Dimensions of Externalizing Symptoms in Children and Adolescents

In this dissertation I focused on the classification system of mental disorders by examining associations between ADHD and ODD symptomatology in children, applying latent factor analysis. In my first article (Junghänel et al., 2020), I applied a newly developed bifactor model (Burns et al., 2020; Eid et al., 2018) – the bifactor S-1 model – in a clinical sample of children aged 6-18 years of age. This model allows for the à priori definition of a general reference factor, which is opposed to the traditional bifactor model in which a general factor is extracted in a more exploratory way. I compared models with difference reference factors according the trait-impulsivity theory (Beauchaine et al., 2010; Beauchaine & McNulty, 2013) and the two- (DSM-5; APA, 2013) vs. three-dimensional (ICD-10/ICD-11; WHO, 1992, 2020) conceptualization of ADHD. Key findings were the following: First, the model showed no statistical problems, such as anomalous factor loadings or collapsing s-factor, frequently found in the traditional bifactor model, strongly facilitating its interpretation. As the general reference factor was defined à priori and associations between s-factors could be assessed, the bifactor S-1 model allowed for a straight-forward interpretation of all included dimensions and their associations with one another. Second, I found the model with ADHD-HI as the general reference factor, which is in line with the two-dimensional conceptualization of ADHD (DSM-5; APA, 2013) to fit the data best. Third, I found two additional stable s-factors, accounting for the remaining variance, constituting ODD and ADHD-IN symptomatology. In the second article (Junghänel et al., 2022) I further elaborated my hypothesis regarding the general reference factor of ADHD and ODD by including AD symptomatology, with that also taking the multidimensional nature of impulsivity into account (Berg et al., 2015) by differentiating between physical and emotional impulsivity. In this article, I addressed several limitations of my first article. I included a more nuanced view of ODD symptomatology, by differentiation between the two frequently found dimensions of ODD-irritability and ODD-defiant behavior (Aebi et al., 2016; Burke et al., 2014; Rowe et al., 2010), used a more representative sample, assessed construct validity, and compared the bifactor S-1 models to a large range of other latent factor models. Key findings were the following: First, a main component representing irritability and emotional impulsivity emerged when assessing the structure of AD. In addition, I found two minor factors constituting positive emotionality and exuberance. Second, the bifactor S-1 model with AD-II as general reference factor fit best in statistical and theoretical terms. About 2/3 (70%) of the variance shared between ADHD and ODD could be explained by AD-II. The remaining third of the variance (30%) were explained by all s-factors taken together. Third, differential correlations between g- and s-factors with emotion regulation skills and the Parent Proxy Anger Scale (Irwin et al., 2012) supported construct validity. In sum, the bifactor S-1 model eliminated frequently found statistical problems and was clearly interpretable due to the general reference factor, which was defined à priori, as well as a result of the possibility to assess partial correlations between the residual s-factors. I found ADHD and ODD to share a large part of common variance that was mainly explained by irritability and emotional impulsivity. These results lead me to suggest AD-II as the common core of ADHD and ODD symptomatology, responsible for the frequently found comorbidities between them. In conclusion, I would like to propose an adaptation and extension of the specifier approach the ICD-11 (WHO, 2020) has already taken for ODD and AD symptomatology, by adding a specifier of chronic irritability to the ODD diagnosis. My suggestion is to understand AD-II as the core component of ADHD and ODD symptomatology. To this core, specifiers such as “with inattention”, “with hyperarousal” or “with defiant behavior” could be added. An examination regarding the extension of this approach into the internalizing spectrum, including mood and anxiety disorders seems promising. My results speak in favor of a more parsimonious, dimensional classification approach of mental disorders and will hopefully encourage further research in this area