31373
‘Play As You Do at Home’: Using the Who Adapted Joint Engagement Rating Inventory (JERI) to Measure Caregiver-Child Interaction
Objectives: to examine observers’ agreement, accuracy, and construct validity of the WHO JERI
Methods: The sample consists of 86 children with ASD (67 males) aged 28-60 months recruited for a randomized controlled trial of the WHO Caregiver Skills Training program. Autism severity (ADOS-2), cognitive abilities (Griffiths-III) and vocabulary (MCDI) measures were collected at baseline. A free play caregiver/child interaction with a standard toy kit suitable for a range of developmental play levels was videorecorded. Ten consecutive minutes were rated both with a measure of social communication (BOSCC) and the WHO JERI by two reliable raters blind to the study’s hypotheses. Rating procedures were applied as per the manual for both measures. For the BOSCC, averaged scores were obtained for two 5-min scenes that were scored separately. For the WHO JERI, ratings of child engagement state (e.g., joint engagement); child behavior (e.g., initiation of communication, expressive language); caregiver behavior (e.g., following-in, affect); and the dyadic interaction (fluency and connectedness, shared routines) were applied to the entire 10-min scene in three separate viewings. Preliminary analysis were conducted on a partial sample (n=22, 17 males). To check agreement, 50% of the corpus was independently rated by a second observer. Estimated observer accuracy was computed using the KappaAcc program (Bakeman & Quera, 2011; Bakeman, 2018). Construct validity of the WHO JERI was investigated by examining the associations of the WHO JERI item scores with theoretically associated variables on other assessments.
Results: Observers' agreement was reasonable: the range of weighted Kappas for ratings within 1 point on the 7-point rating scale was .64–1. Observer accuracy was acceptable, with 8 out of 12 items showing accuracy higher than 99%. JERI Unengaged and Joint Engagement items show strong (> .5) positive and negative correlations with BOSCC and ADOS composite scores and items covering similar constructs (Spearman r’s range= .68 to .89 and -.76 to -.97, respectively). Correlations between JERI child items and corresponding BOSCC items (range = -.75 to -.85) were higher than correlations with expressive vocabulary or developmental level. Caregiver behavior items were weakly correlated with child skills, with the exception of expected relationships: e.g. Following-in was negatively correlated with repetitive behaviors (r=-.67). Interaction items were strongly correlated with the BOSCC engagement item (range = -.71 to -.86).
Conclusions: The WHO JERI shows reasonable inter-rater agreement. Examination of correlations supports the tool’s construct validity, as the patterns of both the direction and size of correlations shown among WHO JERI items and other study variables are consistent with previous studies and theoretical expectations.