Cluster analysis of a pre-referral screening battery : with measures of phonological proficiency, self-concept, social perception, and moral reasoning

No Thumbnail Available
Daniel, Linda Lea
Gridley, Betty E.
Issue Date
Thesis (Ph. D.)
Department of Educational Psychology
Other Identifiers

The purpose of this study was to develop a screening battery that could be used as a pre-referral method of identifying students in need of formal assessment for special education services. Further, the battery was designed to provide guidance regarding what classification should be specifically explored in assessment. The screening battery under study included measures of phonological proficiency, self-concept, social perception, and moral reasoning. These variables were measured by the G-F-W Sound-Symbol Tests (Spelling of Sounds Subtest) (Goldman, Fristoe, & Woodcock, 1974), the Self-Perception Profile for Children (Harter, 1985), My Classmates and Me (a sociometric instrument), and the Moral Reasoning Scale for Children (Daniel, 1992). Research has demonstrated that the constructs measured by these instruments are related in differential degrees to classification status of students.Battery scores of 104 subjects (mean age=10.3 years) were cluster analyzed using Ward's method. Once distinct profiles were identified, they were compared to behavioral and educational histories of subjects to determine if the patterns were related to pre-determined classifications. Step-wise discriminant function analyses were conducted using measures of cognitive ability, reading achievement, emotional adjustment and attention as discriminatingvariables in an effort to externally validate the cluster solution.Cluster analysis of the four battery components yielded four clusters with distinctive profiles. Pair-wise comparisons of the four clusters indicated each was significantly different from the others based on performance on the screening battery (p<.001). These groups were identified as: Low Achievement, Pervasive Dysfunction, Normal, and Low Cognitive Ability. Although these cluster groups were not found to be strictly parallel to previously identified classifications, they were differentiated to varying degrees on several dimensions identified through discriminant analyses.The "Normal Group" was adequately differentiated from the other three groups on external validation indices. This suggested that the screening battery, which was used to establish groups, could predict the need for special education services of the subjects in the non-normal groups.