Reliability Issues and Solutions for Coding Social Communication Performance in Classroom Settings.

Purpose: To explore the utility of time-interval analysis for documenting the reliability of coding social communication performance of children in classroom settings. Of particular interest was finding a method for determining whether independent observers could reliably judge both occurrence and d...

Descripción completa

Detalles Bibliográficos
Publicado en:Journal of Speech, Language & Hearing Research Vol. 49; no. 5; pp. 1058 - 1072
Autores principales: Olswang, Lesley B., Svensson, Liselotte, Coggins, Truman E.
Formato: Artículo
Publicado: American Speech-Language-Hearing Association October 2006
Materias:
Acceso en línea:Ver este registro en EBSCOhost
Descripción
Sumario:Purpose: To explore the utility of time-interval analysis for documenting the reliability of coding social communication performance of children in classroom settings. Of particular interest was finding a method for determining whether independent observers could reliably judge both occurrence and duration of ongoing behavioral dimensions for describing social communication performance. Method: Four coders participated in this study. They observed and independently coded 6 social communication behavioral dimensions using handheld computers. The dimensions were mutually exclusive and accounted for all verbal and nonverbal productions during a specified time frame. The technology allowed for coding frequency and duration for each entered code. Data were collected from 20 different 2-min video segments of children in kindergarten through 3rd-grade classrooms. Data were analyzed for interobserverand intraobserver agreements using time-interval sorting and Cohen's kappa. Further, interval size and total observation length were manipulated to determine their influence on reliability. Results: The data revealed interval sorting and kappa to be a suitable method for examining reliability of occurrence and duration of ongoing social communication behavioral dimensions. Nearly all comparisons yielded medium to large kappa values; interval size and length of observation minimally affected results. Implications: The analysis procedure described in this research solves a challenge in reliability: comparing coding by independent observers of both occurrence and duration of behaviors. Results indicate the utility of a new coding taxonomy and technology for application in online observations of social communication in a classroom setting. Reprinted by permission of the publisher.