Doctoral Dissertations

Eun Jung Oh

Date of Award

5-2000

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Education

Major Professor

Robert L. Williams

Committee Members

Katherine Greenberg, Bill Wallace, Ralph Brockett

Abstract

The purpose of this study was to evaluate the psychometric adequacy of a teacher rating scale of middle school students' work habits. This rating scale consisted of 20 work habits on which teachers rated selected students. A five-point Likert-type scale from a lot like this student to very unlike this student was used in rating the students on each work habit. Twenty-two teachers participated in the rating of 238 middle school students in 6th through 8th grades in 3 middle schools in the United States. Teachers rated the academic achievement of all students and the work habits of a subsample of students fi:om each class. Teachers were asked to identify the 2 students with the best work habits and the 2 with the worst work habits in each class and then rate those 4 students on their specific work habits. The data analysis focused on the item contributions, internal consistency, construct validity, and criterion validity of the rating scale. The main strength of the Student Work Habits Rating Form was its internal consistency, which was in the high .90s. This value greatly exceeds the .70 minimum traditionally required for research use. Thus, this rating form has considerable promise as a reliable means of assessing the work habits of middle school students. Teacher ratings of student achievement were used in assessing the criterion-related validity of the Student Work Habits Rating Form. The work-habits ratings for the 3 achievement groups judged as poor, average, or outstanding were significantly different. In particular, a logistic regression analysis showed that the work-habits ratings could classify correctly students in the poor and outstanding achievement groups virtually 100% of the time. Based on the result of the criterion-related validity of the Student Work Habits Rating Form, the rating scale appears to have substantial potential as a predictor of student performance. The construct validity of the Student Work Habits Rating Form was assessed through a series of factor analyses, culminating in Confirmatory Factor Analysis. This series of analyses points to a general factor of work habits rather than a set of primary and secondary factors. This rating form provides an overall measure of work habits that might be a useful predictor of students' long-term success. In its current form, it should not be used as a diagnostic profile of areas where work habits are strong or weak. Although this rating form appears to have considerable potential as a way to assess student work habits, all items on the rating form are not useful to an equivalent degree. Several conclusions can be reached from the various item analyses; (a) some items are more difficult than others with respect to receiving a favorable rating, (b) progression from low to high on the Likert-type scale is more difficult at the threshold from neutral to positive ratings than at the low end of the scale, and (c) most items discriminate best for students obtaining ratings toward the middle part of the work-habits distribution. As a comprehensive rating scale, the Student Work Habits Rating Form has strong internal consistency and moderately good criterion-related validity. It provides a solid overall assessment of student work habits. Some items appear to be psychometrically stronger than others, but all items (including the psychometrically weaker items) appear to have practical utility. In spite of the psychometric and practical potential of the Student Work Habits Rating Form, several features of this study limit the generalizability of the results and the clinical utility of the rating scale: (a) the specific habits identified in the rating form did not reflect some of the important habits mentioned in the literature; (b) the student sample rated was too limited and unrepresentative of different ethnic groups; (c) too few teachers were used as raters, and their ratings were not cross-validated by other teacher raters; (d) no test-retest reliability assessment for the rating scale was done; and (e) more diverse and precise criterion measures need to be used in validating the scale.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS