**The most important metric is `total_sleep_hours`.** It is the core outcome of the dataset, and `sleep_debt_category` is derived from it. It is also tied to the other outcomes: sleep latency, next-day fatigue and alarm snoozes. This is a judgement based on the dataset's structure, not a statistical ranking of every metric. **Highest records (9.8 hours)** - 17 users share the maximum. The five records shown are USR-01351, USR-01363, USR-01396, USR-01715 and USR-02117. - All 17 are in "Optimal Recovery". - They fall asleep quickly, averaging 22.4 minutes of sleep latency. - Their average next-day fatigue is 1.0, and they average 0.12 alarm snoozes. - Their average post-5pm caffeine is 13.8 mg and their average bedtime phone use is 22.3 minutes. - Deep sleep averages 23.7% and REM 18.4%. **Lowest records (3.2 hours)** - 242 users share the minimum. The five records shown are USR-00001, USR-00003, USR-00025, USR-00070 and USR-00086. - All five shown work in Healthcare / Shift Work, and four of them are Night Owls. - They take much longer to fall asleep, averaging 70.0 minutes of sleep latency. - Their average next-day fatigue is 9.5 and they average 6.0 alarm snoozes. - Their average post-5pm caffeine is 38.1 mg and their average bedtime phone use is 123.5 minutes. - Deep sleep averages 19.9% and REM 18.7%, so REM barely differs from the top group. **Caveats** - Both extremes are ties, so the "highest" and "lowest" records are arbitrary picks from large groups. The group averages describe them better than any one user. - The 3.2-hour group spans two categories, "Moderate Debt" and "Severe Sleep Debt". The category is therefore not a pure function of total sleep hours alone. - The data is observational, so the link between heavy phone use or caffeine and short sleep is an association, not proof of cause. - Bedtime phone minutes are sensitive person-level data, so I only reported group averages for them.