**The most counterintuitive finding is that weekend extra hours are unrelated to everything else in the dataset, even though weekday usage is strongly tied to almost every outcome.** **What the data shows (n = 4,500 students, correlations from steps 8 and 3):** - **Weekday usage:** Daily usage hours track outcomes very tightly. - Mental health: r = **-0.85** - Sleep hours: r = **-0.72** - GPA: r = **-0.71** - Sleep quality: r = **-0.68** - Stress: r = **+0.75** - **Weekend extra hours:** These look like noise. - Daily usage: r = **-0.010** - Mental health: r = **0.004** - GPA: r = **0.012** - Sleep hours: r = **0.017** - **Age:** Age and GPA are also essentially uncorrelated (r = 0.028). **Why it's surprising:** You would expect heavy daily users to also spend more extra time on weekends. You would also expect extra weekend screen time to show at least some link to sleep or mood. Neither happens. Daily usage predicts nearly every outcome, while weekend usage predicts none. **The same pattern in the "Overall_Impact" labels (step 1):** | Overall_Impact | Students | Avg daily hours | Avg mental health | Avg GPA | Avg sleep hours | |---|---|---|---|---|---| | Beneficial | 3,681 | 4.4 | 84.3 | 3.55 | 7.0 | | Neutral | 654 | 8.5 | 63.4 | 3.00 | 5.5 | | Negative | 165 | 12.3 | 47.1 | 2.54 | 4.2 | The labels line up almost perfectly with usage level. Mental health drops about 37 points from the Beneficial to the Negative group, and sleep falls by about 3 hours. The dataset also has no weekday-versus-weekend dose-response, only a weekday one. **Caveat:** This clean, almost deterministic gradient on one variable, paired with complete independence on a closely related one, is more orderly than real survey data usually is. It suggests the data may have been generated from a formula built on daily usage. The correlations are real in this dataset, but I'd be cautious about treating them as causal effects for real students. They are observational, with no levers and no time dimension. The SQL behind these numbers is in steps 1, 3 and 8.