Methodology of Psychological Processes
Life is characterized by constant changes in affect, behavior, and cognition, as people grow, adapt, and interact with their environment. Hence, it is not surprising that researchers from all sub-disciplines within psychology share a key interest in the way processes unfold and accumulate over time. My Ph.D. projects are defined under an ERC Consolidator Grant awarded to Ellen Hamaker. The ERC project is titled: “Coming-of-Age of Process Research: Connecting Theory with Measurement and Modeling” (OPTIMAL) and aims at studying the pairwise links between theory, measurement, and modeling of psychological processes unfolding over multiple time scales.
The psychological phenomena (be it affective, cognitive, or behavioral) can be studied from a relatively large number of measurements—which may be collected via, e.g., self-report questionnaires in the experience sampling method (ESM) and activity trackers—over a set amount of time. Such measurements are often called intensive longitudinal data (ILD) and call for more advanced statistical techniques. My Ph.D. projects approach the methodology of psychological processes from a clinical point of view and aim to provide answers to questions stemming from applied research.
The first project contributes to the measurement–modeling connection, in which I study the extent to which skewness leads to biased estimates. Many of the statistical techniques currently being used are based on the assumption that the data are normally distributed. However, in practice, many ILD measurements violate this assumption. One situation in which this is typically the case is with ESM measures of symptoms or negative affect, which are characterized by a strong floor-effect, wherein the lower responses are more common, leading to high positive skewness. This may lead to spurious results: For instance, it has been suggested that this restriction of variance at the lower end of the scale causes the replicated finding that symptoms are less strongly connected in individuals with lower average scores on those symptoms (cf. Terluin et al., 2016). I make use of lesser-known data-generating mechanisms to simulate ILD that mimic the known characteristics of empirical data well, such as (a) individual differences in mean; (b) individual differences in skewness related to individual differences in mean; and (c) individual differences in autoregression. Then, via multilevel dynamic modeling, I investigate whether the individual differences in autoregression or cross-lagged regressions can result from the restriction of range or if this concern is, in fact, ungrounded.
The second project brings the psychological literature about processes under the spotlight. In this project, I will explore the definition of psychological disorders as provided in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5; American Psychiatric Association, 2013), which forms an essential system for defining psychopathology in both research and practice. I will consider the descriptions of symptoms and evaluate which of these symptoms make reference to some form of dynamics. For instance, in depression, an important symptom is “Markedly diminished interest or pleasure in all, or almost all, activities most of the day, nearly every day,” which from a dynamic viewpoint implies that there is a lack of responsiveness to events. Similarly, important symptoms in posttraumatic stress disorder are emotional distress and physical reactivity after exposure to traumatic reminders, which also clearly pertains to a dynamic pattern that exposes itself over time. I will systematically evaluate all symptoms described in the DSM-5 and determine the extent to which symptoms make reference to particular patterns over time, develop broad classes of different dynamic features, and investigate whether particular classes of disorders make more reference to dynamics than others. This project aims to evaluate to what extent dynamic thinking is inherent to defining psychopathology and how this can be recognized. Additionally, this study will provide vital clues on measuring and modeling psychopathology as defined in the DSM-5.
The third project contributes to the theory–modeling connection, in which I study the cyclical and regime-switching processes. Many psychological processes are described in terms of fluctuations that may have some smooth regularity to them, as in circadian rhythms and other cycles, or that consist of more sudden switches, as in regime-switching. I will investigate, both conceptually and practically, the differences between these processes through (a) investigating the diverse patterns these processes can generate and (b) investigating through simulations how well these two models can be distinguished, and how this depends on the number of time points (T) and the number of cases (N).Finally, my fourth project contributes to the theory–measurement connection by means of studying interval measures and continuous-time processes. Although many processes can be conceptualized as taking place in continuous time, oftentimes, our measurements of these processes pertain to an interval, for instance, a five-second window in observational ratings (cf. Gottman et al., 2005), a fifteen-minute interval in physiological data (cf. Houtveen et al., 2010), or the entire day in self-report (Laurenceau et al., 2005). It is unclear what the consequences of this discrepancy between theory and measurement are. I will discuss this discrepancy from a conceptual point of view and investigate through simulations what its consequences are for our ability to study particular process features that stem from the theory directly.
Prof.dr. E. Hamaker
Dr. Oisín Ryan
ERC Consolidator Grant 2019