1
INTRODUCTION TO A DYNAMIC SYSTEMS APPROACH TO PSYCHOSOCIAL DEVELOPMENT IN ADOLESCENCE
Naomi M.P. de Ruiter, Mandy A.E. van der Gaag, Bertus F. Jeronimus, and E. Saskia Kunnen
Over the past decades, researchers who study adolescent development slowly reached an unnerving conclusion: much of the scientific knowledge on development that we have generated by studying groups, does not directly describe processes that unfold within individuals over time (this so-called non-ergodicity is explained in the next section). This is a shocking realization indeed, as one would expect developmental psychology to describe individual-level processes, and individual differences in change therein. Instead, most of what we have generated is knowledge about groups of individuals, and truths about averages that may or may not be true for any one individual (Blaauw, 2018; Molenaar, 2004). Subsequently we have a huge gap in our knowledge (see also Lichtwarck et al., 2008), which spans all domains of adolescent development. Uncertainty about whether and when group-to-individual generalizations are valid is particularly detrimental for our ability to effectively inform interventions (e.g., Fisher et al., 2018), as these typically deal with changes within individuals over time.
The lack of knowledge on individual processes in the social sciences is at least partly because many advanced tools were developed in the twentieth century to address differences between groups of people, whereas the development of techniques to study individual processes lagged behind (Lamiell, 1998; Rose, 2016). We lacked the practical tools to frequently measure individuals, we lacked the analytical tools to analyze such individual time-serial data, and we lacked the theoretical tools to understand these individual processes. These difficulties made individual process studies somewhat of a rarity.
But now, the field of research into adolescent development is changing in fundamental ways. Over the last decades, we have seen a rapidly increasing amount of research that is more person centered (rather than variable centered) and based on the analysis of processes over time, beyond the comparisons of groups. Technological developments have facilitated this change. There is currently a rapid development of tools for frequent data collection (e.g., apps for smartphones) as well as tools for the analysis of such data (e.g., Hamaker et al., 2015; Kunnen, 2012). And, importantly, theoretical developments have allowed us to understand individual change processes over time. One major theoretical approach that helps researchers in describing individual developmental process is the dynamic systems theory. This theory deals specifically with individual processes of change and stability over time, and is well established in the fields of physics, chemistry, and biology (Broer & Krauskop, 2000; Haken et al., 1990; Von Bertalanffy, 1968). In recent decades, pioneers such as Thelen and Smith (1994) and Van Geert (1994, 2008) used the dynamic systems approach as a theoretical tool to study individuals over time and zoom into processes that lead to change and stability within individuals. These studies showed the potential of dynamic system approaches in developmental psychology.
With this book, we aim to share some of the theoretical insights that have been gained from a dynamic systems approach to the study of psychosocial development in adolescence. In 2012, Kunnen published a book entitled A Dynamic Systems Approach to Adolescent Development, which presented new techniques for the study of adolescent developmental processes from a dynamic systems perspective, and claimed that this approach would generate fundamentally new insights and knowledge. The aim of the current book is to substantiate this claim. Although we cannot be exhaustive, the following chapters demonstrate a broad variety of applications of dynamics systems principles and methods to the study of adolescent development. Importantly, in doing so, this book demonstrates how the application of dynamic system principles contributed to theories in many important areas of adolescent development. Before we present the chapters and the rationale for including them, we first explain the problem of non-ergodicity in more detail, followed by a definition of what we consider to be a dynamic systems approach, and an overview of the consequences that both have for research.
Ergodicity and the relevance of a dynamic systems approach for adolescent development
The problem of ergodicity is becoming more and more salient in developmental research. Non-ergodicity means that the relation that exists between variables in a group of people may be different from the relations that exist within people (Molenaar, 2004, 2008; Molenaar & Campbell, 2009). The famous example involving typists illustrates this nicely: in a group of typists, a positive relation is found between speed and accuracy, because better typists are faster and more accurate (Hamaker, 2012). However, when this static group association is applied to a developmental process within one person (i.e., where we are interested in the change of x and the change of y for individuals), the relation between speed and accuracy is negative: if an individual begins to type faster, this is usually associated with an increase in mistakes. Furthermore, relations between variables can differ between individuals, as person A may be inclined to react to frustration with anger, while person B may become depressed. Thus, in person A, intra-individual correlations over time will show a positive relation between frustration and anger, but no relation between frustration and depression. In person B, frustration will correlate with depression, but not with anger. Furthermore, mechanisms may also differ within individuals over time. A frustrating event may be experienced as a challenge for person C, and result in an increase in motivation when she has slept well and has enough time, but it may result in a drop in motivation when she is tired or has too much to do already. Together, the continuum of differences between relationships at the group (between-person) and individual (within-person) levels, and differences in causal mechanisms over time, are all characteristic of non-ergodicity.
