An R-based data analysis project exploring how social media usage affects students' mental health, sleep, and academic performance across 1,705 survey responses.
This dataset is from Kaggle (accessible here). It contains survey responses from 1,705 students across 11 variables:
- Student_ID: Unique identifier per respondent
- Age: Student age
- Gender: Male / Female
- Academic_Level: High School, Undergraduate, or Graduate
- Country: Country of residence
- Avg_Daily_Usage_Hours: Average hours per day spent on social media
- Most_Used_Platform: Primary social media platform (Instagram, TikTok, LinkedIn, etc.)
- Affects_Academic_Performance: Whether the student reports social media hurts their academics (Yes/No)
- Sleep_Hours_Per_Night: Average nightly sleep hours
- Mental_Health_Score: Self-reported mental health score (higher = better)
- Overall_Impact: Student's overall assessment of social media's effect (Negative, Neutral, Positive)
No missing values or duplicates were found in the dataset.
Problem Statement: Social media use among students is near-universal, but its relationship to mental health, sleep, and academic performance is not always clear. This project uses R to explore those relationships across a diverse student sample and identify which usage patterns and platforms are associated with better or worse outcomes.
Research Questions:
- Does higher daily social media usage correlate with lower mental health scores?
- Is there a relationship between social media use and sleep hours?
- Does social media use affect academic performance, and does sleep play a role in that?
- Do certain platforms correlate with better or worse mental health outcomes?
- Can daily usage hours alone predict mental health score?
- More usage, worse outcomes. Students using social media 6+ hours per day score about 3 points lower in mental health and sleep roughly 2.6 fewer hours per night compared to low-usage peers.
- Academic impact is tied to sleep loss. Students who report social media hurts their academics sleep over an hour less per night on average. The difference is statistically significant (p < 2.2e-16).
- Platform choice matters. LinkedIn and YouTube users report notably better mental health scores than Snapchat and TikTok users.
- Usage hours are a strong predictor. Daily usage hours alone explain 68% of the variance in mental health score (R² = 0.68), which is high for a single predictor in social science research.