Best Practices in Quantitative Methods for Developmentalists, Volume 71, Number 3

The role of quantitative methods in testing developmental hypotheses is widely recognized, yet even very experienced quantitative researchers often lack the knowledge required for good decision-making on methodology. The end result is a disconnect between research and practice in methods. The purpose of this monograph is to fill a gap in the literature by offering a series of overviews on common data-analytic issues of particular interest to researchers in child development. Our hope is that this monograph will make already developed methods accessible to developmentalists so they can understand and use them in their research. We start at the beginning with chapters on data management and measurement, two neglected topics in methods training despite the fact that every investigation should begin with proper consideration of each. We follow with two important topics for developmental research, missing data and growth modeling. Missing data can plague developmental work because participants sometimes miss one or more assessment points. Growth modeling methods offer researchers a true means to assess change over time as compared with cruder methods like difference scores and residualized change scores. Then comes a discussion of mediation and moderation, two tools that can be used to elucidate developmental processes. Because so much developmental science is non-experimental, we include a chapter on selection bias that compares five modeling strategies. Proper attention to data management, measurement, missing data, growth modeling (whenever possible), mediation and moderation, and potential selection bias is guaranteed to result in greater precision in inference-making. Even when researchers make good decisions about methods, it is critical for them to use good judgment about the practical importance of findings, so we conclude with this important discussion. We view this monograph as a first step to getting quantitative researchers started and we believe this reference will help researchers make better-informed decisions about methodology. BEST PRACTICES IN QUANTITATIVE.

METHODS FOR DEVELOPMENTALISTS.

CONTENTS.

EDITORS’ PREFACE vii.

INTRODUCTION TO THE MONOGRAPH.

Kathleen McCartney, Margaret Burchinal, and Kristen L. Bub 1.

I. DATA MANAGEMENT: RECOMMENDED PRACTICES.

Margaret Burchinal and Eloise Neebe 9.

II. MEASUREMENT ISSUES AND PSYCHOMETRICS METHODS.

IN DEVELOPMENTAL RESEARCH.

Richard G. Lambert, Lauren Nelson, Denise Brewer, and Margaret Burchinal 24.

III. MISSING DATA: WHAT TO DO WITH OR WITHOUT THEM.

Keith F. Widaman 42.

IV. GROWTH CURVE ANALYSIS: AN INTRODUCTION TO VARIOUS.

METHODS FOR ANALYZING LONGITUDINAL DATA.

Margaret Burchinal, Lauren Nelson, and Michele Poe 65.

V. CONTEMPORARY ADVANCES AND CLASSIC ADVICE FOR.

ANALYZING MEDIATING AND MODERATING VARIABLES.

Eric Dearing and Lawrence C. Hamilton 88.

VI. SELECTION, DETECTION, AND REFLECTION.

Kathleen McCartney, Kristen L. Bub, and Margaret Burchinal 105.

VII. THE PRACTICAL IMPORTANCE OF FINDINGS.

Roger Bakeman 127.

CONTRIBUTORS 146.

STATEMENT OF EDITORIAL POLICY 149

Roger Bakeman (Ph.D., University of Texas at Austin) is professor of psychology

at Georgia State University, Atlanta, Georgia. He is the author, with

J. M. Gottman, of Observing Interaction: An Introduction to Sequential Analysis

(2nd ed.; 1997), and, with V. Quera, of Analyzing Interaction: Sequential

Analysis With SDIS and GSEQ (1995). His interests include observational

methodology and sequential analysis of observational data.

Denise Brewer (Ph.D., University of North Carolina at Charlotte) is currently

an Assistant Professor at Appalachian State University in the Child

Development Department. She earned her Ph.D. in special education. She

also received her master’s degree from the University of North at Carolina

Chapel Hill in early intervention and family support and her undergraduate

degree from Appalachian State University in birth through kindergarten.

Research interests and background include assessment issues with

young children.

Kristen L. Bub (M.Ed., Harvard Graduate School of Education) is a fifth

year doctoral student in Human Development and Psychology. She earned

her master’s degree in human development, with a concentration in research

methods, from the Harvard Graduate School of Education. Her

research focuses on the role that early education experiences play in children’s

social and academic development.

Margaret Burchinal (Ph.D., University of North Carolina) is a Senior Scientist

and director of the Data Management and Analysis Center at the

Frank Porter Graham Center and Research Professor of Psychology at the

University of North Carolina at Chapel Hill. She is a methodologist who is

best known for her methodological work on longitudinal modeling as well

as for her substantive work on child care.

The role of quantitative methods in testing developmental

hypotheses is widely recognized, yet even very experienced

quantitative researchers often lack the knowledge required

for good decision-making on methodology. The end result is

a disconnect between research and practice in methods. The

purpose of this monograph is to fill a gap in the literature by

offering a series of overviews on common data-analytic

issues of particular interest to researchers in child development.

Our hope is that this monograph will make already

developed methods accessible to developmentalists so they

can understand and use them in their research. We start at

the beginning with chapters on data management and

measurement, two neglected topics in methods training

despite the fact that every investigation should begin with

proper consideration of each. We follow with two important

topics for developmental research, missing data and growth

modeling. Missing data can plague developmental work

because participants sometimes miss one or more assessment

points. Growth modeling methods offer researchers a

true means to assess change over time as compared with

cruder methods like difference scores and residualized

change scores. Then comes a discussion of mediation and

moderation, two tools that can be used to elucidate developmental

processes. Because so much developmental

science is non-experimental, we include a chapter on selection

bias that compares five modeling strategies. Proper attention

to data management, measurement, missing data, growth

modeling (whenever possible), mediation and moderation,

and potential selection bias is guaranteed to result in greater

precision in inference-making. Even when researchers make

good decisions about methods, it is critical for them to use

good judgment about the practical importance of findings,

so we conclude with this important discussion. We view this

monograph as a first step to getting quantitative researchers

started and we believe this reference will help researchers

make better-informed decisions about methodology.