The Wiley Handbook of Developmental Psychopathology (eBook)

Luna C. Centifanti, David M. Williams (Herausgeber)

eBook Download: EPUB

2017
John Wiley & Sons (Verlag)
978-1-118-55453-1 (ISBN)

Infos

The Wiley Handbook of Developmental Psychopathology offers a concise, up-to-date, and international overview of the study of developmental psychopathology.

• Examines the cognitive, neurobiological, genetic, and environmental influences on normal and abnormal development across the lifespan
• Incorporates methodology, theory, and the latest empirical research in a discussion of modern techniques for studying developmental psychopathology
• Considers the legal, societal, and policy impacts of changes to diagnostic categories in the light of the transition to DSM-5
• Moves beyond a disorder-based discussion to address issues that cut across diagnostic categories

Luna C. Centifanti is a Senior Lecturer in the Department of Psychology at the University of Durham, UK and a Fellow of the Wolfson Research Institute for Health and Wellbeing. Dr. Centifanti is a member of the executive board of the Society for the Scientific Study of Psychopathy, and has been recognized for her longitudinal research with the Neville Butler Memorial prize in 2010. Her current research focuses on the developmental and psychophysiological correlates of aggression, decision-making, and callous-unemotional traits in both forensic and community samples of adolescents and young adults.

David M. Williams is a Professor of Developmental Psychology at the University of Kent, UK. His current research focuses on the neuro-cognitive underpinnings of, and relations among, developmental disorders. He is an Associate Editor at the journal Research in Autism Spectrum Disorder and, in 2010, he was presented with the International Society of Autism Research's Young Investigator Award for his research into metacognition in autism.

Textauszug

1
Developmental Trajectories of Psychopathology: An Overview of Approaches and Applications

Nathalie M. G. Fontaine and Isaac T. Petersen

Introduction

Developmental psychopathology concerns the study of the development of psychological disorders (e.g., depression, anxiety, schizophrenia, conduct problems), risk and protective factors, as well as outcomes, through a lifecourse perspective (Cicchetti, 1989; Rutter, 1990). Longitudinal studies, which involve repeated measures of the same variables from the same individuals, are crucial to investigate change (increases or decreases), but also stability, of psychopathology over time. Indeed, unlike cross‐sectional studies, in which different individuals are compared at one time point, longitudinal studies allow for (1) the exploration of within‐individual change (or stability)—that is, how each individual develops over time—and (2) between individual differences—including the investigation of distinct patterns of change (or stability) over time across individuals and factors associated with these distinct patterns (Singer & Willett, 2003).

More specifically, longitudinal data enable testing hypotheses about the development of behaviors, the developmental association between different, yet related behaviors (e.g., hyperactivity and physical aggression) and the factors associated with stability or change of behaviors over time. The identification of factors associated with persistence or change in behavior (increasing or decreasing patterns) can shed light on the vulnerabilities associated with severe and persistent psychopathology. In turn, a better understanding of risk processes in the development of psychopathology can help in the development of effective intervention strategies that target protective factors associated with desistance or decreased levels of psychopathology. Longitudinal data can therefore be useful for testing developmental theories about psychopathology as well as the effectiveness of prevention and treatment programs (Cicchetti & Toth, 1992).

New advances in statistical approaches over the last decades help in maximizing what we can learn using longitudinal data in the field of psychopathology (Nagin, 2005). Methodologists have developed various statistical approaches, which include and are known variously as growth curve models (GCMs), random coefficient models, multilevel models, mixed models, mixed‐effects models, hierarchical linear models, group‐based trajectory models (GBTMs), latent class growth models (LCGMs), and growth mixture models (GMMs) (Muthén, 2004; Nagin, 2005; Singer & Willett, 2003). The approaches have been applied with a rapid rise in many areas of developmental psychopathology research (Curran, Obeidat, & Losardo, 2010; Nagin & Odgers, 2010), including conduct problems (Nagin & Tremblay, 1999), depression/anxiety (Côté et al., 2009), callous‐unemotional traits (Fontaine, Rijsdijk, McCrory, & Viding, 2010), and substance use problems (Hu et al., 2008). These approaches have also been applied to assess heterogeneity in treatment responses to clinical trials (Muthén et al., 2002).

Statistical approaches for longitudinal data can be complex in terms of selecting the optimal approach, fitting the models to the data and interpreting the findings with respect to hypothesis and theory (Curran et al., 2010). In this chapter, we introduce some of the approaches and applications, particularly to non‐technical readers, including researchers, clinicians, and graduate students, who may not have yet received an extensive training in this area. References to more detailed and complete technical developments on these approaches are offered for interested readers. We first provide an overview of the approaches, with a focus on GCM, GBTM and GMM, namely approaches focusing on developmental trajectories (Muthén, 2004; Nagin & Odgers, 2010). Next, we present selected examples of applications of these models in the field of developmental psychopathology and clinical psychology. Finally, we discuss methodological considerations when applying these models and interpreting the findings.

