Search

Item does not contain fulltext This chapter discusses theoretical foundations of network analysis by outlining different ways in which networks can be used. In the broadest of these ways, often denoted the network approach, the researcher views a substantive phenomenon through the lens of networks. In this case, networks mainly function to organize observations and to suggest theoretical ideas. The network approach can subsequently be specified in at least two ways. First, by constructing psychometric network models, which formulate a probability distribution for a set of observations, typically by representing variables as nodes and conditional associations between these variables as edges. Second, by constructing network theories, which offer putative explanations of empirical phenomena. Thus, where network theories are tied to a particular empirical domain, network models are generic, i.e., independent of any particular domain. Finally, the chapter discusses ways in which the relation between network approaches, network theories, and network models can be understood.

This chapter discusses time-varying network models, which allow for network models with parameters that change across time. These models are relevant when one is interested in how individuals change across time. The chapter introduces the reader to time-varying models and discusses various methods to estimate them, which make different assumptions about the true time-varying nature of parameters. Finally, the chapter illustrates one of the methods by fitting a time-varying Gaussian graphical model (GGM) to a time series of mood related measurements.

This chapter introduces the reader to studying psychological dynamics operating on network models. The chapter will illustrate such dynamics using the Ising model, an undirected network model for binary variables in which nodes mutually influence one-another. First, the chapter discusses the basics of the Ising model. Second, the chapter introduces the reader to dynamics emerging from the Ising model, such as polarization and hysteresis. These dynamics will be illustrated using the example of attitude networks. Third, the chapter will illustrate how the Ising model can be used to model cross-sectional phenomena as an alternative to latent trait theories. General intelligence will be used as an example for this illustration to show how the positive manifold and block structures in correlation matrices can arise from a network implemented in the Ising model.

This chapter introduces pairwise Markov random fields (PMRFs), a class of models of which the parameters can be represented as an undirected network. In this undirected network nodes represent variables and edges represent the strength of association between two variables after conditioning on all other variables included in the model. The chapter focuses on specific classes of PMRFs often used in network psychometrics: Gaussian graphical models (GGM; a network of partial correlations) for continuous data, Ising models for binary data, and mixed graphical models (MGM) for data with different types of variables. PMRFs can be interpreted in multiple ways: the models can be used as a general statistical modeling framework, as an exploratory tool to investigate predictive relationships between variables, as a tool to generate causal hypotheses, as a causal model itself, and as an exploratory tool to uncover latent variables. The chapter concludes with an introduction to estimating PMRFs from data using the bootnet and psychonetrics packages.

Het ministerie van VWS neemt verschillende maatregelen om de verspreiding van het coronavirus (SARS-CoV-2) onder controle te krijgen. Zo is het belangrijk dat mensen zich laten testen bij klachten of na contact met een besmet persoon. Een onderdeel van de aanpak van VWS is dat grote groepen mensen zich meerdere keren laten testen, ook als ze geen klachten hebben (grootschalig testen). Hiervoor kunnen bijvoorbeeld alle inwoners van een dorp of buurt worden uitgenodigd als er daar veel mensen besmet zijn. VWS wil weten welke omstandigheden stimuleren dat zoveel mogelijk mensen zich laten testen. In twee gemeenten – Dronten en Bunschoten - is onderzocht welk effect twee factoren hebben op het aantal mensen dat zich laat testen en hoe vaak ze dat doen: de afstand tot een testlocatie en de informatie in een uitnodigingsbrief. Het blijkt dat meer mensen zich laten testen als de testlocatie dichtbij is, op minder dan 2 kilometer van hun woning. Ook laten ze zich dan vaker testen. Het effect van de informatie in de uitnodigingsbrieven is niet goed te bepalen. Dat komt omdat er naast de brieven ook in de media (radio, sociale media, reclameborden) veel aandacht aan de pilots is besteed. In beide gemeenten zijn twee soorten brieven gestuurd met een verschillende inhoud. Een deel van de inwoners ontving een brief met alleen basisinformatie over grootschalig testen. Het andere deel ontving een brief waarin de basisinformatie werd aangevuld met informatie over het risico op een besmetting met het coronavirus en de verspreiding ervan. De opkomst was bij beide groepen inwoners bijna even hoog. De Corona Gedragsunit van het RIVM heeft dit onderzoek in samenwerking met de Universiteit van Amsterdam, GGD’en en gemeenten uitgevoerd.

