Dyadic Data Analysis

by Kenny, David A.; Kashy, Deborah A.; Cook, William L.; Simpson, Jeffry A.

• ISBN13: 9781572309869
• ISBN10: 1572309865
• Edition: 1st
• Format: Hardcover
• Copyright: 2006-07-28
• Publisher: The Guilford Press

Summary

Interpersonal phenomena such as attachment, conflict, person perception, helping, and influence have traditionally been studied by examining individuals in isolation, which falls short of capturing their truly interpersonal nature. This book offers state-of-the-art solutions to this age-old problem by presenting methodological and data-analytic approaches useful in investigating processes that take place among dyads: couples, coworkers, or parent-child, teacher-student, or doctor-patient pairs, to name just a few. Rich examples from psychology and across the behavioral and social sciences help build the researcher's ability to conceptualize relationship processes; model and test for actor effects, partner effects, and relationship effects; and model the statistical interdependence that can exist between partners. The companion website provides clarifications, elaborations, corrections, and data and files for each chapter.

Author Biography

David A. Kenny, PhD, is Board of Trustees Professor in the Department of Psychology at the University of Connecticut, and he has also taught at Harvard University and Arizona State University. He served as first quantitative associate editor of Psychological Bulletin. Dr. Kenny was awarded the Donald Campbell Award from the Society of Personality and Social Psychology. He is the author of five books and has written extensively in the areas of mediational analysis, interpersonal perception, and the analysis of social interaction data.

 

Deborah A. Kashy, PhD, is Professor of Psychology at Michigan State University (MSU). She is currently senior associate editor of Personality and Social Psychology Bulletin and has also served as associate editor of Personal Relationships. In 2005 Dr. Kashy received the Alumni Outstanding Teaching Award from the College of Social Science at MSU. Her research interests include models of nonindependent data, interpersonal perception, close relationships, and effectiveness of educational technology.

 

William L. Cook, PhD, is Associate Director of Psychiatry Research at Maine Medical Center and Spring Harbor Hospital, and Clinical Associate Professor of Psychiatry at the University of Vermont College of Medicine. Originally trained as a family therapist, he has taken a lead in the dissemination of methods of dyadic data analysis to the study of normal and disturbed family systems. Dr. Cook’s contributions include the first application of the Social Relations Model to family data, the application of the Actor-Partner Interdependence Model to data from experimental trials of couple therapy, and the development of a method of standardized family assessment using the Social Relations Model.

