 29% showing a standard score change of +1 point in all domains. Less than 19% showed no standard score change or a 1- point change. Figure 8. Standard Score Change Distribution (+1 Point, No Change, or -1 Point Change) Domain/Domain Combination Count Percent Communication, Socialization, Daily Living Skills 0 0.00% Communication, Socialization, ABC 41 8.13% Communication, Daily Living Skills, ABC 39 7.74% Socialization, Daily Living Skills, ABC 44 8.73% Communication 40 7.94% Communication and Socialization 3 0.60% Communication and Daily Living Skills 8 1.59% Communication and ABC 14 2.78% Socialization 34 6.75% Socialization and Daily Living Skills 6 1.19% Socialization and ABC 10 1.98% Daily Living Skills 17 3.37% Daily Living Skills and ABC 7 1.39% Adaptive Behavior Composite 0 0.00% All Domains 148 29.37% No change or -1.0 Change 93 18.45% Total 504 100% Figure 9 displays the percentage of participants who experienced Minimal Clinically Important Differences (MCID) where the MCID scale is change in Communication ≥ 2.0, change in Socialization ≥ 2.6, change in Daily Living Skills ≥ 2.6, and change in ABC ≥ 2.0 (Internalizing and Externalizing changes will be shown separately). Overall, the highest MCIDs were observed in the Communication and ABC domains across genders and funding sources. Figure 9. Participants Experiencing MCID From the First Administration to the Most Recent Administration Insurer Gender ABC ≥ 2.0 (%) Communication ≥ 2.0 (%) Daily Living ≥ 2.6 (%) Socialization ≥ 2.6 (%) Non-Medicaid Female 45 45 45 45 Male 58 53 49 56 Medicaid Female 56 56 48 58 Male 51 57 44 45 To further illustrate the MCID performance across the sample, Figure 10 displays the percentage of participants who met and/or did not meet MCID thresholds in at least one domain. Approximately 25% of the participants met MCID thresholds for all the domains while 24% did not meet MCID threshold in any of the domains. More than 75% of the participants met the MCID threshold for at least one domain. Figure 10. Participants Meeting/Not Meeting MCID Threshold in at Least 1 Domain MCID Category Count % All Domains 124 24.60% No MCID Threshold Met 121 24.01% Communication Only 51 10.12% Communication + Socialization + ABC 40 7.94% Communication + Socialization + DLS 0 0.00% Daily Living Skills + Socialization + ABC 38 7.54% Communication + Daily Living + ABC 34 6.75% Socialization Only 32 6.35% Communication + ABC 16 3.17% Daily Living Skills Only 15 2.96% Socialization + ABC 9 1.79% Communication + Daily Living Skills 9 1.79% Daily Living Skills + ABC 7 1.39% Daily Living Skills + Socialization 5 0.99% Communication + Socialization 3 0.60% ABC Only 0 0.00% Total 504 100% Figure 11 provides an account of Internalizing scores from first administration to the most recent administration based on the percentage of participants who fall into one of three scoring categories; Average = 1-17, Elevated = 18-20, and Clinically Significant = 21-24. Due to the nature of the scoring ranges between the three categories, substantial movement across categories is less likely between Average and the other categories (Elevated and Clinically Significant) and the results indicate there were no substantial changes in categorization percentages for Internalizing scores across first to most recent administrations. Figure 11. Percent of Participants Within Internalizing Categories First Administration Funding Source Gender Average Elevated Clinically Significant Non-Medicaid Female 7 57 37 Male 11 64 25 Medicaid Female 11 42 47 Male 8 60 32 Most Recent Administration Funding Source Gender Average Elevated Clinically Significant Non-Medicaid Female 6 58 35 Male 11 64 25 Medicaid Female 11 44 46 Male 8 60 32 Figure 12 provides an account of Externalizing scores from first administration to the most recent administration based on the percentage of participants who fall into one of three scoring categories; Average = 1-17, Elevated = 18-20, and Clinically Significant = 21-24. Due to the nature of the scoring categories, substantial movement across categories is less likely between Average and the other categories (Elevated and Clinically Significant). The results indicate there were changes for non-Medicaid Males and Females in reduction in the percent of participants in the Clinically Significant category and increase in the Average category, which indicates an overall reduction in external maladaptive behavior for the sample. Figure 12. Percent of Participants with Externalizing Categories First Administration Insurer Gender Average Elevated Clinically Significant Non-Medicaid Female 20 53 27 Male 25 51 24 Medicaid Female 25 51 25 Male 18 57 25 Most Recent Administration Insurer Gender Average Elevated Clinically Significant Non-Medicaid Female 39 55 6 Male 36 51 13 Medicaid Female 33 52 15 Male 25 58 17 Vineland-3 Statistics A number of statistical analyses were conducted to