In clinical research, populations are often selected on the sum-score of diagnostic criteria such as symptoms. Estimating statistical models where a subset of the data is selected based on a function of the analyzed variables introduces Berkson’s bias, which presents a potential threat to the validity of findings in the clinical literature. The aim of the present paper is to investigate the effect of Berkson’s bias on the performance of the two most commonly used psychological network models: the Gaussian Graphical Model (GGM) for continuous and ordinal data, and the Ising Model for binary data. In two simulation studies, we test how well the two models recover a true network structure when estimation is based on a subset of the data typically seen in clinical studies. The network is based on a dataset of 2,807 patients diagnosed with major depression, and nodes in the network are items from the Hamilton Rating Scale for Depression (HRSD). The simulation studies test different scenarios by varying (1) sample size and (2) the cut-off value of the sum-score which governs the selection of participants. The results of both studies indicate that higher cut-off values are associated with worse recovery of the network structure. As expected from the Berkson’s bias literature, selection reduced recovery rates by inducing negative connections between the items. Our findings provide evidence that Berkson’s bias is a considerable and underappreciated problem in the clinical network literature. Furthermore, we discuss potential solutions to circumvent Berkson’s bias and their pitfalls.

Networks have been recently proposed as plausible models of psychological phenomena in several domains, such as personality psychology and psychopathology. In these fields, nodes represent variables such as cognitions, behaviors, emotions, motivations, and symptoms, and edges represent their pairwise associations. Edge weights are typically estimated using regularized partial correlations, for instance via the graphical lasso. In several situations, it is necessary to compute networks on observations that belong to different classes (e.g., patients vs. controls). Previous studies estimated either a single network for all classes or several networks in each class independently. These strategies may be both suboptimal. The Fused Graphical Lasso (FGL) has been recently proposed for dealing with such situations (Danaher et al., 2014), but it has never been applied to psychology before. FGL allows simultaneously estimating multiple partial correlation networks from observations belonging to different classes. FGL does not assume that the networks are similar, but if similarities are present, they are exploited to improve parameter estimates. This method requires setting two tuning parameters: One is akin to the graphical lasso parameter and controls sparsity, the second one controls the amount of similarity among classes. We developed an R package that implements automatic tuning parameter selection according to information criteria (AIC, BIC, and extended BIC) or relying on k-fold crossvalidation. We present FGL from a theoretical point of view, discuss its performance in simulation studies, and show examples of its applications to personality psychology and psychopathology.

Spousal bereavement can cause a rise in depressive symptoms. This study empirically evaluates 2 competing explanations concerning how this causal effect is brought about: (a) a traditional latent variable explanation, in which loss triggers depression which then leads to symptoms; and (b) a novel network explanation, in which bereavement directly affects particular depression symptoms which then activate other symptoms. We used data from the Changing Lives of Older Couples (CLOC) study and compared depressive symptomatology, assessed via the 11-item Center for Epidemiologic Studies Depression Scale (CES-D), among those who lost their partner (N = 241) with a still-married control group (N = 274). We modeled the effect of partner loss on depressive symptoms either as an indirect effect through a latent variable, or as a direct effect in a network constructed through a causal search algorithm. Compared to the control group, widow(er)s' scores were significantly higher for symptoms of loneliness, sadness, depressed mood, and appetite loss, and significantly lower for happiness and enjoyed life. The effect of partner loss on these symptoms was not mediated by a latent variable. The network model indicated that bereavement mainly affected loneliness, which in turn activated other depressive symptoms. The direct effects of spousal loss on particular symptoms are inconsistent with the predictions of latent variable models, but can be explained from a network perspective. The findings support a growing body of literature showing that specific adverse life events differentially affect depressive symptomatology, and suggest that future studies should examine interventions that directly target such symptoms. (PsycINFO Database Record (c) 2015 APA, all rights reserved).

