Background: A great part of the literature has confirmed the importance of both child and parents reports as source of factual information, especially for childhood emotional syndromes. In our study we aimed at: (i) calculating mother-child agreement and (ii) evaluating factorial structure of the Screen for Child Anxiety Related Emotional Disorders (SCARED) questionnaire in an Italian clinical sample. The novelty of this contribution is two-fold: first, from a clinical point of view, we investigated the parent-child agreement level and examined separately the factorial structures of both parent and child versions of the SCARED for the first time in an Italian clinical sample. Second, unlike previous studies, we used statistical approaches specifically suited to account for the ordinal nature of the collected variables.Method: In a clinical sample of 171 children and adolescents aged 8–18 and their mothers we evaluated inter-rater agreement using weighted kappa indices to assess agreement for each item belonging to a certain SCARED subscale. Exploratory factor analysis for ordinal data was then performed on the polychoric correlation matrix calculated on SCARED items. Differences in the numbers of symptoms reported by children and parents were evaluated as well.Results and Conclusions: Our results reveal moderate to strong mother-child agreement. A significant age effect is present. Two different factorial solutions emerged for parent and child SCARED versions (a 5 factor structure for parents and a 6 factor solution in the child version, including a new factor “Worry about Parents”). This study confirmed the importance of evaluating both child and parent reports in assessment protocols for anxiety disorders. Our findings could help clinicians to determine which information, and from which rater, must be accounted for in evaluating treatment decisions. Moreover, we find that patients characteristics, such as gender and age, should be taken into account when assessing agreement.

The latent space item response model (LSIRM) is a newly-developed approach to analyzing and visualizing conditional dependencies in item response data, manifested as the interactions between respondents and items, between respondents, and between items. This paper provides a practical guide to the Bayesian estimation of LSIRM using three open-source software options, JAGS, Stan, and NIMBLE in R. By means of an empirical example, we illustrate LSIRM estimation, providing details on the model specification and implementation, convergence diagnostics, model fit evaluations and interaction map visualizations.