determine the potential relationship(s) between relevant variables and any possible predictive power of outcomes. Those variables included gender, funding source, diagnosis severity levels, treatment modality, and age at first administration. The following section provides a summary of the analyses conducted and the resulting conclusions. Figure 13 illustrates the level of variance associated with each Adaptive Behavior domain according to funding source type. There is a substantial amount of variance for each domain and for each funding source, with the highest levels of variability associated with the Socialization domain. Figure 13. Variance Domain Medicaid Count Medicaid Variance Non-Medicaid Count Non- Medicaid Variance ABC 352 56.37 152 75.58 Communication 352 93.53 152 100.23 Daily Living 352 97.02 152 117 Socialization 352 133.86 152 134.6 Figure 14 displays the T-test results testing for statistical significance across Adaptive Behavior domains. At the p<0.05 level, there was slight statistical significance for the socialization domain as compared to the other domains. Figure 14. T-Test Across Adaptive Behavior Domains Domain t-statistic p-value (p < 0.05) ABC -1.2 0.231 Communication 0.47 0.641 Daily Living Skills -0.96 0.339 Socialization -2.03 0.043 Figure 15 displays the confidence intervals for domain standard score change for the Adaptive Behavior domains. We can see in this data that the Socialization domain has the largest effect size of the domains, but it is still relatively small. The Cohen’s d calculations suggest a smaller and therefore minimal differences across the domains. Figure 15. Confidence Intervals for Domain Standard Score Change Domain Cohen's d 95% CI Lower 95% CI Upper ABC 0.124 -0.067 0.314 Communication -0.046 -0.236 0.144 Daily Living Skills 0.096 -0.094 0.287 Socialization 0.197 0.007 0.388 A T-test for statistical significance across domains by gender and funding source was conducted. Figure 16 illustrates there is a statistically significant difference (p<0.05) between males across funding sources. Males who were supported by non-Medicaid sources showed more significant improvement as compared to males supported by Medicaid. There were no other sources of statistical significance in this comparison. Figure 16. T-Test Results – Domain by Gender and Funding Source for Adaptive Behavior Domains Domain Gender Medicaid Mean Medicaid SD Non-Medicaid Mean Non-Medicaid SD p-value ABC Male 2.521 7.357 3.909 8.702 0.129 Female 2.953 8.000 2.387 8.697 0.753 Communication Male 3.498 9.377 3.603 9.481 0.919 Female 4.329 10.578 1.871 11.935 0.317 Daily Living Male 2.180 9.619 3.421 11.096 0.289 Female 2.612 10.597 2.645 9.796 0.987 Socialization Male 2.573 11.680 5.645 11.520 0.016 Female 2.894 11.283 2.161 11.691 0.764 A regression analysis (Figure 17) of the relationship between the age at first administration and domain standard score change was conducted. The results indicate that older age is associated with smaller improvements. Additionally, statistically significant relationships (p < 0.05) were found between the ABC, Communication, Daily Living Skills, and Socialization domains. This suggests that how one performs on one domain has some relationship to how one might perform on another. It is also important to note that ABC is a composite score, so there should be some relationship between ABC and the other domains. Finally, the R-squared values are low, which indicates that age at first administration explains only a small portion of the variance in standard score changes. Figure 17. Age at First Administration as a Predictor of Positive Standard Score Change Domain Coefficient P-value R-squared Internalizing -0.0123 0.4821 0.008 Externalizing -0.0187 0.3914 0.011 ABC -0.0452 0.0278 0.064 Communication -0.0528 0.0193 0.072 Daily Living Skills -0.0496 0.0225 0.069 Socialization -0.0581 0.0157 0.078 Additionally, a regression analysis was conducted to assess whether or not Severity level is a predictor of positive domain standard score change (Figure 18). Severity level was not statistically significant (p > 0.05) predictor. The model explains less than 1% of variance in average standard score change (R² = 0.008). The results suggest that severity level is not a strong predictor of overall improvement across domains. Figure 18. Regression Analysis of Severity Level as a Predictor of Positive Standard Score Change. Predictor Coefficient p-value 95% CI Lower 95% CI Upper Intercept 0.9734 0.574 -2.441 4.388 SC Severity 1.2308 0.609 -3.509 5.971 RRB Severity -0.967 0.687 -5.708 3.774 Lastly, in relation to the Vineland-3 results, a regression analysis (Figure 19) was conducted to determine if age and hours of service were predictors of domain MCID thresholds being met. Neither age nor hours of service were a significant predictor of MCID improvement (p<0.05). Figure 19. Regression Analysis of Age and