About 17% of humanity goes through an episode of major depression at some point in their lifetime. Despite the enormous societal costs of this incapacitating disorder, it is largely unknown how the likelihood of falling into a depressive episode can be assessed. Here, we show for a large group of healthy individuals and patients that the probability of an upcoming shift between a depressed and a normal state is related to elevated temporal autocorrelation, variance, and correlation between emotions in fluctuations of autorecorded emotions. These are indicators of the general phenomenon of critical slowing down, which is expected to occur when a system approaches a tipping point. Our results support the hypothesis that mood may have alternative stable states separated by tipping points, and suggest an approach for assessing the likelihood of transitions into and out of depression.

In recent years, research on dynamical systems in psychology has emerged, which is analogous to other fields such as biology and physics. One popular and promising line of research involves the modeling of psychological systems as causal systems or networks of cellular automat. The general hypothesis is that noticeable macroscopic behavior—the co-occurrence of aspects of psychology such as cognitive abilities, psychopathological symptoms, or behavior—is not due to the influence of unobserved common causes, such as general intelligence, psychopathological disorders, or personality traits, but rather to emergent behavior in a network of interacting psychological, sociological, biological, and other components. This dissertation concerns the estimation of such psychological networks from datasets. While this line of research originated from a dynamical systems perspective, the developed methods have shown strong utility as exploratory data analysis tools, highlighting unique variance between variables rather than shared variance across variables (e.g., factor analysis). In addition, this dissertation shows that network modeling and latent variable modeling are closely related and can complement one-another. The methods are thus widely applicable in diverse fields of psychological research. To this end, the dissertation is split in three parts. Part I is aimed at empirical researchers with an emphasis on clinical psychology, and introduces the methods in conceptual terms and tutorials. Part II is aimed at psychometricians and methodologists, and discusses the methods in technical terms. Finally, Part III is aimed at R users with an emphasis on personality research.

The International Classification of Diseases and Related Health Problems (ICD) and the Diagnostic and Statistical Manual of Mental Disorders (DSM) represent dominant approaches to diagnosis of mental disorders. However, it is unclear how these alternative systems relate to each other when taking into account the symptoms that make up the disorders. This study uses a network approach to investigate the overlap in structure between diagnostic networks pertaining to ICD-10 and DSM-IV-TR. Networks are constructed by representing individual symptoms as nodes, and connecting nodes whenever the corresponding symptoms feature as diagnostic criteria for the same mental disorder. Results indicate that, relative to the DSM-IV-TR network, the ICD-10 network contains (a) more nodes, (b) lower level of clustering, and (c) a higher level of connectivity. Both networks show features of a small world, and have similar (of “the same”) high centrality nodes. Comparison to empirical data indicates that the DSM-IV-TR network structure follows comorbidity rates more closely than the ICD-10 network structure. We conclude that, despite their apparent likeness, ICD-10 and DSM-IV-TR harbour important structural differences, and that both may be improved by matching diagnostic categories more closely to empirically determined network structures.

This chapter provides a general introduction of network modeling in psychometrics. The chapter starts with an introduction to the statistical model formulation of pairwise Markov random fields (PMRF), followed by an introduction of the PMRF suitable for binary data: the Ising model. The Ising model is a model used in ferromagnetism to explain phase transitions in a field of particles. Following the description of the Ising model in statistical physics, the chapter continues to show that the Ising model is closely related to models used in psychometrics. The Ising model can be shown to be equivalent to certain kinds of logistic regression models, loglinear models and multi-dimensional item response theory (MIRT) models. The equivalence between the Ising model and the MIRT model puts standard psychometrics in a new light and leads to a strikingly different interpretation of well-known latent variable models. The chapter gives an overview of methods that can be used to estimate the Ising model, and concludes with a discussion on the interpretation of latent variables given the equivalence between the Ising model and MIRT., In Irwing, P., Hughes, D., and Booth, T. (2018). The Wiley Handbook of Psychometric Testing, 2 Volume Set: A Multidisciplinary Reference on Survey, Scale and Test Development. New York: Wiley

In recent literature, the Gaussian Graphical model (GGM; Lauritzen, 1996), a network of partial correlation coefficients, has been used to capture potential dynamic relationships between psychological variables. The GGM can be estimated using regularization in combination with model selection using the extended Bayesian Information Criterion (Foygel and Drton, 2010). I term this methodology GeLasso, and asses its performance using a plausible psychological network structure with both continuous and ordinal datasets. Simulation results indicate that GeLasso works well as an out-of-the-box method to estimate a psychological network structure.

