A statistical method which will appeal to two groups in particular: those who are currently using the more traditional technique of exploratory factor analysis; and those who are interested in the analysis of covariance structures, commonly known as the LISREL model. The first group will find that this technique may be more appropriate to the analysis of their research problems; while the second group will find that confirmatory factor analysis is a useful first step to understanding the LISREL model, for this book, and its companion volume, Covariance Structure Models, are designed to be read consecutively. The proofs presented are simple, but the reader must feel comfortable with matrix algebra in order to understand the model.
The procedures, collectively known as discriminant analysis, allow a researcher to study the difference between two or more groups of objects with respect to several variables simultaneously, determining whether meaningful differences exist between the groups and identifying the discriminating power of each variable.
Recent advances in statistical methodology and computer automation are making canonical correlation analysis available to more and more researchers. This volume explains the basic features of this sophisticated technique in an essentially non-mathematical introduction which presents numerous examples. Thompson discusses the assumptions, logic, and significance testing procedures required by this analysis, noting trends in its use and some recently developed extensions.
Describes ARIMA or Box Tiao models, widely used in the analysis of interupted time series quasi-experiments, assuming no statistical background beyond simple correlation. The principles and concepts of ARIMA time series analyses are developed and applied where a discrete intervention has impacted a social system. '...this is the kind of exposition I wished I had had some ten years ago when venturing into the world of autoregressive, moving-average (ARIMA) models of time-series analysis...This monograph nicely lays out a method for assessing the impact of a discrete policy or event of some importance on behavior which can be continuously observed...If widely used, as I hope, it will save a generation of social scientists from the labor of having to learn this methodology the hard way...' -- Helmut Norpoth, State University of New York
The updated second edition offers expanded discussions of the chi square test of significance and the potential measures of association available for use with categoric data. Reviewing basic techniques in analysis of nominal data, this paper employs survey research data on party identification and ideologies to indicate which measures and tests are most appropriate for particular theoretical concerns. This book serves as an ideal primer for Volume 20, Log-Linear Models.
Provides an introduction to the fundamentals of scaling theory and construction, focusing on a variety of unidimensional scaling models. The authors present an overview and comparative analysis of such techniques as Thurstone scaling, Likert scaling, Guttman scaling, and unfolding theory, with emphasis on their varying conceptions of dimensionality. 'The aim of this series is to make the assumptions and practices of quantitative analysis more readily accessible to students and research workers with a limited background in statistics or mathematics...earlier works in the series certainly achieve this aim, and are, on the whole, lucidly written and of a generally high standard. The current two volumes maintain this standard.' -- Personality and Individual Differences, Vol 3 1982
Empirical researchers, for whom Iversen's volume provides an introduction, have generally lacked a grounding in the methodology of Bayesian inference. As a result, applications are few. After outlining the limitations of classical statistical inference, the author proceeds through a simple example to explain Bayes' theorem and how it may overcome these limitations. Typical Bayesian applications are shown, together with the strengths and weaknesses of the Bayesian approach. This monograph thus serves as a companion volume for Henkel's Tests of Significance (QASS vol 4).
Discusses the innovative log-linear model of statistical analysis. This model makes no distinction between independent and dependent variables, but is used to examine relationships among categoric variables by analyzing expected cell frequencies.
An introduction to the underlying principles, central concepts, and basic techniques for conducting and understanding exploratory data analysis -- with numerous social science examples.
