Scoring the NBAS: To recode or not to recode
Lester (1984) proposed a set of six clusters for the standard 27 items of the National Behavioral Assessment Scale (NBAS). Nine of these items were recorded to change the scale values for the categories. This analysis examined the impact of recording on the psychometric properties of the NBAS. First, correlations were compared between recorded and nonrecorded items within the clusters. Second, iteritem consistencies were computed within each cluster to determine the extent to which the variables measured a common behavioral dimension. Third, a principal components analysis was done to determine the extent to which recoding makes the clusters more distinguishable. The results indicated that the original scale resulted in greater within-cluster correlations and higher interitem consistencies. The recoding method did not result in more distinguisable behavioral clusters. Overall, the original scale provided more optimal psychometric properties.
Neonatal Behavioral Assessment Scale, Principal components analysis, Psychometrics, Reliability, Scoring
McCollam, Karen M., Susan E. Embretson, Frances Degen Horowitz, and D. Wayne Mitchell. "Scoring the NBAS: To recode or not to recode." Infant Behavior and Development 19, no. 1 (1996): 63-69.
Infant Behavior and Development