Overview of the Approaches

This section presents an overview of three approaches applied to longitudinal data, and more specifically to trajectory modeling: GCM, GBTM and GMM. We selected these three approaches because they share a common analytical goal, namely to examine differences or variability across members of a population in their developmental trajectories (Nagin & Odgers, 2010). A developmental trajectory can be defined as the course of a behavior over time or age (Nagin, 1999). Because these trajectory analyses examine longitudinal data with more than two measurement occasions, they have key advantages over analytical approaches that do not examine trajectories (Beauchaine, Webster‐Stratton, & Reid, 2005; Gueorguieva & Krystal, 2004). First, trajectory analyses tend to have better reliability and greater power to detect behavioral change than simple pre–post or difference score designs. Second, trajectory analyses have greater flexibility with unbalanced designs, unequal spacing of time points, and tolerate missing data, unlike repeated measures analysis of variance (ANOVA). Third, trajectory analyses are less likely to inflate the Type I error rate than are repeated measures ANOVA analyses, which have more strict assumptions (e.g., sphericity). Fourth, trajectory analyses often allow multiple outcomes to be examined in the same analysis. Although these approaches share a common goal, they make distinct assumptions about the distribution of trajectories in the population. Figure 1.1 presents hypothetical trajectories according to the GCM, GBTM, and the GMM approaches. In a nutshell, it is assumed with GCM that all individuals come from the same population and can be described by the same parameters of change. It is not assumed, however, that individuals’ change is identical—the model captures the average developmental trend and person‐specific variations around the average trend using the same parameters of change. GBTM and GMM, by contrast to GCM, assume that some individuals come from distinct subpopulations, as captured by different subgroups. These subgroups can be described by different parameters of change. In other words, the models allow different individuals to follow different trajectories, but only GBTM and GMM allow subgroups (that are not captured by model predictors) of individuals with qualitatively different forms of change.

Figure 1.1 Hypothetical Trajectories According to the GCM, GBTM, and the GMM Approaches.

A number of differences exist between GCM, GBTM and GMM approaches. Researchers in developmental psychopathology often have to decide what approach they should apply. We present below a brief summary of the assumptions underlying each approach to help readers decide the optimal strategy for a given research question or hypothesis. Table 1.1 presents a summary of the key outputs of the three different approaches.

Table 1.1 Summary of the key outputs of the GCM, GBTM, and the GMM approaches.

GCM

GBTM

GMM

Intercept and slope of development

✔

✔ (for each trajectory)

✔ (for each trajectory)

Individual‐specific effects (random effects)

✔

✔ (for each trajectory)

Proportion of the population following each developmental trajectory

✔

✔

GCM

In their simplest form, GCMs typically fit a best‐fit straight line to each individual’s trajectory of change over time. Each individual’s line is allowed to have a different starting point (intercept) and direction and steepness of change (slope). Each individual’s best‐fit line is slightly adjusted to take into account the trajectories of the other individuals in the sample—a phenomenon known as shrinkage because individuals’ GCM estimates are shrunk towards the mean estimate for the sample, making the GCM estimates more reliable (i.e., having less measurement error) than estimates from simple regression (Hox, 2010). Based on theory and/or how well the model fits the data, the modeler can decide whether each parameter (intercept, slope) is the same (fixed effect) or allowed to differ (random effect) between individuals. GCM can be extended to consider nonlinear forms of change, such as polynomial (e.g., quadratic), exponential, and logistic forms.

GCM can be fit in a structural equation modeling (SEM) or hierarchical linear modeling (HLM; also known as multilevel modeling, mixed modeling, or mixed‐effects modeling; Raudenbush & Bryk, 2002) framework. In general, SEM is more advanced and flexible than HLM. SEM, unlike HLM, allows specifying latent variables that represent the common variance among observed (manifest) variables, and have less measurement error. Unlike HLM, SEM also allows specifying multiple outcomes in the same analysis and more flexibility in specifying correlated residuals (which, if residuals covary, would violate assumptions if unspecified). However, SEM typically requires a larger sample size than does HLM. In addition, HLM is more flexible when...

Produkt-Details

• Erscheint lt. Verlag: 24.8.2017
• Reihe/Serie: Wiley Clinical Psychology Handbooks
• Reihe/Serie: Wiley Clinical Psychology Handbooks
• Reihe/Serie: Wiley Clinical Psychology Handbooks
• Sprache: englisch
• Themenwelt: Geisteswissenschaften ► Psychologie ► Klinische Psychologie
• Themenwelt: Medizin / Pharmazie ► Gesundheitsfachberufe
• Schlagworte: Clinical psychology • Klinische Psychologie • <p>Developmental psychology, psychology, cognitive development, neurobiology, genetics, comorbidity, parenting, DSM 5, DSM V, mental health, health policy, diagnosis, psychiatry, marriage and family, abnormal development, child psychology, adolescent psychology, clinical psychology, speech therapy, language therapy, psychopathology</p> • Psychologie • Psychology
• ISBN-10: 1-118-55453-1 / 1118554531
• ISBN-13: 978-1-118-55453-1 / 9781118554531