Design and Analysis : A Researcher's Handbook. Geoffrey Keppel , Thomas D. The fourth edition of Design and Analysis continues to offer a readily accessible introduction to the designed experiment in research and the statistical analysis of the data from such experiments. Unique because it emphasizes the use of analytical procedures, this book is appropriate for all as it requires knowledge of only the most fundamental mathematical skills and little or no formal statistical background.
|Genre:||Health and Food|
|Published (Last):||16 June 2018|
|PDF File Size:||8.58 Mb|
|ePub File Size:||19.37 Mb|
|Price:||Free* [*Free Regsitration Required]|
Design and Analysis A Researcher's Handbook. Geoffrey Keppel Thomas D. Wickens University of California, Berkeley. Includes bibliographical references and index. ISBN alk.
Social sciences-Statistical methods. Factorial experiment designs. Social sciences-Research. Wickens, Thomas D.
K44 '. Credits and acknowledgments borrowed from other sources and reproduced, with permission, in this textbook appear on appropriate page within text.
W Jersey, Pearson Prentice Hall. All rights reserved. Printed in the United States of America. This publicar tion is protected by Copyright and permission should be obtained from the publisher prior to any prohibited reproduction, storage in a retrieval system, or transmission in any form or by any means, electronic, mechanical, photocopying, recording, or likewise.
For information regarding permission s , write to: Rights and Permissions Department. Pearson Education LTD. Contents 5 Analysis of '! Author Index Preface for a reader at that level to follow them and to adapt them to new data. Our reader must also be able to use the quantitative methods as part of an overall research plan. We have, therefore, set our numerical examples in context of plausible, if greatly simplified, research designs, so that we can describe both the statistical analysis and how that analysis serves as an inferential tool.
Although we have structured this book as a text, we also envision a second use. The books from which one learns technical material are those to which one returns to review and apply them. We know or at least hope that many of our readers will use This fourth edition of Design and Analysis is the result of a collaboration that began this book as the handbook of its title and will return to it over the years when using informally about 25 years ago, when TOW was spending a sabbatical year at Berkeley the analysis of variance.
With this application in mind, we have included extensive and attended GK's seminar on experimental design. We have continued our discus- coverage of the analytical analysis of the larger designs.
Indeed, we expect that some sions of research design and statistical analysis over the years. TOW was intimately of the later portions-'--Part VI, in particular-will gain their full usefulness only once familiar with the earlier editions of this book, using it as a text in his graduate courses the reader has had some practical research experience. Our col- The earlier editions of this book emphasized the computational aspect of the anal- laboration on this new revision is the natural extension of our discussions.
With our ysis of variance. Since they were written, the accessibility of computer programs has equal participation, we have created an essentially new book. We have rewritten it greatly increased. Quite sophisticated and comprehensive packages of statistical pro- throughout, reorganized the chapters, and introduced much new material. A researcher Our topic is the analysis of variance, the major statistical method used to analyze now will use a computer for any major analysis.
We have considerably modified this experimental data in psychology and the behavioral sciences. It gives the researcher edition of the book to accommodate this change. Thus, we have added Chapter 9 on a way to compare the means from a series of sets of scores and to determine whether computer applications, and we have completely rewritten our treatments of unequal they differ. More than just identifying overall differences, however, the analysis of sample sizes and the analysis of covariance Chapters 14 and 15 to emphasize the variance provides a very specific set of procedures that allow a researcher to express conceptual basis of these techniques.
Our treatment of the multifactor designs has particular questions about the study and to measure and test them. Much of the likewise been adjusted to be concordant with computer use. The changes here have power and the subtlety of the analysis of variance lies in these analytical techniques, let us emphasize the principles of the analysis instead of the arithmetic.
On the other which are necessary both when planning a study and when interpreting its results. Different laboratories are, quite appropriately, The early part of the book contains a detailed discussion of the foundation block of committed to different programs. Moreover, the programs differ too much from one to the analysis of variance-the experimental design with independent groups of subjects, another and change t'oo frequently for us to tie our discussion to a single program or each of whom contributes a single observation to the analysis.
We then elaborate this package. A description of a computer package that is nearly current can be extraodi- structure by expanding it in several ways-to studies in which the groups are classified narily confusing to a learner.
We suggest that an instructor supplement our treatment in two or more ways, to studies with muitiple observations, and to studies with several with examples of whatever computer programs are locally available.
By the end of this book, a student or researcher will be Notwithstanding the importance of the computer, we feel strongly that real facility able to put these parts together and apply them to the analysis of any complex design. Without with numerical examples, the analyses appropriate to them. Where possible, we cover experience with the actual computation at least of the simpler designs , a user's statistical arguments in the context of data analysis and interpretation. For example, skills are brittle-he or she has little ability to go outside the standard analyses or when we discuss the statistical assumptions that underlie an analysis, we both give to notice and accommodate unusual characteristics of the data.
With the greater their mathematical form and describe how they relate to experimental procedures. Real data sets require this flexibility. We wrote the book with a particular reader in mind: a student at the advanced This book is intended for use in a one-semester course or a two-quarter sequence undergraduate or beginning graduate level who is about to engage in experimental in experimental design and statistical analysis-for a shorter course, an instructor research, but who has little or no formal mathematical or statistical background.
We might treat just the single-factor and two-factor designs Chapters and What can we conclude from studying subjects in this single condition? Very little, except to describe the situation. We can infer nothing about the relative importance of the various characteristics in influencing the speed with which the maze is learned. Although this condition suffers from the same problem as the first when the behavior of the subjects is considered alone, the purpose of the experiment springs into focus when we compare the two treatment conditions.
More specifically, the critical difference between the two conditions is the difficulty of the maze, and any difference observed between the two conditions will be attributable to this particular difference hard versus easy.
