Optimizing Measurement of Communication in Autistic Children and Their Caregivers: Dependability of the...

Despite widespread use of caregiver-child interactions (CCX) to evaluate outcomes in autism interventions, little research has examined the dependability of this measure. This study applied generalizability theory to determine optimal conditions for dependable assessment of child communication and caregiver implementation of communication intervention strategies. Twenty caregiver-child dyads participated in structured interactions across six daily activities on two occasions. Young autistic child communication was measured via tele-health using the Individual Growth and Development Indicators (IGDI) and caregiver implementation using the Naturalistic Developmental Behavioral Intervention Fidelity (NDBI-Fi) scale. Both measures demonstrated strong generalizability when multiple observations were conducted, but significantly lower dependability with single observations. For child communication, three to five activities across two occasions provided optimal dependability (Φ = 0.69-0.82). For caregiver implementation, either four activities on one occasion (Φ = 0.80) or two activities across two occasions (Φ = 0.81) yielded dependable estimates. The greatest sources of error variance were person-by-occasion (54%-59%) and person-by-activity interactions (20%-28%), with all six activities contributing similarly to measurement error. Results indicate that researchers should prioritize collecting multiple CCX observations across different activities and occasions when sampling via telehealth. For optimal feasibility and generalizability, we recommend two observations of 3-4 play-based activities to balance measurement precision with participant burden and intervention relevance.