Non-ergodicity is sometimes also described as Simpson’s paradox, which is an illustration of how misleading averages can be. The paradox includes instances of erroneous conclusions drawn from group averages when causality is attributed or when “time” is not considered in correlational associations. The paradox has been found to be a common feature in findings concerning psychological processes (Fisher et al., 2018; Kievit et al., 2013; Molenaar, 2004). For example, a study by Keijsers (2016) about parent–adolescent relationships shows that, on a group-level, adolescents who are closely monitored by their parents show less delinquent behavior, whereas this relation is inverted on the individual level: an adolescent will show more delinquent behavior if the parents start monitoring more closely. Another example is a study by Van der Gaag and colleagues (2016) on micro-level identity development. They demonstrated that the commonly found relation between a large amount of exploration of alternatives and weaker commitments does not hold on the individual level: on an individual level this relation varied wildly, from being positive for one half of the individuals to negative for the other half.
The realization that the majority of psychological processes are non-ergodic has far-reaching consequences. It means, for example, that knowledge and insight that is derived from group analyses can often not be assumed to hold for individual development, which requires additional evidence. When studies find that, on average, positive development is related to a high level of some variable in a group of adolescents, it does not necessarily mean that increasing that variable in individuals will stimulate optimal development. Yet most of our knowledge nowadays is derived from group studies, and many intervention programs – explicitly aimed at development within individuals – are based on such studies. To address this issue, many researchers are becoming interested in studying developmental processes on an individual level as well. A dynamic systems approach can provide theoretical and methodological tools for studying individual processes and changes over time. As explained in detail below, this means the study of individual systems, which does not necessarily mean an individual person.
The ergodicity problem highlights the need for an individualized, process-oriented approach, which dynamic systems theory can offer. Additionally, a dynamic systems approach is ideal for the study of adolescent development in particular, as it is characterized by rapid and jumpy changes (irregular patterns, turbulence, bumps, delays, and regressions).
How can we define dynamic systems based research?
Given that the focus of this book is “dynamic systems based research” it is necessary for us to first define what we mean by dynamic systems based research. To start with, we would like to state that there is not one strict definition of “dynamic systems research” (for an overview of different schools and centers see https://en.wikipedia.org/wiki/Dynamical_system). Even within our own research group in Groningen – or indeed among the editors of this very book – there is no consensus. In fact, now that dynamic system approaches are become increasingly popular, this stream of research is becoming more diversified, development that we believe should be embraced and celebrated. Indeed, one of the objectives of this book is to illustrate this diversity. While we think that some criteria are necessary in order to provide guidance to interested researchers, these criteria are not exclusive. There are many ways of applying a dynamic systems approach to research, each with its own purpose and strengths. What is important is that all of these approaches share theoretical principles and some basic methodological characteristics that tie them together and make them “dynamic systems based research.” These principles and characteristics are outlined below as assumptions that characterize how dynamic systems research is done – as an approach, which may also form the basis of a dynamic systems theory of development.
Assumptions
First, we believe that the most important assumption in dynamics systems approach is that development and change is always individual based. This assumption reflects the phenomenon of non-ergodicity (described above). The individual-based approach to development does not necessarily refer to the study of an individual person, but pertains to the study of individual systems. An important first step in any dynamic systems study is deciding what will be the target system of the study. This choice depends on the phenomenon in question and can in principle be defined at any level: from the level of a cell, to an individual, to dyads, to groups, to whole societies, or cultures. After defining a system, the researcher may then ask whether or not relations between variables within that individual system (for example, within-person affect and autonomy) or trajectories of development of the system, may vary between different individual systems. Thus, from a dynamic systems approach, the objective does not have to be solely to understand individual development, the objective may also be to find similarities or differences between individuals in order to generalize a specific finding to a larger group. At the core of the individual-based assumption is that the individuals that make up a group are always the starting point of the analysis.
Second, a pivotal assumption in a dynamic systems approach is that development is inherently iterative, such that yt+1 = f(yt), causing a system to continuously evolve over time (see Figure 1.1). Simply put, this means that the...