Hours of Service Compared to MCID Thresholds. Domain Age Coefficient p-value Hours Coefficient p-value Communication -0.0087 p = 0 .680 0.0009 p = 0.507 Daily Living Skills -0.0031 p = 0.550 0.0008 p = 0.550 Socialization -0.0074 p = 0.719 0.0012 p = 0.408 ABC -0.0036 p = 0.858 0.0009 p = 0.561 With regards to the Maladaptive behavior domains (Internalizing and Externalizing) a Paired-Samples t-test was conducted for Internalizing Standard and Externalizing Standard scores between first administrations to most recent administrations. The paired-samples t-test comparing internalizing scores from first Vineland-3 administration to scores from the most recent Vineland-3 administration results were M = 19.71, SD = 1.95 and most recent administration results were M = 19.26, SD = 2.06, yielding an analysis result of t(472) = -5.49, p<.0001. The change in Internalizing scores from the first administration to the most recent administration is statistically significant (p < .05). The negative mean change indicates a slight decrease in Internalizing scores overall, which suggests improvement (reduction) in maladaptive internalizing behaviors. With regards to externalizing scores the paired-samples t-test comparing externalizing scores from first Vineland-3 administration to scores from the most recent Vineland-3 administration results were M = 18.83, SD = 2.42 and most recent administration results were M = 18.29, SD = 2.49, yielding an analysis result of t(495) = -5.97, p<.0001. The change in Externalizing scores from the first administration to the most recent administration is highly statistically significant (p < .05). The negative mean change indicates a decrease in Externalizing scores overall, which suggests improvement (reduction) in maladaptive Externalizing behaviors. Child and Family Quality of Life - Version 2 (CFQL-2) Descriptives This section is devoted to providing the descriptive statistics for those participants who had the CFQL-2 administered. The process of evaluating CFQL-2 results primarily focused on assessing Total QoL and changes in Total Score. Figure 20 displays the gender and funding source data. The majority of the 431 participants (54%) were males who received support from Medicaid with Medicaid funded participants accounting for 71% of the sample. Figure 20. Gender and Funding Source Distribution Funding Source Gender Count Percent Medicaid Female 73 17 Male 231 54 Non-Medicaid Female 28 7 Male 99 23 Note. N=431 Figure 21 illustrates the number of participants who participated in CFQL-2 administrations over time. All participants in the sample (N=431) had at least one CFQL-2 administration with 30.86% having only one administration, 48.96% having two, and the remaining number accounting for 20.19%. Figure 21. CFQL-2 Count of Administrations # of CFQL-2 Administrations # of Participants % of Participants 1 133 30.86 2 211 48.96 3 72 16.71 4 11 2.55 5 4 0.93 Note. N=431 Figure 22 illustrates the average score (and score ranges) for each CFQL-2 scale across all participants at the last administration. The average Total QoL Score at last administration was 92.36 with a range of 51.0-125.0. Figure 22. Average Scale Scores at Last Administration Scale Average Score at Last Administration (Range) Child QoL 12.98 (0-20) Family QoL 13.30 (0-20) Caregiver QoL 14.09 (0-20) Financial QoL 11.28 (0-15) Social Network QoL 15.13 (0-15) Partner Relationship QoL 11.55 (0-20) Coping QoL 11.72 (4-15) Total QoL 92.36 (Range: 51.0-125.0) Figure 23 displays the average percentage of participants who improved across scale scores throughout the intervention programming period according to all CFQL-2 administrations. Positive Change was defined as a ≥+1 standard score change. More than half the sample indicated improvement in Family QoL and Caregiver QoL, and just under 50% showed improvement in Child QoL. Total QoL indicated that overall 59% of the participants showed improvement. Figure 23. Percentage of Participants Demonstrating Positive Change in Quality of Life Improvement Scale % of Participants Demonstrating Improvement Child QoL 49.33 Family QoL 53.69 Caregiver QoL 52.01 Financial QoL 41.61 Social Network QoL 37.58 Partner Relationship QoL 25.84 Coping QoL 44.97 Total QoL 58.72 Figure 24 represents the percentage of participants Quality of Life (QoL) scaled scores where improvement occurred as well as the category of change exclusively at the time of the most recent CFQL-2 administration. Improvement was determined based on the following scale thresholds: Child QoL, Family QoL, Caregiver QoL, Social Network, and Partner Network QoL (High = 16-20, Adequate = 8-15, and Low = 0-7); Financial QoL (High = 12-15, Adequate 6-11, and Low = 0-5), and Coping QoL (High = 12-15, Adequate = 8-11, and Low = 4-7). The Change categories were defined as: Same = No meaningful change from previous score, Increased = positive change (≥+1 point), and