Social science researchers commonly employ univariate models, yet multivariate models offer the advantage of depicting multiple outcomes and the dependencies between these outcomes simultaneously....Show moreSocial science researchers commonly employ univariate models, yet multivariate models offer the advantage of depicting multiple outcomes and the dependencies between these outcomes simultaneously. De Rooij and Groenen (2021) introduced the MELODIC family, a multivariate approach for addressing these multiple binary outcome variables. The assessment focused on diagnoses and the scales of a behavioural screening questionnaire (SDQ) in a Northern Netherlands outpatient population. Predictive validity gauged content-related scale performance in diagnosis prediction. Discriminative validity was confirmed if only the diagnosis-related scale accurately predicted the presence of the diagnosis. AUC values were used for these comparisons. No significant differences emerged among the models. Since we did not find direct differences across the different models. We tried to elucidate the cause of these non-significant differences by post-hoc analyses: scatterplots and Brier scores. The scatterplots of the ordering of probabilities for a univariate multiple logistic regression and a MELODIC model did not offer any more insights. The Brier scores yielded no additional insights either. There was no more evidence for the predictive and discriminative validity when the MELODIC model was used. The benefit of the utilization of a MELODIC model was that it allowed for the inclusion of all predictors and outcome variables at the same time, eliminating the requirement for multiple separate univariate models and thus reducing the likelihood of errors.Show less
As the availability of data becomes more widespread and computational technology develops, the need to model several outcome variables at once increases. To do this for multiple dichotomous outcome...Show moreAs the availability of data becomes more widespread and computational technology develops, the need to model several outcome variables at once increases. To do this for multiple dichotomous outcome variables, De Rooij and Groenen (2021) proposed the MELODIC family for simultaneous binary logistic regression in a reduces space. As an added feature, 4 different forms of regularization on the singular values were proposed to let the algorithm itself select the true dimensionality of the data set. In this paper a simulation study was performed to provide empirical evidence for the functionality of the regularization features on data sets with differing numbers of subjects, predictor variables, and outcome variables. The results show that the hard-thresholding regularization consistently estimates the correct dimensionality with regularization on the logarithm of the singular values slightly outperforming the regularization on the unaltered singular values.Show less