New Statistical Procedures for the Social Sciences: Modern Solutions to Basic Problems

New Statistical Procedures for the Social Sciences: Modern Solutions to Basic Problems

New Statistical Procedures for the Social Sciences: Modern Solutions to Basic Problems

New Statistical Procedures for the Social Sciences: Modern Solutions to Basic Problems

Synopsis

This unique volume addresses the inadequacies of basic statistical methods that standard textbooks tend to ignore. The author introduces new procedures with accompanying tables that illustrate the practicality of the methods. Concentrating on basic experimental designs that are central to research in the social sciences, Wilcox describes new nonparametric techniques, two-way ANOVA designs, and new results related to the analysis of covariance and repeated measure design. This book serves as the ideal reference and supplement to standard texts by making the statistical advances of the last thirty years accessible to graduate students and researchers.

Excerpt

The first time I taught a graduate level analysis of variance course, I based my lectures on four recent and popular textbooks. My impression from these books was that not much had changed since I took the course in graduate school. However, when I started reading the journal articles on this topic, it soon became clear that many of the best known procedures should never be used. in particular, many journal articles had demonstrated that if certain commonly made assumptions are violated, procedures that are typically used can give highly unsatisfactory results. Moreover, new procedures had been developed for dealing with these problems, but there was no convenient way for students and researchers to learn these new techniques. the primary goal in this book is to supplement existing texts by bringing together many of these new procedures. Some procedures were previously impractical because appropriate tables were not available. This is the first book to include many of these tables.

A particularly important problem with the most frequently used procedures is the assumption that treatment groups have equal variances. It is probably fair to say that most researchers in the social sciences are more than willing to assume equal variances, yet among statisticians doing research on the analysis of variance, it is agreed that equal variances should not be assumed. in fact it has been demonstrated, using real data, that newer procedures that allow unequal variances can lead to drastically different results than those obtained using more conventional techniques where equal variances are assumed. What led to the widespread belief that equal variances can always be assumed, and why did statisticians change their mind about this view? How should the problem of unequal variances be addressed? Most researchers control power in terms of standardized differences, yet others find this approach to be completely unsatisfactory. What are the alternatives and what are their relative merits. What are the best multiple comparison procedures currently available? How should regression lines be compared in the analysis of covariance? These issues, and others, are discussed.

A difficulty with writing this book is that many results cannot be included because of space limitations. There is no doubt that nearly every statistician has opinions about what should be covered and what should be omitted in a book like this, and any two statisticians are not going to agree exactly on how to resolve this problem. Nevertheless, it is hoped that the new results covered in this book will be useful in many practical situations. One thing that should be evident from this book is that statistics is an evolving science. the approach used to solve the most basic questions may not be appropriate tomorrow.

Although the motivation for this book was to cover recent developments and to supplement existing texts, there is the problem that the training of students and researchers in statistics varies . . .

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.