Decreased = negative change (≤+1 point). Overall, 41% of participants showed an increase in Total QoL, 33% remained the same, and 26% decreased with 72% of the participants experiencing Adequate QoL according to the assessment. Figure 24. Percent of Participants Showing Improvement and Change Status at Last Administration Scale High Adequate Same Low Increased Decreased Child 15.31 81.67 43.62 2.32 34.11 22.27 Family 35.03 51.04 42.23 12.99 37.12 20.65 Caregiver 38.05 56.38 42.46 3.94 35.96 21.58 Financial 69.61 22.04 51.74 7.42 28.77 19.49 Social Network 0 0 46.4 0 25.99 27.61 Partner Relationship 0 0 58.47 0 17.63 23.43 Coping 72.85 25.06 45.48 2.09 31.09 23.43 Total 27.84 71.69 33.18 0.46 40.6 26.22 Figure 25 provides a more detailed breakdown of Clinically Significant Change (CSC) where CSC is defined for Total Score QoL standard score change as a change that is ≥7 raw score points or ≥0.3 increase in the score average, and the score change for the other domain scores is ≥3-4 raw score points (Frazier, 2023). In order to capture change in scores over time, those participants with only one CFQL-2 administration were excluded from this analysis. This resulted in a reduced participant sample of N=298. The results indicated that 48% (N=142) of the participants in the reduced sample experienced CSC. There were 222 (75%) males and 76 (25%) females in the sample. Medicaid recipients represented 71% (N=211) of the sample with 29% (N=87) receiving support from a non-Medicaid source. Just under half the participants (N=142) demonstrated CSC with the percentage of males being higher than females. Neither the Medicaid recipient group nor non-Medicaid recipient group demonstrated more CSC than the other despite the sample including significantly more Medicaid recipients (N=211). Figure 25. Participants Experiencing Clinically Significant Change Number Percent Total 142 47.65% Male 111 50.00% Female 31 40.79 Medicaid 101 47.87 Non-Medicaid 41 47.13 Note. N=298 CFQL-2 Statistics A number of statistical analyses were conducted to determine the potential relationship(s) between relevant variables and any possible predictive power of the CFQL-2 outcomes. Those variables included gender, funding source, diagnosis severity levels, treatment modality, and age at the first administration onset of administration. The following section provides a summary of the analyses conducted and the resulting conclusions. Figure 26 displays the results of a regression analysis testing whether or not Age at First Administration, Medicaid Status, ASD Severity Level, Initial Total QoL, Average Service Hours/Week, or Gender are significant predictors of Clinically Significant Total QoL standard score change where Clinically Significant standard score change equals any change in Total QoL Average ≥0.30 item average. The results indicate that Age at First Administration, Medicaid Status, ASD Severity Level, Average Number of Service Hours Per Week and Gender were not significant predictors (p>0.05) of Clinically Significant Change in the Average Total QoL Score. Initial Total QoL (i.e., the first Total QoL score captured) however, was a significant predictor (p<0.05) of CSC indicating that the higher the initial Total QoL score, the less likely the participant will be to show improvement in scores over time and the more likely a participant with a lower initial Total QoL score would be to show improvement. Additionally, a chi-square test of independence was conducted to examine the relationship between Medicaid status and Clinically Significant QoL improvement. The results indicated that 127 out of 211 (60.2%) Medicaid participants achieved clinically significant improvement and 48 out of 87 (55.2%) of non-Medicaid participants achieved clinically significant improvement. The difference, however, was not statistically significant, χ²(1) = 0.45, p = .503, indicating that Medicaid status was not associated with the likelihood of clinically significant QoL improvement. Figure 26. Regression Analysis of Predictors of Clinically Significant Changes in Total QoL Scores Predictor OR CI Lower CI Upper p Intercept 25.15 1.12 565.21 0.042 Age at First Administration 1 0.92 1.1 0.949 Medicaid Status 1.25 0.56 2.83 0.584 ASD Severity Level 1.17 0.71 1.92 0.548 Initial Total QoL Score 0.96 0.93 0.98 0.001 Average Hours/Week Between First and Most Recent Administration 1.17 0.81 1.71 0.406 Gender 1.3 0.54 3.11 0.559 Caregiver Satisfaction A 4-question Likert-scale based caregiver satisfaction survey was conducted to capture more anecdotal information related to caregiver experience and provide a mechanism for caregivers to give feedback regarding services. Of the 504 participants in the sample, a portion (N=162) returned the satisfaction survey. The survey asked general questions about satisfaction and experience. Figure 27 provides a detailed account of the questions and