Academic Stress, Sleep Quality, and Lifestyle Factors Among University Students: A Cross-Sectional Dataset

Description

This cross-sectional survey dataset explores the intricate relationships between academic stress, sleep quality, and lifestyle factors among university undergraduate students. The dataset contains complete responses from 300 students across various major disciplines (including Law, NFE, CSE, EEE, Civil, Pharmacy, and BBA) with zero missing data, capturing 31 distinct variables spanning demographics, academic performance ($CGPA$), mental health scales, and daily habits. The study operates under three primary hypotheses: first, that poorer subjective sleep quality and shorter nightly sleep durations significantly associate with higher self-reported academic stress, depression, and anxiety ($H_1$); second, that elevated psychosocial and financial stress negatively impact $CGPA$ and increase behavioral maladaptations like absenteeism ($H_2$); and third, that robust social support, proper diet, and physical activity serve as positive buffers against high anxiety and depression scores ($H_3$). Key data insights reveal that 26.33% of the surveyed students experience severe to extreme academic stress levels, scoring a 4 or 5 on a standard 1–5 ordinal scale. Furthermore, a severe sleep deficit is prevalent across the cohort; while 53.0% rate their sleep quality as subjectively "Good", 53.0% of the entire sample averages only 5–6 hours of sleep per night. This widespread strain manifests in notable academic maladaptations, with a striking 32.3% of students admitting to skipping classes specifically due to the downstream effects of poor sleep or overwhelming stress. Interestingly, the data demonstrates a "CGPA Paradox" where high academic stress is distributed evenly across all performance brackets. High-performing students with a $CGPA$ above 3.70 experience severe academic strain at nearly identical rates to those with lower academic standings, indicating that high performance tiers do not shield students from psychological pressure. Data was gathered via a standardized digital questionnaire distributed across university student platforms, resulting in a balanced cohort consisting of 51.7% female and 48.3% male respondents with an average age of 21.9 years. Data cleaning involved standardizing categorical text strings and removing trailing whitespaces to make the files immediately ready for analytical pipelines. For future reuse, researchers can utilize this data to run non-parametric analyses (such as Spearman Rank Correlations for the ordinal mental health metrics), Chi-Square tests for behavioral groupings, or Binary Logistic Regression models to calculate Odds Ratios ($OR$) for outcomes like class absenteeism based on sleep restriction and anxiety levels. Because the study design is cross-sectional, users should interpret findings as associational rather than causal.

Steps to reproduce

To reproduce or extend this dataset, researchers should follow a structured three-phase process: instrument design, distribution, and data pipeline cleaning. First, construct a digital survey comprising the 31 structured variables, ensuring psychometric scales for academic stress, depression, and anxiety are mapped to a standard 1 to 5 ordinal rating system. Incorporate categorical lifestyle choices (e.g., bedtime ranges, caffeine intake, screen time) alongside discrete academic indices like continuous CGPA. Second, distribute the survey digitally across diverse university student networks and community communication channels to minimize departmental bias. Data should be gathered until achieving a cross-sectional cohort of approximately 300 complete student records with a balanced demographic distribution (e.g., near-equal gender splits and represented cohorts living on-campus, off-campus, and with family). Third, initiate the data pre-processing pipeline. Clean the raw data by standardizing categorical text outputs (such as converting inconsistent response labels into unified binary string descriptors), stripping any trailing whitespaces from column headers, and handling any accidental structural nulls or merged spreadsheet cells to ensure the resulting tabular dataset is cleanly formatted and completely optimized for statistical computing environments like R, SPSS, or Python.

Institutions

• Daffodil International University

  Dhaka Division, Dhaka

Categories

Licence