This new combination volume of three-books-in-one, dealing with the topic of artifacts in behavioral research, was designed as both introduction and reminder. It was designed as an introduction to the topic for graduate students, advanced undergraduates, and younger researchers. It was designed as a reminder to more experienced researchers, in and out of academia, that the problems of artifacts in behavioral research, that they may have learned about as beginning researchers, have not gone away.
For example, problems of experimenter effects have not been solved. Experimenters still differ in the ways in which they see, interpret, and manipulate their data. Experimenters still obtain different responses from research participants (human or infrahuman) as a function of experimenters' states and traits of biosocial, psychosocial, and situational origins.
Experimenters' expectations still serve too often as self-fulfilling prophecies, a problem that biomedical researchers have acknowledged and guarded against better than have behavioral researchers; e.g., many biomedical studies would be considered of unpublishable quality had their experimenters not been blind to experimental condition.
Problems of participant or subject effects have also not been solved. We usually still draw our research samples from a population of volunteers that differ along many dimensions from those not finding their way into our research. Research participants are still often suspicious of experimenters' intent, try to figure out what experimenters are after, and are concerned about what the experimenter thinks of them.

Contrasts are statistical procedures for asking focused questions of data. Researchers, teachers of research methods and graduate students will be familiar with the principles and procedures of contrast analysis included here. But they, for the first time, will also be presented with a series of newly developed concepts, measures, and indices that permit a wider and more useful application of contrast analysis. This volume takes on this new approach by introducing a family of correlational effect size estimates. By returning to these correlations throughout the book, the authors demonstrate special adaptations in a variety of contexts from two group comparison to one way analysis of variance contexts, to factorial designs, to repeated measures designs and to the case of multiple contrasts.

This book, first published in 1985, is for those seeking appropriate statistical approaches to research data that is devoted entirely to the topic of contrasts. Contrast analysis permits us to ask more focused questions of our data. In return for a small amount of simple computation, we get greater statistical power, and can make clearer substantive interpretations of the research results. Contrast analysis should be employed in the context of the analysis of variance whenever the numerator degrees of freedom are greater than one (which is probably most of the time). Unfortunately, it is employed relatively rarely by behavioural and social scientists. This book makes it possible for non-mathematical data analysts to avail themselves of contrasts, and thereby simply and efficiently to address the focused questions posed by their theories, hypotheses, and hunches. A wide range of researchers in the behavioural and social sciences, education and business will find this book an invaluable resource.

Contrasts are statistical procedures for asking focused questions of data. Researchers, teachers of research methods and graduate students will be familiar with the principles and procedures of contrast analysis included here. But they, for the first time, will also be presented with a series of newly developed concepts, measures, and indices that permit a wider and more useful application of contrast analysis. This volume takes on this new approach by introducing a family of correlational effect size estimates. By returning to these correlations throughout the book, the authors demonstrate special adaptations in a variety of contexts from two group comparison to one way analysis of variance contexts, to factorial designs, to repeated measures designs and to the case of multiple contrasts.