Evaluating the Verbal Behavior Milestones Assessment and Placement Program (VB-MAPP) Scores Using Principal Components Analysis

Introduction The Verbal Behavior Milestones Assessment and Placement Program (VB-MAPP) is an extensive tool used to assess children with autism and other developmental disabilities who have language delays. Applied behavior analysis (ABA) professionals frequently use the VB-MAPP to create personalized intervention programs catering to each child's needs. The lack of studies examining the VB-MAPP at the pretest, posttest, and differential scores using principal components analysis (PCA) suggests an opportunity to conduct PCAs on these different VB-MAPP scores. In doing so, researchers could better understand the VB-MAPP's dimensionality and factor structure at these levels. This, in turn, could inform the development of more effective assessment strategies and intervention plans for individuals with language and social communication challenges. Materials and methods From January 2018 to July 2021, The Oxford Center in Brighton and Troy, Michigan, treated autistic children using ABA therapy. A convenience sample of 13 children was retrospectively analyzed using VB-MAPP, which evaluates various behavioral milestones using a pretest-posttest design. Descriptive data analysis and internal consistency reliability estimates (using Cronbach's alpha) were calculated for pretest, posttest, and difference scores. A Wilcoxen signed-rank test was conducted to determine the statistical significance between the pretest and posttest. Correlation matrices were inspected for relevant relationships between VB-MAPP scales, and a PCA with orthogonal rotation was also performed on this pretest, posttest, and difference scores. Results The mean age for the children was 4.083 years ± 1.083 years, (95%CI 3.64, 4.36). Around 66.6% of the children had an autism severity level of three, 33.3% had a severity level of two, and none were at level one. Cronbach's alpha for internal consistency reliability of the pretest, posttest, and difference scores, indicating excellent reliability with values of 0.948 for the pretest and 0.937 for the posttest, respectively. The difference scores had a lower but acceptable reliability coefficient of 0.752. PCA on the pretest scores identified three factors that explain 85.584% of the total variation, indicating that these components capture most of the data's structure. The posttest PCA also identified three factors, accounting for 84.293% of the variance, suggesting a similar complexity and good model fit as the pretest. PCA revealed four factors explaining 82.317% of the variation for the difference scores. The increase in factors suggests that changes between pretest and posttest scores are complex, likely due to the ABA treatment, and require an additional component to represent the data adequately. There is a good model fit; the underlying structure is more complex than the pretest or posttest alone. Conclusions Robust coefficient alphas combined with the shift to a more detailed factor structure post-ABA treatment highlight ABA therapy's diverse and multi-faceted impact on children. The increase from three to four principal components indicates a richer and more nuanced pattern of improvements across different domains of verbal and social behavior. This detailed factor structure is a testament to the comprehensive and individualized nature of ABA treatment, reflecting the therapy's effectiveness in addressing specific needs and fostering broad developmental gains in children.