While there is a call for the use of novel statistical inference methods in the social sciences, the scientific community is slow in adapting to using such statistical methods. A possible reason for this is that state-of-the-art statistical methods are mostly implemented in the statistical programming language R whereas empirical social scientists more often rely on propitiatory statistical software such as SPSS. We introduce JASP, an graphical environment for statistical analyses that combines the strength of R, open-source and state-of-the-art statistical methods, with the flexibility and accessibility of proprietary statistical software. JASP provides a rich graphical user interface and implements a wide range of classical as well as Bayesian analyses. JASP is freely available for Windows, Mac OS X and Linux under an open source license from https://jasp-stats.org.

This paper explores the suitability of social media networks (SMNs) as a means of influencing the public’s decision-making process regarding vaccinations, specifically a vaccination to protect girls against HPV, a virus associated with cervical cancer. Parents of girls in the target cohort were invited to online discussion forums where they could discuss their opinions on the vaccination. We varied the posts on the forums in different experimental condition, such that they were exposed to promotion of the vaccination in one of four different ways, and coming from one of two different sources, i.e., peers or government health representatives. Following the health belief model (HBM), these messages served as cues to action. After their active participation on the forums, participants filled out a ques-tionnaire with items related to the HBM. Analyses revealed no effect of our experimental manipula-tions of the cue to action. However, using an exploratory novel network analysis approach, we find that the HBM does not adequately account for influence via SMNs. Specifically we show that vaccina-tion decisions are not taken in social isolation, a fact thus far ignored by various forms of the HBM. Implications for studies assessing the use of online channels for health communication are discussed.

Networks have been recently proposed for modeling dynamics in several kinds of psychological phenomena, such as personality and psychopathology. In this work, we introduce techniques that allow disentangling between-subject networks, which encode dynamics that involve stable individual differences, from within-subject networks, which encode dynamics that involve momentary levels of certain individual characteristics. Furthermore, we show how networks can be simultaneously estimated in separate groups of individuals, using a technique called the Fused Graphical Lasso. This technique allows also performing meaningful comparisons among groups. The unique properties of each kind of network are discussed. A tutorial to implement these techniques in the “R” statistical software is presented, together with an example of application.

In recent years, network analysis has been applied to identify and analyse patterns of statistical association in multivariate psychological data. In these approaches, network nodes represent variables in a data set, and edges represent pairwise conditional associations between variables in the data, while conditioning on the remaining variables. This Primer provides an anatomy of these techniques, describes the current state of the art and discusses open problems. We identify relevant data structures in which network analysis may be applied: cross-sectional data, repeated measures and intensive longitudinal data. We then discuss the estimation of network structures in each of these cases, as well as assessment techniques to evaluate network robustness and replicability. Successful applications of the technique in different research areas are highlighted. Finally, we discuss limitations and challenges for future research. Network analysis allows the investigation of complex patterns and relationships by examining nodes and the edges connecting them. Borsboom et al. discuss the adoption of network analysis in psychological research.

We argue that it is useful to distinguish between three key goals of personality science—description, prediction and explanation—and that attaining them often requires different priorities and methodological approaches. We put forward specific recommendations such as publishing findings with minimum a priori aggregation and exploring the limits of predictive models without being constrained by parsimony and intuitiveness but instead maximizing out-of-sample predictive accuracy. We argue that naturally occurring variance in many decontextualized and multidetermined constructs that interest personality scientists may not have individual causes, at least as this term is generally understood and in ways that are human-interpretable, never mind intervenable. If so, useful explanations are narratives that summarize many pieces of descriptive findings rather than models that target individual cause–effect associations. By meticulously studying specific and contextualized behaviours, thoughts, feelings and goals, however, individual causes of variance may ultimately be identifiable, although such causal explanations will likely be far more complex, phenomenon-specific and person-specific than anticipated thus far. Progress in all three areas—description, prediction and explanation—requires higher dimensional models than the currently dominant ‘Big Few’ and supplementing subjective trait-ratings with alternative sources of information such as informant-reports and behavioural measurements. Developing a new generation of psychometric tools thus provides many immediate research opportunities. © 2020 European Association of Personality Psychology