Maze difficulty is the independent variable, and the pair of conditions An experiment consists of a carefully worked-out and executed plan for data collec- constitutes an experiment. We could have created additional two-group experiments tion and analysis. Treatment conditions are chosen to focus on particular features simply by varying one of the other critical features and holding the others constant, of the testing environment.
An appropriate source of subjects is established. These in which case the treatments might be food and water, satiated and hungry, or rat conditions are administered to subjects in such a way that observed differences in the average values can be unambiguously attributed to the differences among the various and hamster.
Most experiments consist of more than two treatment conditions. If we were treatment conditions. In essence, a well-designed experiment permits the inference of interested in comparisons among species in a given learning task, for example, we causation.
This chapter describes a number of important features of experimentation. We will rats, hamsters, and mice. The choice of animal would be dictated by our substantive consider throughout the interdependent problems of experimental design along with questions. If we were interested in the effects of food deprivation on learning, we our presentation of the formal statistical procedures. We will consider performance of some sort measured and recorded following the administration of the three categories based on the fundamental nature of the manipulation.
In this section, we will elaborate on this relatively simple idea. The first type of manipulation, qualitative independent variables or categor- ical independent variables, represents variations in kind or in type. Experiments The Independent Variable designed to study different methods for teaching reading or of different kinds of drugs In the simplest experiment, a researcher creates.
So is the study mentioned above and treats them differently. Frequently these groups are created by some operation that compares rats, hamsters, and mice.
When the qualitative independent variable performed by the researcher, such as forming a control group and an experimental involves more than two conditions, it can be viewed. Suppose an exper- male subjects. The critical difference between the groups is called the independent iment is designed to study the effects of different rewards on the speed of learning.
Once these groups have been defined, some form of behavior is observed Rats that have been deprived of food and water are assigned to one of the following and measured for each subject. The research question concerns how the average value three conditions: of this measured behavior depends on the group classification.
A comparison between conditions Studies in some fields of psychology examine naturally occurring groups that have 2 and 3 permits a similar determination of the effects of the addition of food to the received different treatments.
For example, an organizational psychologist may com- water reward. Most experiments involving a qualitative independent variable can be pare the incentive systems used in two different factories, or a health researcher may analyzed as a set of smaller, more focused experiments, a topic we will discuss in detail compare the relative effectiveness of two diet programs administered in different clinics.
Although these studies have much in common with an experiment-in particular, two A second type of manipulation, quantitative independent variables or con- different treatments are compared-they lack a critical component of a true experi- tinuous independent variables, consists of variables that represent variation in ment.
The treatments are applied to naturally occurring groups the factories or the amount-amount of food deprivation, variations in dosage, loudness of the masking clinics , and these may differ for various other reasons than the particular treatment. Variables of this sort usually include treatment Differences in productivity or in weight loss may have been influenced by other factors, conditions that define the range of values of interest to the researcher and several such as the kind of person who works at a particular factory or attends a specific clinic.
The effects of quantitative independent variables are usually analyzed As with the classification factors, the difficulties are not with the statistical methods differently from those produced by qualitative manipulations.
Here, researchers con- per se, but with the interpretation of their outcome. When using a quasi-experimental centrate on the overall relationship between the variation of the independent variable factor, a researcher must try to establish that the groups do not differ on whatever- and changes in behavior. The goal of the analysis is to determine the nature or shape incidental characteristics there may be that also affect the final scores. Suppose that the independent variable is the number of hours of food deprivation and that we will be measuring the trials required to learn a difficult The Dependent Variable maze.
How are we to describe the relationship? Presumably we will find an increase in performance as the animais become more hungry. But will the increase occur steadily Suppose we have completed the first steps in an experimental investigation: the de- as the number of hours increases, or will there be no effect at first, then a steady velopment of a meaningful hypothesis, the choice of the subjects, and the selection of increase?
Keppel & Wickens. Libro. 2004. Design and Analysis. A Researcher%27s Handbook-Pearson. 3.pdf
Design and Analysis A Researcher's Handbook. Geoffrey Keppel Thomas D. Wickens University of California, Berkeley. Includes bibliographical references and index. ISBN alk. Social sciences-Statistical methods. Factorial experiment designs.
ISBN 13: 9780135159415
View larger. Additional order info. The fourth edition of Design and Analysis continues to offer a readily accessible introduction to the designed experiment in research and the statistical analysis of the data from such experiments. Unique because it emphasizes the use of analytical procedures, this text is appropriate for the advanced undergraduate or beginning graduate student, as it requires knowledge of only the most fundamental mathematical skills and little or no formal statistical background. This book is also useful as a source and guide to application for researchers who require assistance in both planning a study and analyzing its results.
Design and Analysis: A Researcher's Handbook, 4th Edition
The fourth edition of Design and Analysis continues to offer a readily accessible introduction to the designed experiment in research and the statistical analysis of the data from such experiments. Unique because it emphasizes the use of analytical procedures, this book is appropriate for all as it requires knowledge of only the most fundamental mathematical skills and little or no formal statistical background. Topics include: single- and two-factor designs with independent groups of subjects; corresponding designs with multiple observations; analysis of designs with unequal sample sizes; analysis of covariance; designs with three factors, including all combinations of between-subjects and within-subject factors; random factors and statistical generalization; and nested factors. This book lives up to its name as a handbook, because of its usefulness as a source and guide to researchers who require assistance in both planning a study and analyzing its results. Designed to bridge the gap between elementary texts in statistics and experimental design and professional source books, this volume provides students with the basic information necessary to design and analyze meaningful experiments in the behavioral, social, and biological sciences. Fourth edition to be available Fall, Convert currency.