Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Records of 575 children evaluated for special education placement were analyzed to assess the descriptive validity of classification as practiced by Louisiana pupil appraisal teams. Classification categories, assigned by the assessment teams, included No Special Class, Learning Disability, Slow Learner, Mild Mentally Retarded, and Moderate Mentally Retarded. Results of discriminant analysis were evaluated in terms of the two attributes of descriptive validity, homogeneity and logical consistency of the categories. Clinical classification was judged satisfactory in homogeneity but unsatisfactory in logical consistency. within-groups homogeneity was demonstrated by significant differences among groups on seven variables summarized by the factors of Adaptive Behavior, IQ, Achievement, and Demographic Information. Classification accuracy, a quantitative estimate of homogeneity, was 65% agreement between clinical classification and group membership predicted from a statistical model of clinical classification. The comparison of clinical classifications with criteria mandated by the Louisiana guidelines for special education placement produced evidence of inadequate logical consistency. Overall agreement between clinical classification and classification by a computer program based on the guidelines (the Guideline Model) was almost 42%. The ELP criterion score ranges specified in the guidelines were frequently ignored by the assessors, possibly because the ELP measure lacked wide professional acceptance. Assessors also deviated from adaptive behavior and achievement guideline criteria. One serious shortcoming of clinical classification was the placement of over one-fifth of the retarded group without mandated evidence of inadequate adaptive behavior. In addition, clinical classification placed disproportionate numbers of black children in categories wth pejorative labels. Possible inadequacies in the Louisiana guidelines were identified by comparisons between Clinical, Guideline, and IQ Model classification schemes. First, classification of 75% of referred children as No Special Class would "mainstream" many children who would still require special help. Second, the use of ELP as a measure of learning potential did not completely eradicate demographic differences among groups, although this was its major purpose. Third, guideline classification accuracy was low for two of the four groups. Examination and discussion of clinical-guideline discrepancies was recommended. Dialogue among assessors in the field and guideline authors could improve both clinical assessment and the guidelines.