COVID-19-related school closures caused unprecedented and prolonged disruption to daily life, education, and social and physical activities. This disruption in the life course affected the well-being of students from different age groups. This study proposed analyzing student well-being and determining the most influential factors that affected student well-being during the COVID-19 pandemic. With this aim, we adopted a cross-sectional study designed to analyze the student data from the Responses to Educational Disruption Survey (REDS) collected between December 2020 and July 2021 from a large sample of grade 8 or equivalent students from eight countries (n = 20,720), including Burkina Faso, Denmark, Ethiopia, Kenya, the Russian Federation, Slovenia, the United Arab Emirates, and Uzbekistan. We first estimated a well-being IRT score for each student in the REDS student database. Then, we used 10 data-mining approaches to determine the most influential factors that affected the well-being of students during the COVID-19 outbreak. Overall, 178 factors were analyzed. The results indicated that the most influential factors on student well-being were multifarious. The most influential variables on student well-being were students’ worries about contracting COVID-19 at school, their learning progress during the COVID-19 disruption, their motivation to learn when school reopened, and their excitement to reunite with friends after the COVID-19 disruption.

The demand for diagnostic feedback has triggered extensive research on cognitive diagnostic models (CDMs), such as the deterministic input, noisy output “and” gate (DINA) model. This study explored two Q-matrix specifications with the DINA model in a statewide large-scale mathematics assessment. The first Q-matrix was developed based on five predefined content reporting categories, and the second was based on the post hoc coding of 15 attributes by test-development experts. Total raw scores correlated strongly with the number of skills mastered, using both Q-matrices. Correlations between the DINA-model item statistics and those from the item response theory analyses were moderate to strong, but were always lower for the 15-skill model. Results highlighted the trade-off between finer-grained modeling and less precise model estimation.

Background: Variable selection is a central challenge in logistic regression, particularly in high-dimensional biomedical applications where correlated predictors and limited sample sizes complicate reliable identification of relevant variables. This study aims to systematically compare three widely used variable selection approaches - Boruta, LASSO, and Elastic Net - under a range of data-generating conditions and to illustrate their performance using an ovarian cancer miRNA dataset. Methods: We conducted a simulation study across 36 logistic regression scenarios varying in sample size, dimensionality, predictor correlation, and effect magnitude. Performance was evaluated using true-positive and false-positive selection rates. In addition, all three methods were applied to a real-world serum miRNA expression dataset, and discriminative performance was assessed using the area under the receiver operating characteristic curve (AUC). Results: Boruta, Elastic Net, and LASSO exhibited distinct variable selection behaviors across simulation scenarios. Boruta maintained strong true-positive recovery while controlling false positives in most settings, particularly when predictors were highly correlated. Elastic Net consistently achieved high sensitivity but produced comparatively large false-positive rates. LASSO showed the most conservative behavior, recovering fewer true predictors while maintaining low false-positive rates across nearly all scenarios. In the ovarian cancer miRNA application, all three methods achieved similarly strong test-set AUC performance, despite marked differences in the size of the selected biomarker panels. Conclusions: The results demonstrate clear trade-offs among the three methods. Boruta offers a favorable balance between sensitivity and specificity in highly correlated settings. Elastic Net prioritizes sensitivity at the cost of increased false discoveries, whereas LASSO provides stricter false-positive control with reduced sensitivity. These findings offer practical guidance for selecting variable selection methods in logistic regression, particularly for high-dimensional biomedical applications.