Robert Cudeck - US grants

This project will develop statistical theory that is needed to estimate the sampling variability of parameter estimates in the exploratory factor analysis model, will implement the theory in a computer program to be made available to interested researchers, and will apply the method in a series of reanalyses to several data sets collected to study the structure of personality. Much of the work to be performed is technical; however, the immediate practical benefit will be inferential tools available for the first time in factor analytic research. The work fills a void that has existed for decades in the literature. With the program one will be able to evaluate a variety of hypotheses, construct interval estimates for single parameters, and confidence regions for groups of parameters. Although the proposed research has a statistical component, it was motivated by its relevance for addressing empirical measurement problems. Consequently, a second objective is to apply the new methods in reanalyses of data sets from the domain of personality measurement. A series of studies is proposed. We will systematically re-analyze several data sets generously made available by Prof. L. R. Goldberg in his investigation of the Big Five factor model of phenotypical personality characteristics. Available are data recently collected by Tellegen that also bear on the same topic. The objective of the study is to use the new statistical tools to identify robust markers of the factors underlying these data. The hope. is to achieve homogeneous and coherent subsets of variables across the samples. Because procedures used in these analyses will be based on sound statistical methodology applied in a consistent fashion, it should be possible to obtain solutions that are especially clear. This should be valuable in contributing to the understanding of the nature and composition of the Big Five.

Funding from NSF will support a special occasion conference, "Factor Analysis at 100: Historical Developments and Future Directions." The conference marks the 100th anniversary of the publication of Charles Spearman's seminal and monumental article, "General Intelligence, Objectively Determined and Measured," in the American Journal of Psychology, which outlined the framework of a new statistical tool in support of his psychological theory about intelligence. The statistical method was factor analysis. In the 99 years since it was introduced, factor analysis has become as integral to the development of psychological science as any method or procedure used in the study of human behavior. Vigorous methodological research and development continues today on factor analysis and its various extensions and generalizations, and these methods serve a critical role in substantive research in many areas of psychology as well as other sciences. The overarching objective of the conference is to examine the history, current state and significant issues, and future directions of the theory, methodology and application of factor analysis. Through this examination, we hope to clarify the state of the art and future directions of research, and to facilitate and enhance the quality of continued research with respect to both methodology and application. These objectives will be achieved by gathering together the most prominent researchers in the field along with an audience of scientists and students with a special interest in learning more about the field and carrying on the development and proper use of factor analysis and related methods. The bulk of the funding provided by NSF will be used to promote and support attendance and participation at the conference by graduate students, university faculty, and researchers with an interest in the theory and application of factor analysis.