Recent reasoning in psychopathology sees mental disorders as phenomena that emerge, at least in part, from complex patterns of symptom-symptom interactions. Borderline Personality Disorder (BPD) is characterized by instability of self-image, interpersonal relationships, and affects, as well as by marked impulsivity. In this study, we investigated the dynamical interplay of BPD symptoms over time. We employed an Ecological Momentary Assessment (EMA) protocol assessing BPD symptoms five times a day for 31 days in a sample of 156 participants. Participants also completed a comprehensive assessment of personality, including BPD, before the EMA, after the EMA, and one month later. We examined BPD symptoms at two timescales: The short scale of EMA and the long scale offered by the three questionnaires. At the short timescale, we employed a multilevel vector auto-regressive network model, which provided insights into the contemporaneous and cross-lagged relationships among symptoms. At the long timescale, we observed a decrease in all BPD symptoms immediately after EMA and after one month, suggesting beneficial effects of the self-reflection induced by EMA. We discuss how each analysis gave unique insights into the dynamics of BPD.

Network analysis represents a novel theoretical approach to personality. Network approaches motivate alternative ways of analyzing data, and suggest new ways of modeling and simulating personality processes. In the present paper, we provide an overview of network analysis strategies as they apply to personality data. We discuss different ways to construct networks from typical personality data, show how to compute and interpret important measures of centrality and clustering, and illustrate how one can simulate on networks to mimic personality processes. All analyses are illustrated using a data set on the commonly used HEXACO questionnaire using elementary R-code that readers may easily adapt to apply to their own data.

IntroductionReproducibility is a defining feature of science, but the extent to which it characterizes current research is unknown. Scientific claims should not gain credence because of the status or authority of their originator but by the replicability of their supporting evidence. Even research of exemplary quality may have irreproducible empirical findings because of random or systematic error.RationaleThere is concern about the rate and predictors of reproducibility, but limited evidence. Potentially problematic practices include selective reporting, selective analysis, and insufficient specification of the conditions necessary or sufficient to obtain the results. Direct replication is the attempt to recreate the conditions believed sufficient for obtaining a previously observed finding and is the means of establishing reproducibility of a finding with new data. We conducted a large-scale, collaborative effort to obtain an initial estimate of the reproducibility of psychological science.ResultsWe conducted replications of 100 experimental and correlational studies published in three psychology journals using high-powered designs and original materials when available. There is no single standard for evaluating replication success. Here, we evaluated reproducibility using significance and P values, effect sizes, subjective assessments of replication teams, and meta-analysis of effect sizes. The mean effect size (r) of the replication effects (Mr = 0.197, SD = 0.257) was half the magnitude of the mean effect size of the original effects (Mr = 0.403, SD = 0.188), representing a substantial decline. Ninety-seven percent of original studies had significant results (P < .05). Thirty-six percent of replications had significant results; 47% of original effect sizes were in the 95% confidence interval of the replication effect size; 39% of effects were subjectively rated to have replicated the original result; and if no bias in original results is assumed, combining original and replication results left 68% with statistically significant effects. Correlational tests suggest that replication success was better predicted by the strength of original evidence than by characteristics of the original and replication teams.ConclusionNo single indicator sufficiently describes replication success, and the five indicators examined here are not the only ways to evaluate reproducibility. Nonetheless, collectively these results offer a clear conclusion: A large portion of replications produced weaker evidence for the original findings despite using materials provided by the original authors, review in advance for methodological fidelity, and high statistical power to detect the original effect sizes. Moreover, correlational evidence is consistent with the conclusion that variation in the strength of initial evidence (such as original P value) was more predictive of replication success than variation in the characteristics of the teams conducting the research (such as experience and expertise). The latter factors certainly can influence replication success, but they did not appear to do so here. Reproducibility is not well understood because the incentives for individual scientists prioritize novelty over replication. Innovation is the engine of discovery and is vital for a productive, effective scientific enterprise. However, innovative ideas become old news fast. Journal reviewers and editors may dismiss a new test of a published idea as unoriginal. The claim that “we already know this” belies the uncertainty of scientific evidence. Innovation points out paths that are possible; replication points out paths that are likely; progress relies on both. Replication can increase certainty when findings are reproduced and promote innovation when they are not. This project provides accumulating evidence for many findings in psychological research and suggests that there is still more work to do to verify whether we know what we think we know.