Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 2 appraisals
Behavior research methods
Measuring complex constructs in large-scale text with computational social mixed methods
A growing convergence between social science and machine learning enables, in principle, large-scale analyses of complex social phenomena through text. Yet, approaches leveraging supervised text classification based on human-annotated data for statistical analysis often treat conceptual validity and technical performance as separate challenges, impairing measurement quality. We provide guidelines to bridge this gap in what we call computational social mixed methods pipelines across three stages: data annotation, model training, and statistical analysis. Building on best practices and our own methodological innovations, such as "Iterative Annotation" and "Training on Confident Examples", we address recurring pitfalls like unbalanced training data or stagnant model performance. We also discuss when large language models constitute a viable alternative to transformer-based classifiers. Using a case study on countering online hate, we illustrate how consequently integrating social science and machine learning expertise improves the validity and comparability of computational social science.
24 July 2026
Read appraisal →Journal of affective disorders
What retrospective and dynamic assessments tell us about youth depression: A network analysis perspective.
PURPOSE: Traditional assessments of depressive symptoms often rely on retrospective self-reports, which may be affected by cognitive and memory biases. Few studies have compared retrospective and dynamic (real-time) assessments to examine the consistency and structure of depressive symptom reporting. This study aimed to compare retrospective and dynamic assessments of depressive symptoms in youth using network analysis to explore symptom-level associations and clustering. METHODS: Ninety Brazilian adolescents and young adults (mean age = 18 years), with and without depression, completed the Short Mood and Feelings Questionnaire (SMFQ) every other day for 14 days via a smartphone-based chatbot (dynamic assessment). At the end of the 2-week period, they completed the same questionnaire retrospectively. Network analyses were conducted using Exploratory Graph Analysis (EGA) and Dynamic Exploratory Analysis (DEA) to identify symptom communities and compare network structures across both assessment methods. RESULTS: Both retrospective and dynamic assessments revealed three symptom communities; however, the composition and structure of these communities differed. Retrospective assessments showed stronger connections among cognitive symptoms, while dynamic assessments displayed a more balanced distribution, with stronger associations between somatic and affective symptoms. IMPLICATIONS: Findings highlight significant differences in depressive symptom networks between retrospective and dynamic assessments. Dynamic methods may offer less biased and more ecologically valid insights into youth depression, underscoring the importance of real-time data collection in clinical assessment and research.
16 Apr 2026
Read appraisal →