Qualitative and probability-based analysis of student open-ended responses on academic stress and mental health

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Likert-scale instruments capture how much stress students report, but not what shape that stress takes. We present a thematic and probability-based analysis of open-ended survey responses from 217 graduate students at a private university in central India, collected in July 2025. Students described how academic stress had affected their mental health and proposed institutional changes they considered necessary. Seven stressor themes were extracted through inductive coding following the Braun and Clarke framework. Anxiety symptoms and assignment accumulation each appeared in close to 29% of valid narratives. Female students reported higher rates on five of seven themes; the gap was largest for sleep and physical health (28.1% against 16.2% for male respondents). Conditional probability analysis placed the lift for exam scheduling stress co-occurring with sleep disruption at 1.31. VADER sentiment scoring identified anxiety narratives as carrying the most negative polarity. Classification of student-proposed solutions by locus of control showed that students reporting structural stressors (scheduling, attendance rules, faculty relations) consistently demanded structural remedies. The probability framework provides a replicable method for extracting granular stressor profiles from open-ended feedback.