Discriminant analysis between dropouts and non-dropouts in Newfoundland

• Main Author: Stack, Terence Leo
• Format: Thesis
• Language: English
• Published: Memorial University of Newfoundland 1973
• Subjects: Newfoundland
• Online Access: https://research.library.mun.ca/7427/; https://research.library.mun.ca/7427/1/Stack_TerenceL.pdf; https://research.library.mun.ca/7427/3/Stack_TerenceL.pdf

The purpose of this study was to develop a model which could be used in the identification and prediction of potential school dropouts. This model has attempted to determine the extent to which certain selected variables studied could all be used by schools in the future, special instrumentation was avoided. -- The variables used were the five Canadian Test of Basic Skills subtests Vocabulary, Reading Comprehension, English Usage, Mathematical Concepts, and Mathematical Problem Solving; school achievement in the form of grade average and overage variables; the two mental ability variables of verbal and non-verbal IQ from the Canadian Lorge-Thorndike Group Intelligence Test; father's occupational level; mother's level of education and father's level of education; number of natural parents in the home; two dwelling area variables community 1 and community 2; percent of time absent; and the nine variables involving teacher ratings of student self-control, courtesy, leadership, co-operation, attitude toward criticism, concentration, attention, tenacity and self-reliance. -- Twenty- three of these initial twenty-five variables were found to discriminate statistically between the two groups from the analysis of variance. The two exceptions, number of natural parents and community variable 1, were eliminated from further consideration. The school achievement variables of grade average and overage were also eliminated from the discriminant analysis because a new promotion policy in the district was eliminating the strength of these variables for future studies. -- The remaining twenty-one variables provided the model that discriminated between the two groups. The most important variables were verbal IQ, absence, self-reliance, co-operation, vocabulary and mother's level of education. These 6 variables accounted for 79 percent of the between groups variance between the dropouts and non-dropouts. This model would correctly classify 87.3 percent of the sample. -- An extreme difference was found between the verbal and ...