answer options. Figure 27. Caregiver Satisfaction Survey Questions Scale How likely are you to recommend services to a family with a child similar to yours? Scale 1-10; 1 = not at all likely and 10 = Extremely likely Have services been effective for your child? Very Effective, Effective, Unsure, Ineffective, Very Ineffective What aspect of services is most important to you? Accepts my insurance, Caregiver training, Effective Therapies, Experienced BCBAs®, No waitlist, Virtual/Video format, All the above, Other In the last month, how many times have you considered moving your child to residential services and/or calling emergency services to support your child? Never 1-2 Times 3+ Times The responses to the questions provide a better understanding of how the caregivers who responded view the services they are receiving. Figure 28 provides a summary of the responses for each of the 4 survey questions. A large portion of the respondents indicated being extremely likely (84%) to recommend services with only 16% answering the question with a score of 9 or less. An overwhelmingly large portion of caregivers indicated services were very effective or effective (95%), with only 2% indicating services were ineffective or very ineffective. Just under half the respondents (41%) indicated Effective Therapies as the most important aspect of services followed by the virtual/video format (21%) and having Experienced BCBAs® (20%) with only 18% of respondents indicating Caregiver Training, Accepts My Insurance, No Waitlist, All the Above, and Other combined. Lastly, 94% of respondents indicated they had never considered moving their child to residential services and/or calling emergency services with 4% indicated they’d considered the options 1-2 times, and 2% indicated they’d made those considerations 3 or more times. Figure 28. Caregiver Survey Responses; Percent of Participants Question Responses Percent How likely are you to recommend services to a family with a child similar to yours? 0 – 1 5 6 7 8 9 10 0% 1% 1% 3% 4% 7% 84% Have services been effective for your child? Very Effective Effective Unsure Ineffective Very Ineffective 69% 26% 3% 1% 1% What aspect of services is most important to you? Effective Therapies Virtual/Video Format Experienced BCBAs® Caregiver Training Accepts My Insurance No Waitlist All The Above Other 41% 21% 20% 7% 4% 4% 2% .06% In the last month, how many times have you considered moving your child to residential services and/or calling emergency services to support your child? 1-2 Times 3 or More Times Never 4% 2% 94% A chi-square test of significance (X2(3,N = 162) = 1.85, p = 0.603) was conducted to assess the relationship between positive change in Vineland-3 Scores across domains and customer satisfaction. There was no significance between Vineland-3 domain score change and customer satisfaction (p>0.05). In addition, a chi-square test of significance (X2(21,N = 162) = 16.57, p = 0.737) was conducted to evaluate the association between caregiver satisfaction and what aspect of services were indicated as being most important. There was no statistically significant association between caregiver satisfaction and what aspect of services were indicated as being most important (p>0.05). Lastly, a chi-square test of significance (X2(6,N = 162) = 23.82, p = 0.0006) was conducted to assess the relationship between caregiver satisfaction and frequency of considering residential or emergency services. The results indicate there is a significant statistical association between caregiver satisfaction and the frequency of considering residential or emergency services (p>0.05) which suggests that families who considered crisis interventions more frequently are more likely to report lower satisfaction and those who did not consider crisis interventions are more likely to report higher satisfaction.

Conclusions:

The present study evaluated the outcomes of a technology-enabled, focused ABA service model for individuals diagnosed with Autism Spectrum Disorder (ASD) and explored caregiver perceptions of service effectiveness and satisfaction. Findings indicate that the technology enabled, BCBA®-delivered model produced measurable improvements in adaptive functioning and quality of life for many participants, while also reducing maladaptive behavior and revealing important considerations for caregiver experience. Prior research on ABA interventions where the focus was on more comprehensive in-person intervention programs generally shows that participants demonstrate positive changes across Vineland-3 adaptive behavior domains through participation in those programs. For example, Aitken, Lazerwitz, Eash, Hattangadi, Mukherjee, Marco, and Shapiro (2025) found that out of a sample of 1,225 children diagnosed with ASD who participated in comprehensive in-person based services, 61% demonstrated improvement within multiple domains of the