Experiment Design and Statistical Methods For Behavioural and Social Research

David R. Boniface

• Publisher: Routledge
• Year: 2019
• eBook ISBN: 9781351449298
• Topic: Matematica
• Subtopic: Probabilità e statistica

Experiment Design and Statistical Methods introduces the concepts, principles, and techniques for carrying out a practical research project either in real world settings or laboratories - relevant to studies in psychology, education, life sciences, social sciences, medicine, and occupational and management research.
The text covers:

repeated measures

unbalanced and non-randomized experiments and surveys

choice of design

adjustment for confounding variables

model building and partition of variance

covariance

multiple regression
Experiment Design and Statistical Methods contains a unique extension of the Venn diagram for understanding non-orthogonal design, and it includes exercises for developing the reader's confidence and competence. The book also examines advanced techniques for users of computer packages or data analysis, such as Minitab, SPSS, SAS, SuperANOVA, Statistica, BMPD, SYSTAT, Genstat, and GLIM.

Statistical Design and
Analysis for Basic
Experiments

PART
1

1	Introduction

1.1 STRUCTURE AND SCOPE OF PART ONE

1.1.1 Structure

This chapter sets out the framework in which the material of this part of the book is located and identifies the aims of the design of experiments.

Chapter 2 presents examples of each of the four experiment designs dealt with. It includes an introduction to some of the concepts and issues relevant to them.

Chapter 3 presents the concepts of design and analysis for experiments in a degree of detail sufficient for understanding the later material.

Chapters 4,5,6,7 each deal in detail with one of the four designs that were introduced in Chapter 2.

Chapter 8 extends the analysis of the designs of Chapters 4,5,6,7 to suit them to particular research issues which occur commonly in practice.

Chapter 9 is concerned with the number of individuals to be included in the research and the choice of appropriate design.

1.1.2 Scope

Part One introduces designs, analyses, principles and techniques for comparing alternative conditions in experimental research.

In all experiments dealt with it is assumed that the response of the individuals taking part is measured on a continuous scale. A continuous scale is one in which the numerical values refer to an underflying continuum of amount or quantity. It is further supposed that the measurement scale has the equal value interval property (i.e. one unit has equal value over the whole scale).

The reader is assumed to have completed a basic non-mathematical course in statistical methods and to be familiar with the basic ideas of hypothesis testing, t-tests, correlation and regression.

1.2 INFERENCE FOR DESCRIPTIVE AND EXPERIMENTAL RESEARCH

Descriptive research is essentially an exercise in gathering data. The data may be gathered by direct observation, questionnaire or some other method. Some considerable intervention in the lives of individuals may be involved: for example, they may be asked to keep a diary or follow a special diet. Such intervention is made only to provide the conditions under which the observations are to be made; the intervention is not made in order to provide a comparison with the absence of intervention or with some alternative form of intervention.

In descriptive research the design could take one of several forms. It may be a case study; for example, an account of the development of speech in a child with a particular learning difficulty. It may be a study of a sample of individuals; for example, a survey of the extent of examination nerves in a sample of students.

Sometimes research is carried out with very limited aims. A nursing manager may want to carry out a small research project whose end result will be an improved oganization of a hospital ward. In this case there may be no intention to generalize the results of the research to other hospital wards. Very often, however, the researcher wishes to obtain knowledge from the research which can be applied elsewhere. This is true whichever form of descriptive research design is used. In other words, the researcher intends the findings of the particular study to be generalized to other individuals or situations.

Generalizing the results of research can be based on common-sense judgements of the similarity of situations. Such judgements have an important place in scientific work. However, there is also available a formal method for generalizing the findings from descriptive research. This is the method of statistical inference.

Statistical inference uses the mathematics of probab...

Table of contents

• Cover
• Title Page
• Copyright Page
• Table of Contents
• Preface
• Part One: Statistical Design and Analysis for Basic Experiments
• Part Two: Unbalanced, Non-Randomized and Survey Designs
• Part Three: Analysis for Further Experiment Designs
• Appendix A: Hints on use of computer programs
• Appendix B: Additional exercises for Chapters 5–13
• Appendix C: Solutions to exercises for Chapters 4–13
• Appendix D: Approximate degrees of freedom for test of significance for simple effects in BW and WW designs
• Appendix E: Rationale for approximate sample size formula
• Appendix F: Tables of critical values
• References
• Index