Teacher Education, Educational Leadership & Policy ETDs

Publication Date

5-10-1971

Abstract

The major problem for investigation was to get useful information out of missing data, where missing data were defined as items not answered on a returned questionnaire. The author developed a scheme of analysis which can provide information not otherwise obtainable to a researcher using questionnaires. A model developed to categorize missing data was based on the reaction of the subject answering the questionnaire. The model rests on the notion the subject's avoidance reaction to the questionnaire item is his measurable attribute at that time. The mathematical development of the scheme of analysis uses the dichotomy of possible answers--the subject either answered the question, or he didn't. The statistical implications of this approach were considered using chi-square as the analytical tool. The problem of implementing the scheme of analysis on a computer was solved by using assembly language for differentiating blanks from zeros. The procedure used to test the analysis scheme involved both a hypothetical set of data points (eight variables for five groups of 25 or 26 subjects, each group having a random assignment of blank and non-blank responses) and a previously acquired set of real subject reactions (15 variables for four groups of 12 subjects, each group representing a different elementary grade level). The questionnaire given to the elementary grade students dealt with the subject's language usage (Spanish or English) with members of their families. The results of the analysis on the questionnaire given to the elementary students showed a significantly strong association with maternal grandparents together, such that association with the maternal grandmother tended to predict association with the maternal grandfather. There was no such significant association found with paternal grandparents. This finding was considered significant since it was not apparent until missing data analysis was performed. The major conclusions reached were that missing data analysis provides the researcher with information about his sample which he could not otherwise obtain, and that missing data analysis can be implemented on the computer.

Document Type

Dissertation

Language

English

Degree Name

Educational Leadership

Level of Degree

Doctoral

Department Name

Teacher Education, Educational Leadership & Policy

First Committee Member (Chair)

Stoughton Bell II

Second Committee Member

Dale Sparks

Third Committee Member

Albert William Vogel

Download

Share

COinS