Is my sleep better on days I use the steam room?

• Question Verbatim from You: “Is my sleep better on days I use the steam room?”
• Interpretation of the Question (an attempt to make the question slightly more precise, computable and scientific)
  • “My sleep” we are interpreting as all 27 measures your OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com ring reports for a night, from its scores to your heart rate, plus the next morning's readiness. You asked us to look at every one rather than pick one.
  • “Days I use the steam room” we are interpreting as days with a steam-room session in your messages to the PerSci chatbotThe PerSci chatbot is a text-message assistant you log your day to in plain words; a session mentioned there is labelled by hand before any analysis uses it. More at persci.app. A day no message mentions counts as a day without, except 5 days whose messages suggest a session the labelling missed, which are left out.
  • “Better” we are interpreting as different on the night that began that evening, in either direction, by more than chance would make, allowing for testing 27 measures at once.
  • Precisely: on steam nights, does any of your OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com sleep measures differ from your other nights by more than chance would make?

Not clearly, based on the analysis below, and within its assumptions and limits.

• Against all your other nights, none of the 27 measures differed by more than chance would make. Your sleep score averaged 81 on steam nights and 79 on the others.
• Against the nights around them, you went to bed earlier on 21 of your 29 steam nights, by about 24 minutes, and OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com's timing score was higher on 25 of 29. The timing score is the one result that passes the bar for chance across all 27 measures.
• So: not a clear yes, and not a clear no. Earlier bedtimes on steam days are a lead to test on your next steam sessions, not a finding.

• Data sources:
  • Devices: OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com ring data, fetched from OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com daily; the nights used here run to 2026-10-09, the last steam day. Each night is filed under the evening it began.
  • Textual sources: steam room data: your messages to the PerSci chatbotThe PerSci chatbot is a text-message assistant you log your day to in plain words; a session mentioned there is labelled by hand before any analysis uses it. More at persci.app, each labelled by hand, the last in October 2026.
• Dates: 2026-01-07 to 2026-10-09, 276 days: 29 steam nights and 242 other nights.
• People and Organizations Named or Anonymized:
  • Named (in your template names, your notes, or what you told us): Equinox
  • Anonymized (shown by alias, inside your own notes and quotes too): None

Nights after days you used the steam room, against nights after days you did not.

• 29 steam nights
• 242 other nights
• 5 days left out: their messages suggest a session the labelling missed

Every measure your OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com ring reports for a night, in three groups:

• OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com's scores: Sleep score, Readiness, next morning, Sleep score part: timing, Sleep score part: total sleep, Sleep score part: efficiency, Sleep score part: restfulness, Sleep score part: REM sleep, Sleep score part: deep sleep, Sleep score part: time to fall asleep
• When, and how long: Bedtime, Wake time, Time in bed, Total sleep, Light sleep, Deep sleep, REM sleep, Awake in bed, Time to fall asleep, Sleep efficiency, Restless periods
• Heart, breathing and temperature: Lowest heart rate, Average heart rate, Heart rate variability (HRV), Breathing rate, Blood oxygen, Breathing disturbance index, Temperature, next morning, against your usual

This was the test written down before any sleep number was looked at. It compares your steam nights with every other night in the record, and it found no measure that differed by more than chance would make, allowing for 27 measures tested at once. The closest was bedtime: about a 1-in-12 chance of a gap that big by luck alone (p = 0.08, raw, before allowing for the 27 measures; allowing for them, q = 0.83).

How much it could see: with 29 steam nights, the smallest difference this test could have seen was about 4.9 points on the sleep score, or 28 min of sleep. At the lucky end of what the data allow (the top of the 95% plausible range), a steam night could add up to about 5.5 points to the sleep score, or 30 min of sleep; more than that is unlikely. Counting the 5 left-out days as days without steam gives the same answer.

For p-values, corrections for testing many measures, effect sizes and the smallest detectable differences, see The statistics behind this below.

Smallest difference this could have seen: If steam really changed this by at least this much, a test like this would show it about 4 times in 5. A smaller real change could easily go unseen, so 'no difference found' means 'no difference this big'.

• Comparing with all your other nights mixes in ten months of slow change: seasons, routines, a new schedule. That adds noise, and a small difference can hide in it.
• Comparing each steam night with the nights around it takes that slow change out, so a small, steady difference shows up.
• Both are right about what they measure. The second is the sharper of the two here, but it was chosen after the first had been seen, so what it shows is a lead, not a finding.

• Steam days are gym days. Every session with a gym named was at Equinox, so this is about the steam room and the workout around it together. Whether the earlier bedtimes come from the steam or the workout is the next question, below.
• The closer comparison was chosen after the registered one. A first simple version (“above your usual night”) was dropped because several OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com scores sit at their top value most nights; its results had been seen. A first look at your first 20 steam days, as a check that the pipeline worked, also came before this one: 20 of these 29 steam nights were in it.
• Matched on the same weekday, nothing passes the bar. Comparing each steam night only with the same weekday within three weeks, the timing score and bedtime both weaken. That leaves fewer nights to compare with, which may be all it is, but a weekday habit cannot be ruled out.
• Some steam days may be counted as days without. A day no message mentions counts as a day without steam. At least 1 more day (2026-07-25) looks like a missed session; days like that make a real difference harder to see, not easier.
• The ring measures sleep, not how rested you feel.
• This describes a pattern in your data. It is not medical advice.

• What does my steam room data look like? — every session in the record, described without any question asked of it
• Does the steam room help my knee arthritis? — set against your stair-descent times; an article on Substack
• What does my Oura data look like? — the same OuraOura makes a ring worn on the finger that measures sleep, heart rate, heart rate variability and skin temperature overnight, and scores each night. More at ouraring.com nights, described without any question asked of them

All nights: difference in means (steam minus other), permutation p over 20,000 shuffles of which days were steam days, Benjamini-Hochberg q across the 27 measures, and the smallest difference detectable at 80% power. Around it: steam nights higher out of those compared, mean of the paired differences, sign-flip p over 20,000 flips, q across the 27. Same weekday: q for the same paired comparison, against the same weekday within three weeks. Hours are in hours, bedtime in hours of the clock. A lead is q below 0.1.