Vineland-3. In the present study, the greatest improvements were observed in Communication and Socialization; however more than 80% of participants exhibited at least a one-point improvement in one or more adaptive behavior domains and approximately 29% improved across all adaptive behavior domains. In the maladaptive behavior domains 38% of the participants showed improvement in the Internalizing domain and 34% showed improvement in the Externalizing domain. These outcomes translate to significant reductions in either the frequency, intensity, and/or duration of maladaptive behavior. This study provides additional insight into outcomes for families participating in ABA therapy when compared with previous studies that included only the adaptive functioning domains (Ostrovsky et al, 2023; Aitken et al, 2025) Furthermore, over 75% of participants met Minimal Clinically Important Difference (MCID) thresholds in at least one domain as well. The results, at the very least, clearly point to the potential clinical utility of this technology enabled service model as a means of addressing targeted skill acquisition and behavior reduction. Quality of life outcomes, as measured by the CFQL-2, also demonstrated encouraging trends. Nearly 59% of participants showed improvement in overall quality of life scores, and 48% achieved clinically significant change. Regression analyses indicated that initial (baseline) QoL scores were predictive of improvement, with lower baseline scores associated with greater gains. This finding suggests individuals with less adaptive behavior and lower perceived quality of life at the outset of therapy may experience more pronounced benefits from intervention over time as compared to those with more adaptive behavior and higher perceived quality of life at the outset. Recent systematic reviews and meta-analyses examining the role of child age, baseline cognitive ability, and also the amount of intervention received, suggest findings across studies remain inconsistent and advocate the need for further research with controlled designs to better understand how pre-program characteristics are associated with intervention-related outcomes (e.g., Vivanti et al., 2014; Whitehouse et al., 2020). Caregiver satisfaction data revealed overwhelmingly positive opinions of service effectiveness, with 95% of respondents rating services as effective or very effective and 84% reporting they were extremely likely to recommend services. However, exploratory analyses examining associations between satisfaction and other variables yielded mixed results. No significant relationship was found between satisfaction and Vineland-3 score improvements or between their satisfaction and the aspect of services the caregivers identified as being most important. These findings suggest that caregiver satisfaction may be additionally influenced by factors beyond measurable clinical outcomes, such as communication quality, accessibility, and perceived responsiveness of providers.Conversely, a significant association between caregiver satisfaction and the frequency of caregivers considering residential or emergency services (χ²(6, N = 162) = 23.82, p = .0006, Cramér’s V = .27) did emerge. Families who reported considering crisis interventions more frequently were also more likely to express lower satisfaction in services, whereas those who never considered such interventions reported higher satisfaction in services. This moderate effect size underscores the importance of addressing caregiver stress and crisis risk as part of service delivery, as these factors may substantially impact overall satisfaction and engagement throughout the course of therapy. Limitations A number of study limitations should be noted. First, the present study relied on deidentified archival data, which limited the ability to control for potentially confounding variables such as socioeconomic status, comorbid conditions, and caregiver stress levels; there was also no measure of caregiver engagement or individual treatment goals/goal mastery. Second, the caregiver satisfaction survey was completed by only a subset of participants who represented less than half the sample of participants (approximately 32%). That reality introduces potential response bias depending upon the nature of those who completed the survey and their histories with other services, existing familiarity with ABA methodology, and their familiarity with alternative service delivery options. Third, the absence of a control group limits our ability to draw direct conclusions about the effectiveness of the technology-enabled targeted model compared to traditional in-person comprehensive services. Fourth, variability in the number of Vineland-3 administrations and service hours across participants may have influenced reported outcomes. And finally, there are other Vineland-3 scores that may be more sensitive to