Table of Contents

Basic Definitions and Overview	1	(24)
Nonindependence	4	(2)
Basic Definitions	6	(8)
Distinguishability1	6	(1)
Between-Dyads, Within-Dyads, and Mixed Variables	7	(2)
Level of Measurement	9	(1)
Idiographic and Nomothetic Analyses	10	(1)
Dyadic Designs	11	(3)
Data Organization	14	(6)
Standard Design	14	(5)
Other Designs	19	(1)
A Database of Dyadic Studies	20	(1)
Overview of the Book	21	(2)
Summary and Conclusions	23	(2)
The Measurement of Nonindependence	25	(28)
Interval Level of Measurement	27	(13)
Distinguishable Members	27	(5)
Indistinguishable Members	32	(8)
Categorical Measures	40	(3)
Distinguishable Members	40	(2)
Indistinguishable Members	42	(1)
Consequences of Ignoring Nonindependence	43	(3)
What Not to Do	46	(2)
Power Considerations	48	(3)
Power of the Test of r	48	(2)
Power of Kappa	50	(1)
Summary and Conclusions	51	(2)
Analyzing Between- and Within-Dyads Independent Variables	53	(25)
Interval Outcome Measures and Categorical Independent Variables	54	(16)
One Between-Dyads Independent Variable	54	(6)
Multiple Between-Dyads Independent Variables	60	(1)
One Within-Dyads Independent Variable	61	(4)
Multiple Within-Dyads Variables	65	(2)
One Between-Dyads and One Within-Dyads Variable	67	(2)
General Case	69	(1)
Interval Outcome Measures and Interval Independent Variables	70	(4)
Between-Dyads Independent Variable	70	(2)
Within-Dyads Independent Variable	72	(2)
General Case	74	(1)
Categorical Outcome Variables	74	(1)
Within-Dyads Independent Variable	74	(1)
Between-Dyads Independent Variable	75	(1)
Summary and Conclusions	75	(3)
Using Multilevel Modeling to Study Dyads	78	(22)
Mixed-Model ANOVA	79	(3)
Multilevel-Model Equations	82	(3)
Multilevel Modeling with Maximum Likelihood	85	(2)
Adaptation of Multilevel Models to Dyadic Data	87	(11)
Computing the Intraclass Correlation and Variance Explained with Multilevel Models	93	(2)
Random Effects with Dyads	95	(1)
Distinguishable Dyads	95	(3)
Summary and Conclusions	98	(2)
Using Structural Equation Modeling to Study Dyads	100	(19)
Steps in SEM	101	(5)
Specification	101	(1)
Identification	102	(1)
Estimation	103	(1)
Testing	104	(2)
Confirmatory Factor Analysis	106	(2)
Path Analyses with Dyadic Data	108	(3)
SEM for Dyads with Indistinguishable Members	111	(5)
Model Fit and Model Comparisons	113	(1)
Example	114	(2)
Double-Entry Method	116	(1)
Summary and Conclusions	116	(3)
Tests of Correlational Structure and Differential Variance	119	(25)
Distinguishable Dyads	121	(14)
Equality of Variances	121	(1)
Equality of Correlations	122	(5)
Latent-Variable Correlations	127	(2)
Omnibus Test of Distinguishability	129	(2)
Categorical Variables	131	(4)
Indistinguishable Dyads	135	(6)
Correlations	136	(3)
Latent Variable Correlations	139	(1)
Comparing Intraclass Correlations	140	(1)
Variances	141	(1)
Summary and Conclusions	141	(3)
Analyzing Mixed Independent Variables: The Actor-Partner Interdependence Model	144	(41)
The Model	145	(2)
Conceptual Interpretation of Actor and Partner Effects	147	(5)
The Relative Size of Actor and Partner Effects	147	(2)
Partner-Oriented Interaction Effects	149	(1)
Actor-Partner Interactions	150	(2)
Estimation of the APIM: Indistinguishable Dyad Members	152	(18)
Pooled-Regression Method	152	(6)
Estimating the APIM with Multilevel Analysis	158	(10)
Estimating the APIM with SEM	168	(2)
Estimation of the APIM: Distinguishable Dyads	170	(9)
Pooled-Regression Method	170	(3)
Multilevel Modeling	173	(5)
Structural Equation Modeling	178	(1)
Power and Effect Size Computation	179	(2)
Specification Error in the APIM	181	(1)
Summary and Conclusions	182	(3)
Social Relations Designs with Indistinguishable Members	185	(38)
The Basic Data Structures	188	(3)
Round-Robin Design	188	(1)
Block Design	189	(1)
Block-Round-Robin Designs	190	(1)
Other Designs	190	(1)
Model	191	(20)
The SRM Components	191	(3)
Estimation of SRM Effects	194	(4)
The SRM Variances	198	(6)
The SRM Reciprocity Correlations	204	(4)
Self-Actor and Self-Partner Correlations	208	(1)
Actor and Partner Correlations with Individual-Difference Variables	209	(1)
Dyadic Analysis of Relationship Effects	210	(1)
Multivariate Correlations	210	(1)
Details of an SRM Analysis	211	(4)
Software	211	(1)
Statistical Issues	212	(1)
