Equilibrium Causal Models: Connecting Dynamical Systems Modeling and Cross-Sectional Data Analysis

Ryan, Oisín; Dablander, Fabian

doi:10.31234/osf.io/q4d9g

Cited by 7 publications

(2 citation statements)

References 75 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is not clear what a cross sectional bi-directional effect means without further study, perhaps contemporaneous change in A causes change in B while change in B causes change in A. Alternatively, these causal effects may have played out in the past shaping the current distribution of the data, but the variables may no long be amenable to change. A paper explores the limits of what can be learned about temporally unfolding processes from cross-sectional data (Ryan & Dablander, 2022) used the backshift algorithm (Rothenhäusler et al, 2015), which has clear similarities to the multi-group SEM model that we outlined above. Another apparent sources of indeterminacy that might arise the presences of interindividual differences in causal processes, which would require further adaptation of the models and put additional requirement on the data (e.g., the requirement of time-series data).…”

Section: Discussionmentioning

confidence: 99%

Three methods that advance tests of complex psychological theory using structural equation models.

Nivard¹,

Tamimy²,

Dolan³

2023

Preprint

View full text Add to dashboard Cite

Theories of psychological constructs like personality, pathology, or cognition, give rise to models that implicitly or explicitly impose a causal structure on mental processes. As there are rarely feasible ways to experimentally isolate and/or individually manipulate psychological variables, evaluating these causal structures is challenging. Competing causal theories, like latent variable and network theories, of psychological constructs are often evaluated based on correlational data, which rarely provide satisfactory conclusions on their relative merits. We discuss three methods that can improve our collective ability to distinguish between models derived from competing theories of psychological constructs. These methods significantly expand the set of models that are empirically identified in structural equation models (SEM) by respectively implementing penalized instrumental variable regression, modeled heteroskedasticity and modeled non-normality. To ease their application in psychological sciences we reformulate all three methods into SEMs. Once formulated as a SEMs, the core identification strategies can readily be adapted to the context of models applied widely in psychological science like growth-, (intense) longitudinal-, multivariate- or measurement-models.

show abstract

Section: Discussionmentioning

confidence: 99%

Three methods that advance tests of complex psychological theory using structural equation models.

Nivard¹,

Tamimy²,

Dolan³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Conceptually, a number of researchers in the causal modeling literature have shown that, under certain conditions, cyclic causal models fit to cross-sectional data may be interpreted as reflecting causal relations between equilibriums or resting states of a dynamic system (Iwasaki & Simon, 1994;Dash, 2005;Strotz & Wold, 1960;Spirtes, 1995;Mooij et al, 2013;Weinberger, 2021;Bongers et al, 2022). From this perspective, cyclic causal relations should be interpreted as a kind of coarse-grained or time-averaged representation of (reciprocal) causal re-lations between processes that evolve over time; for a detailed treatment of cyclic equilibrium causal models in the context of psychological modeling, we refer readers to Ryan and Dablander (2022).…”

Section: Introductionmentioning

confidence: 99%

Discovering cyclic causal models in psycho-logical research

Park,

Waldorp,

Ryan

2024

adv.in/psych

View full text Add to dashboard Cite

Statistical network models have become popular tools for analyzing multivariate psychological data. In empirical practice, network parameters are often interpreted as reflecting causal relationships – an approach that can be characterized as a form of causal discovery. Recent research has shown that undirected network models are likely to perform poorly as causal discovery tools in the context of discovering acyclic causal structures, a task for which many alternative methods are available. However, acyclic causal models are likely unsuitable for many psychological phenomena, such as psychopathologies, which are often characterized by cycles or feedback loop relationships between symptoms. A number of cyclic causal discovery methods have been developed, largely in the computer science literature, but they are not as well studied or widely applied in empirical practice. In this paper, we provide an accessible introduction to the basics of cyclic causal discovery for empirical researchers. We examine three different cyclic causal discovery methods and investigate their performance in typical psychological research contexts by means of a simulation study. We also demonstrate the practical applicability of these methods using an empirical example and conclude the paper with a discussion of how the insights we gain from cyclic causal discovery relate to statistical network analysis.

show abstract

Applying continuous-time models to ecological momentary assessments: A practical introduction to the method and demonstration with clinical data

Abplanalp,

Reavis,

et al. 2024

NPP—Digit Psychiatry Neurosci

View full text Add to dashboard Cite

Ecological momentary assessment (EMA) is a frequently used approach among clinical researchers to collect naturalistic data in real time. EMA data can provide insights into the temporal dynamics of psychological processes. Traditional methods used to analyze EMA data, such as hierarchical linear modeling and multilevel vector auto-regression, paint an incomplete picture of the dynamics of psychological processes because they cannot capture how variables evolve outside predefined measurement occasions. Continuous-time models, an analytical approach that treats variables as dynamical systems that evolve continuously, overcome this limitation. Time advances smoothly in continuous-time models, contrasting with standard discrete-time models in which time progresses in finite jumps. This paper presents a practical introduction to continuous-time models for analyzing EMA data. To illustrate the method and its interpretation, we provide an empirical demonstration of a continuous-time model utilizing EMA data of real-time loneliness and mood states (happiness, sadness, and anxiety) from a clinical sample comprising Veterans with a history of mental illness. Psychological variables, such as feelings of loneliness or sadness, can often change many times throughout the day. However, standard ways of analyzing these variables do not accurately capture these changes and fluctuations. Here, we highlight the benefits of continuous-time models, a method that can capture subtle changes in such psychological variables over time.

show abstract

Equilibrium Causal Models: Connecting Dynamical Systems Modeling and Cross-Sectional Data Analysis

Cited by 7 publications

References 75 publications

Three methods that advance tests of complex psychological theory using structural equation models.

Three methods that advance tests of complex psychological theory using structural equation models.

Discovering cyclic causal models in psycho-logical research

Applying continuous-time models to ecological momentary assessments: A practical introduction to the method and demonstration with clinical data

Contact Info

Product

Resources

About