|
| Analyzing Efficacy of Components of the Comprehensive Application of Behavior Analysis to Schooling on Student and Teacher Outcomes |
| Sunday, May 25, 2025 |
| 6:00 PM–6:50 PM |
| Convention Center, Street Level, 147 B |
| Area: EDC; Domain: Applied Research |
| Chair: Susan Buttigieg (Columbia University) |
| CE Instructor: Susan Buttigieg, Ph.D. |
| Abstract: The Comprehensive Application of Behavior Analysis to Schooling (CABAS) is acybernetic system which takes into account the behaviors and outcomes of all ofits parts (students, caregivers, teachers, mentors/supervisors, and administrators)(Singer-Dudek, Keohane, & Matthews, 2021). In this system, the student and theirprogress drives the decision-making of the remaining stakeholders. In this system,performance of all participants is measured, monitored, and supported tomaximize outcomes. Some cornerstones of the CABAS system are the learn unit(Albers and Greer,1991; Bahadourianet al. 2006; Greer, 2002), the TeacherPerformance Rate Accuracy (TPRA) (Ross et al.,2005; Singer-Dudek et al., 2010)and the Early Learner Curriculum and Achievement Record (ELCAR) (Greer et.al.2023). In three studies, we demonstrate how these tools can be used to maximizestudent and teacher outcomes. In one study, we used the TPRA as a teachertraining tool and measured correct delivery of learn unit components. In anotherstudy, we tested the efficiency of Google Gemini Pro 002 on identification ofinstructional trial components. Finally, we tested the convergent and divergentvalidity of the ELCAR curriculum with Preschool Language Scales and ChildhoodAutism Rating Scale-2. |
| Instruction Level: Intermediate |
| Keyword(s): Artificial Intelligence, CABAS, Learn unit, Teacher training |
| Target Audience: Intermediate- BACB holders, BACB supervisors, BACB consultants in school settings Prerequisite skills include holding BACB certification |
Learning Objectives: 1. The attendee will describe the components of the TPRA and give examples of an incorrect response for each component, and how to correct it.
2. The attendee will explain why the TPRA is an integral part of the CABAS system and list a potential benefit of adding something similar to their practice.
3. The attendee will describe the three components of the learn unit and a correct and incorrect example of each. |
| |
| Testing the Effects of Artificial Intelligence Measuring the Three-Term Contingency |
| Lin Du (Teachers College, Columbia University), Susan Buttigieg (Columbia University), Robin Nuzzolo (Fred S Keller School, NY), R. Douglas Greer (Professor Emeritus Columbia University Teachers College and Graduate School of Arts and Sciences), MICHAEL GAO (Alpaca Health), Bao Van (Duke University, Alpaca Health) |
| Abstract: Artificial intelligence (AI), the development of computer systems capable of performing tasks traditionally requiring human input, is becoming increasingly integrated into our daily lives. Although AI is still a relatively new topic within the behavior analytic community, it is gaining attention for its potential applications. This study explores the use of AI in data collection for instructional presentations across three educational programs: following vocal directions, gross motor imitation, and matching identical colors. Researchers videotaped a teacher working with a 5-year-old female in a one-to-one setting, within a quiet room with minimal distractions. The videos were then analyzed using Google Gemini Pro 002, a large language model (LLM), using multi-modal prompts, zero-shot prompting, and chain-of-thought prompting techniques. The AI-generated data on the accuracy of antecedents and student behavior were compared to human data collection. The preliminary results revealed mixed levels of interobserver agreement (33-87%, x= 66) for these particular educational programs. After feedback from two certified behavior analysts was used to prompt AI using more specific language, calibration increased (45-100%, x= 79%). The study is ongoing and results are discussed in terms of collaboration to
consistently achieve a minimum of 80% calibration for each component (antecedent,
behavior, consequence, data, and potential applications of AI to the field of behavior
analysis. |
| |
| Convergent and Divergent Validity of the Early Learner Curriculum and Achievement Record (ELCAR) |
| JeanneMarie Speckman (Fred S. Keller School Teachers College Columbia University), Jessica Singer-Dudek (Teachers College, Columbia University), LIN DU (Teachers College, Columbia University) |
| Abstract: The study investigated the psychometric characteristics of the Early Learner Curriculum and Achievement Record (ELCAR, Greer et al., 2020), a criterion-referenced curriculum and assessment tool for children’s development in language, academic, social and physical repertoires. The ELCAR is both an assessment tool and curriculum and is a comprehensive analytic tool used in the CABAS school system. We recruited 54 preschoolers (30 boys and 24 girls) diagnosed with autism spectrum disorders and other developmental delays and compared their ELCAR scores to other traditional psychometric measures. The ELCAR demonstrated convergent validity with the Preschool Language Scales (PLS-5). The results show moderate to strong positive correlations between the ELCAR total and sub-domain score and PLS-5 auditory and expressive scores. The ELCAR also demonstrated divergent validity with the Childhood Autism Rating Scale-2 (CARS2-ST). Results are broken down by ELCAR domain (e.g. listener, speaker) as well as child demographic (age, classroom ratio). Results indicate moderate to strong correlations between the ELCAR and the PLS-5 and CARS-2. |
| |
| The Effects of the Teacher Performance Rate and Accuracy Scale (TPRA) on the Presentation of Intact Learn Units for Newly Hired Teaching Assistants |
| ROBIN NUZZOLO (Fred S Keller School, NY), R. Douglas Greer (Professor Emeritus Columbia University Teachers College and Graduate School of Arts and Sciences), Lin Du (Teachers College, Columbia University), Susan Buttigieg (Columbia University), Jessica Pino (Fred S. Keller School) |
| Abstract: Behavior analysis is an evidence-based and effective method of instruction which can have enormous effects on a learner's repertoire. The potential progress of the learner is directly tied to the efficacy and efficiency of the instructor. Many school and home-based settings implementing behavior analysis rely on teaching assistants (TAs) and behavior technicians (RBTs) to deliver the majority of behavior analytic instruction to learners. We sought out to test the Teacher Performance Rate and Accuracy Scale (TPRA; Ingham & Greer, 1991; Ross et al., 2005) as a teacher training tool with new TAs who had no prior experience delivering ABA instruction and its effects on accuracy of teacher presentation of learn units (antecedents, behaviors, and consequences). During baseline, the supervisors modeled the presentation of target program with a student, then observed the trainee’s delivery of instruction of the same program with the same student without any feedback. The supervisors then observed the trainees in situ and used the TPRA, which provided specific feedback on the teacher delivery of learn units and modeling when necessary during intervention. Data demonstrated that 1-4 instances of the TPRA with trainer feedback were sufficient in bringing the trainees’ instruction to criterion level across a variety of programs. The results are discussed in terms of effective TA/RBT training, potential for distance training, and comparison to other interventions such as video recording and read-and-do. |
|
| |