change over time and therefore better for longitudinal tracking. For example, the Growth Scale Values (GSV’s) monitor a person's progress in the same domain across multiple assessments versus comparing standard scores from an individual’s performance to that of their peers within a domain. Implications for Practice The findings have potentially important implications for service delivery models in ABA and how the field of practitioners in ABA work to meet the needs of those diagnosed with ASD. Technology-enabled, BCBA®-delivered interventions may offer a viable alternative to traditional high-intensity, center-based comprehensive models, particularly for families seeking flexible and targeted support or those who have difficulty gaining access to therapy. The strong caregiver satisfaction ratings suggest that virtual/video-based services can meet family needs when implemented with fidelity. The combination of positive outcomes and strong caregiver satisfaction raises the possibility that this model could have a multifaceted impact on the socially significant behaviors that impact the independence, well-being, and individual success of those diagnosed with ASD and their families. The significant association between satisfaction and crisis considerations highlights the need for integrated caregiver support strategies, including proactive crisis planning, coordination of care with a family’s other providers, and stress management resources. Families who are facing crisis situations with their children potentially require support beyond what is typically available even in comprehensive in-person services. It’s plausible that a more BCBA® directed and implemented model of services may provide that level of necessary proactive crisis planning and stress management by having the luxury of more frequent connections being created in a more targeted manner. By working directly with caregivers to decrease maladaptive behavior as the primary objective of intervention programming, the BCBA®’s support may more meaningfully improve the family’s experience with services as well as the family’s general quality of life. Future Directions A great deal of research addressing the role, experience, and needs of the caregiver in the context of participating in ABA services is needed. For example, research could look to provide deeper insight into caregiver perceptions and to more specifically identify factors that drive satisfaction beyond child progress alone. Future caregiver research might more closely examine the relation between caregiver satisfaction and treatment fidelity, caregiver satisfaction and service duration (e.g., do satisfied caregivers stay in treatment longer), and level of caregiver involvement and client outcomes. Exploring interventions that specifically target caregiver stress and crisis prevention within remote service frameworks may enhance both satisfaction and overall treatment efficacy. Greater emphasis could be placed on examining the combinations of adaptive and maladaptive skills at therapy onset that equate to specific outcomes rather than more commonly examined variables such as age, gender, diagnostic severity level, and treatment intensity as the primary predictors of outcomes. Additionally, it is feasible to consider the implications of how types of treatment goals, the density of treatment goals, and mastery of goals impact the development of adaptive behavior and the reduction of maladaptive behavior over the course of intervention programming. The relationship between these factors alone could have direct influence over what scores are obtained on both the Vineland-3 and the CFQL-2 as well as the types of intervention outcomes achieved. Lastly, as noted as a study limitation, GSVs may be more sensitive to change over time and therefore more effective at longitudinal tracking rather than relying on standard scores and/or standard score changes as the primary tracking mechanism(s). The present study provided an overview of the outcomes obtained from the implementation of a technology-enabled ABA service model on Vineland-3 and CFQL-2 standard scores. Future research should work to more specifically compare technology-enabled ABA services with that of traditional delivery models through an examination of clinical outcomes, caregiver preferences, and BCBA® experiences. Longitudinal studies are needed to assess the durability of adaptive skill gains and quality of life improvements over time for recipients of both types of service models. Given the results however, it’s clear that the use of technology-enabled services offers great promise toward addressing a broad range of needs in a potentially more economical and accessible fashion without sacrificing the high quality of services provided or lessening the socially significant impact on the lives of children and families the practice of ABA has become synonymous with.