**The most surprising finding is that weekend extra hours have almost no relationship to anything, while weekday-style daily usage predicts nearly everything.** - **Daily usage correlates strongly with every outcome.** Across all 4,500 students, daily usage hours correlate at about −0.85 with Mental_Health_Index and −0.72 with sleep hours. They also correlate at −0.71 with GPA and −0.68 with sleep quality, and at +0.75 with perceived stress. - **Weekend_Extra_Hours is essentially uncorrelated with all of them.** Its correlation with daily usage is −0.01, with mental health 0.004, with GPA 0.012 and with sleep hours 0.017. So heavy daily users don't systematically add more weekend hours, and extra weekend hours don't track better or worse wellbeing. - **Age is also unrelated to GPA.** The correlation is 0.03. If weekend and weekday usage were the same underlying habit, both would show up in the outcomes, but only daily usage does. The pattern is also unusually clean for observational data on people. The ordering of Overall_Impact groups (Beneficial, Neutral, Negative) lines up almost perfectly with average daily usage: | Overall_Impact | Students | Avg daily usage (hrs) | Avg mental health | Avg GPA | |---|---|---|---|---| | Beneficial | 3,681 | 4.4 | 84.3 | 3.55 | | Neutral | 654 | 8.5 | 63.4 | 3.00 | | Negative | 165 | 12.3 | 47.1 | 2.54 | This looks like a single dominant usage "dose" driving every outcome, with weekend hours and age adding nothing. That makes the data look more engineered than organic. Caveats: these are Pearson (linear) correlations on observational data, so they show association, not causation. I did not test non-linear weekend effects or check for subgroup differences. The figures come from the correlation query across all 4,500 rows (steps 0–11) and the Overall_Impact breakdown (step 1).