Bayesian Network Structure and Predictability of Autistic Traits

Unpacking the overlap between Autism and ADHD in adults: A multi-method approach

The overlap between Autism and Attention-Deficit Hyperactivity Disorder (ADHD) is widely observed in clinical settings, with growing interest in their co-occurrence in neurodiversity research. Until relatively recently, however, concurrent diagnoses of Autism and ADHD were not possible. This has limited the scope for large-scale research on their cross-condition associations, further stymied by a dearth of open science practices in the neurodiversity field. Additionally, almost all previous research linking Autism and ADHD has focused on children and adolescents, despite them being lifelong conditions. Tackling these limitations in previous research, 5504 adults - including a nationally representative sample of the UK (Study 1; n = 504) and a large pre-registered study (Study 2; n = 5000) - completed well-established self-report measures of Autism and ADHD traits. A series of network analyses unpacked the associations between Autism and ADHD at the individual trait level. Low inter-item connectivity was consistently found between conditions, supporting the distinction between Autism and ADHD as separable constructs. Subjective social enjoyment and hyperactivity-impulsivity traits were most condition-specific to Autism and ADHD, respectively. Traits related to attention control showed the greatest Bridge Expected Influence across conditions, revealing a potential transdiagnostic process underlying the overlap between Autism and ADHD. To investigate this further at the cognitive level, participants completed a large, well-powered, and pre-registered study measuring the relative contributions of Autism and ADHD traits to attention control (Study 3; n = 500). We detected age- and sex-related effects, however, attention control did not account for the covariance between Autism and ADHD traits. We situate our findings and discuss future directions in the cognitive science of Autism, ADHD, and neurodiversity, noting how our open datasets may be used in future research.

Unique Variable Analysis: A Network Psychometrics Method to Detect Local Dependence

The local independence assumption states that variables are unrelated after conditioning on a latent variable. Common problems that arise from violations of this assumption include model misspecification, biased model parameters, and inaccurate estimates of internal structure. These problems are not limited to latent variable models but also apply to network psychometrics. This paper proposes a novel network psychometric approach to detect locally dependent pairs of variables using network modeling and a graph theory measure called weighted topological overlap (wTO). Using simulation, this approach is compared to contemporary local dependence detection methods such as exploratory structural equation modeling with standardized expected parameter change and a recently developed approach using partial correlations and a resampling procedure. Different approaches to determine local dependence using statistical significance and cutoff values are also compared. Continuous, polytomous (5-point Likert scale), and dichotomous (binary) data were generated with skew across a variety of conditions. Our results indicate that cutoff values work better than significance approaches. Overall, the network psychometrics approaches using wTO with graphical least absolute shrinkage and selector operator with extended Bayesian information criterion and wTO with Bayesian Gaussian graphical model were the best performing local dependence detection methods overall.

A network perspective on abnormal child behavior in primary school students

Abnormal child behavior is a complex entity including among others disorders such as ADHD, a neuropsychiatric disorder interfering with the normal development of the child, and combining many comorbidities. There are no specific tools designed to assess abnormal child behavior, although well-known ADHD screening tools such as the CTRS-R:S go beyond the disorder itself and cover the wider construct. This can be understood as the complex system of mutually influencing entities, that is why the aim of this study was to investigate the items of the CTRS-R:S as a psychological network. We analyzed a data set of 525 children in French-speaking primary schools from Belgium, and estimated a Bayesian Gaussian Graphical Model with a Bayes Factor of 30. We identified highly connected nodes from each community of the CTRS-R:S to build an additional network of communities. In the domain network, Oppositional and Cognitive skills share the strongest connection in the network. The 28-item network was overall positively connected albeit the presence of some negative connections. The spinglass algorithm identified eight domains in the CTRS-R:S network that causally influence each other. The network approach is a useful framework to investigate the domains of abnormal child behavior as well as the interplay between individual symptoms.