Significance Testing	213	(1)
Power	214	(1)
Model Assumptions	215	(3)
Social Relations Analyses: An Example	218	(3)
Summary and Conclusions	221	(2)
Social Relations Designs with Roles	223	(40)
SRM Studies of Family Relationships	224	(1)
Design and Analysis of Studies	225	(1)
The Model	226	(5)
The Components	227	(1)
Generalized and Dyadic Reciprocity	228	(1)
Estimation of Variance in the SRM with Roles	228	(1)
Family Subsystems	229	(2)
Application of the SRM with Roles Using Confirmatory Factor Analysis	231	(3)
The Four-Person Design	234	(5)
The Base Model	235	(4)
Illustration of the Four-Person Family Design	239	(6)
Model Evaluation	240	(1)
Parameter Estimates	240	(3)
Respecifications of the Base Model	243	(2)
The Three-Person Design	245	(7)
Base Model	245	(3)
Illustration	248	(2)
Alternative Three-Person Models	250	(2)
Multiple Perspectives on Family Relationships	252	(1)
Means and Factor Score Estimation	253	(6)
Constraints on the SRM Effects	257	(1)
Estimation of SRM Effects for a Particular Family	258	(1)
Power and Sample Size	259	(1)
Summary and Conclusions	260	(3)
One-with-Many Designs	263	(33)
Design Issues	264	(3)
Measuring Nonindependence	267	(4)
Indistinguishable Partners	267	(2)
Distinguishable Partners	269	(2)
The Meaning of Nonindependence in the One-with-Many Design	271	(1)
Univariate Analysis with Indistinguishable Partners	272	(12)
Naive Analysis	273	(2)
Between-Within Analysis	275	(2)
Multilevel Analysis	277	(7)
Univariate Estimation with Distinguishable Partners	284	(6)
Repeated-Measures Analysis of Variance and MANOVA	284	(1)
Multilevel Analysis	285	(2)
Structural Equation Modeling	287	(3)
The Reciprocal One-with-Many Design	290	(4)
Indistinguishable Partners	290	(3)
Distinguishable Partners	293	(1)
Summary and Conclusions	294	(2)
Social Network Analysis	296	(21)
Definitions	296	(2)
The Representation of a Network	298	(3)
Network Measures	301	(8)
Centrality	301	(3)
Reciprocity	304	(1)
Triads	305	(1)
Subgroups	306	(3)
Density	309	(1)
The p1 Model	309	(5)
The Model	309	(4)
Relation to the SRM	313	(1)
Another Look at the Sampson Data	314	(1)
Summary and Conclusions	314	(3)
Dyadic Indexes	317	(25)
Item Measurement Issues	319	(2)
Measures of Profile Similarity	321	(6)
Types	321	(4)
Which Measure to Use?	325	(2)
Mean and Variance of the Dyadic Index	327	(3)
Average Measure	327	(1)
Tests of Heterogeneity	328	(2)
Stereotype Accuracy	330	(4)
Differential Endorsement of the Stereotype	334	(1)
Pseudo-Couple Analysis	335	(2)
Idiographic versus Nomothetic Analysis	337	(1)
Illustration	338	(2)
Summary and Conclusions	340	(2)
Over-Time Analyses: Interval Outcomes	342	(39)
Cross-Lagged Regressions	344	(12)
Analysis Model	347	(1)
Computer Programs	348	(3)
Example	351	(5)
Over-Time Standard APIM	356	(3)
Specification of Error Structure	357	(1)
Measurement of the Causal Variable	358	(1)
Example	358	(1)
Growth-Curve Analysis	359	(11)
Definition of Time Zero	360	(1)
Functional Form	360	(3)
Structural Equation Modeling	363	(4)
Multilevel Modeling	367	(1)
General Error Models	368	(1)
Example	369	(1)
Cross-Spectral Analysis	370	(6)
Nonlinear Dynamic Modeling	376	(2)
Summary and Conclusions	378	(3)
Over-Time Analyses: Dichotomous Outcomes	381	(25)
Sequential Analysis	383	(5)
The Model	384	(2)
Illustration	386	(2)
Statistical Analysis of Sequential Data: Log-Linear Analysis	388	(8)
Logit Model	389	(3)
Significance Testing and Aggregation across Dyads	392	(2)
The Assumption of Stationarity in Time-Series Analysis	394	(1)
Beyond the Basic Analysis	395	(1)
Statistical Analysis of Sequential Data: Multilevel Modeling	396	(3)
Event-History Analysis	399	(5)
Censored Scores	401	(2)
Individual-Level Events	403	(1)
Discrete Analysis	403	(1)
Summary and Conclusions	404	(2)
Concluding Comments	406	(21)
Specialized Dyadic Models	406	(7)
Mutual Influence	406	(3)
Common-Fate Models	409	(4)
Going Beyond the Dyad	413	(3)
Triads	413	(1)
Groups	414	(1)
Network Autocorrelation	414	(2)
Conceptual and Practical Issues	416	(5)
The Meaning of Nonindependence	416	(4)
Design and Measurement Issues in Dyadic Research	420	(1)
Recruitment and Retention	421	(1)
The Seven Deadly Sins of Dyadic Data Analysis	421	(3)
The Last Word	424	(3)
References	427	(18)
Index	445