A meta-analysis of locus of control research appearing in the journals of the American Psychological Association from 1963 through 1977.

Abstract

A narrative and descriptive-statistical review of findings receiving unequivocal support across studies was also presented. The review was provided organization by an assignment of studies to content categories. The reliability of the assignment procedure was found to be .74.

The following conclusions were drawn: (1) the meta-analysis of research method may be productively applied to the review of studies heterogeneous with respect to variables measured, (2) the reliability of study or article-rating scales may be enhanced by the development and utilization of specific scoring criteria, and (3) the construct validity of the LC construct received little support from the quantitative analyses conducted in the present study.

The research literature in psychology and education had expanded at an astounding rate in recent years, but the traditional narrative review was considered incapable of organizing, analyzing, and synthesizing the vast bodies of data that had accumulated. An improved, more systematic and objective method of review, the meta-analysis of research, had been developed, but it had been applied only to studies homogeneous with respect to variables measured, and it assigned equal weight to the findings of studies varying widely with respect to important sampling, design, and analysis considerations. The present study investigated possibilities for a meta-analysis of studies heterogeneous with respect to variables measured, and took into account, also, variations in quality of studies. The locus of control research literature was selected for the analysis, as research had been expanding rapidly in this area, but objective, systematic analyses had not been undertaken.

Since a systematic review of some 2200 locus of control (LC) studies was not feasible, analysis was confined to a sample of the literature. All LC studies appearing in the journals of the American Psychological Association from 1957 through 1977 which met certain specified criteria were included in the study. No studies published before 1963 qualified for the sample. The sample consisted of 150 studies.

Investigations of the construct validity of the LC construct, as well as possible differentials related to variations in quality of studies, LC measurement procedures, design of studies, source of derivation of hypotheses, and criteria for LC group assignments, were accomplished through examinations, by means of the chi square test statistic, of differences in proportions of positive to negative findings across categories-developed. A total of 10 chi square tests were conducted. Type I error probability was set at alpha for the collection of comparisons. Of the 10 comparisons, only one attained significance at the .01 level.

Levels of quality of studies were assessed by means of the Quality of Study Rating Form, developed for the study. The development and utilization of point-by-point scoring criteria was found to improve the reliability of the form, which was found to be .73 for the final form used in the study.