No abstract

This study utilized multiple regression analysis to develop equations to predict a committee's decision on whether to readmit flunked-out college students and to develop a second equation to predict grade point average the first quarter after readmitting such students. Data for similar students for a second academic year provided a hold-out group to cross-validate the regression equations. The committee's decision to readmit students could be predicted fairly well (cross-validity = .61) from the variables of setting realistic goals, math test score, number of quality points short of a passing average, and a self-analysis of failure. The attempt to predict the grade point average the first quarter after readmission was much less successful (cross-validity = .32). A different set of student factors seemed to be influential in accounting for a committee's decision to readmit a student and for the student's subsequent grade performance if admitted.

The purpose of this project was to develop tests to measure the six facets of Bloom's Taxonomy of Educational Objectives for the Cognitive Domain. Eight short stories with different themes were written, and tests were developed to measure Bloom's six different levels. After revisions of the items and stories, 487 college students read all 8 stories and answered multiple-choice items and vocabulary quizzes for each story. Factor and reliability analyses yielded inconclusive results regarding Bloom's taxonomy.

The major aim was to develop personality-inventory scales for the discrimination between two groups of problem drivers—traffic violators and accident-repeater drivers—and better-than-average drivers. A secondary aim was to attempt to improve prediction of problem drivers through multiple-regression equations. An experimental instrument including 395 items was developed, taking into account a very wide variety of personal qualities that had previously been found to characterize problem drivers or that were newly hypothesized to be potentially discriminating. The instrument was administered to approximately 2000 drivers. Problem-driver criterion groups were set up, each characterized by a different degree of seriousness of violation or accident record. Item analyses yielded scales that were found to be predictive on cross validation, also scales for the detection of test-taking biases. Multiple-regression equations were derived incorporating the scale scores with variables pertaining to biographical data and were cross-validated. Useful degrees of prediction were demonstrated from the scales alone and from the multiple-regression equations.

No abstract