Multiple comparisons / Alan J. Klockars, Gilbert Sax.

Format:

Book

Author/Creator:

Klockars, Alan J.

Contributor:

Sax, Gilbert.

Series:

Quantitative applications in the social sciences ; no. 07-061.

Sage university papers. Series, Quantitative applications in the social sciences ; no. 07-061

Language:

English

Subjects (All):

Multiple comparisons (Statistics).

Physical Description:

87 pages : illustrations ; 22 cm.

Place of Publication:

Beverly Hills : Sage Publications, [1986]

Summary:

Describes the most important methods used to investigate differences between levels of an independent variable within an experimental design. Readers will learn not only how to conduct multiple comparisons in experimental designs but also how to better understand and evaluate published research. "A highly readable introduction to multiple comparison methods, which demands little from its reader in the way of background other than some familiarity with analysis of variance." --The Statistician

Contents:

1. Multiple Comparisons 7

The Null Hypothesis 11

Variance Estimates or Mean Squares 11

The F-Ratio 13

Decision Rules and Statistical Significance 13

Decision Errors 14

Selecting the Most Appropriate Multiple-Comparison Test 15

Selecting a Multiple-Comparison Test and Type I Errors 16

Definitions and Meanings of Type I Errors in Multiple-Comparison Tests 16

Types of Multiple Comparisons 18

Test Statistics 19

2. A Priori Comparisons 22

Orthogonal Comparisons 22

Characteristics of Orthogonal Comparisons 24

Decision Rule for Evaluating Orthogonal Comparisons 27

Strategies for Developing Sets of Orthogonal Comparisons 28

Protection Levels for Orthogonal Comparisons 30

Orthogonal Polynomials 31

Nonorthogonal Comparisons 35

Summary and Recommendations 38

3. Post Hoc Comparisons: Pairwise Methods 39

Rationale for Range Tests 40

Sampling Distribution of Ranges 40

Tukey's Honestly Significant Difference Test (HSD) 43

Newman-Keuls Test 44

Tukey's Wholly Significant Difference Test (WSD) 47

The Duncan Test 47

Least Significant Difference Test (LSD) 50

Pairwise Comparisons and Type I Error 51

Comparisons of Pairwise Methods 53

Dunnett's Test 57

4. Post Hoc Comparisons: The Scheffe Test 58

Controls for Reducing Type I Errors in Post Hoc Comparisons 60

Conducting the Scheffe Test 60

Characteristics of the Scheffe Test 61

5. Multiple Comparisons in Factorial Designs 65

Familywise Type I Error 67

A Nonquantitative Approach for Examining Interactions 67

Simple Main Effects 69

Multiple Comparisons on Interactions 70

Range Tests and Interactions 71

Planned Comparisons and Interaction Effects 74

6. Additional Considerations 76

Unequal Sample Sizes 76

Violations of Assumptions 79.

Notes:

Bibliography: pages 85-86.

ISBN:

0803920512

OCLC:

14347977