Dear habits…

Contents

1. Stats2the numbers, sketched and coloured in
2. Stories3one entry for every habit, in the order they began
3. Pocket4inside the back cover

Stats

• 161
  Habits tracked
• 28,412
  Daily entries
• Nov 2022
  Tracking since

Stories

1. Retired

   Sleep before 2330

   Oct 2023 – Feb 2024

   I started this in October 2023 because I was convinced my real problem was sleep. My consistency kept slipping, I was going to bed far too late, and the advice around me all said the same thing: fix the bedtime and the rest follows. So the goal was simple, in bed before 23:30 every night, and I wanted to keep it going as long as I could.

   • 13 of 91
     nights before 23:30
   • 6
     nights in a row, my best
   • 1 Feb 2024
     last night logged

   It faded fast. The first week was the best it ever got, and after that the monthly hit rate only went down.

   Hit rate by month. Share of logged nights before 23:30. October is only 12 nights because it began on the 17th.

   label	Nights before 23:30
   Oct	50%
   Nov	16%
   Dec	4%
   Jan	7%

   Share of logged nights before 23:30. October is only 12 nights because it began on the 17th.

   What I never did was check whether the theory was true. This habit's own log can't answer it, but from April 2024 I tracked my bedtime in a separate habit. Here is when I actually fell asleep.

   When I actually fell asleep. 854 nights, April 2024 to September 2026, in half-hour steps. Orange bars are the nights that met the goal.

   label	Nights
   <22:30	10
   22:30	11
   23:00	20
   23:30	14
   00:00	61
   00:30	65
   01:00	117
   01:30	123
   02:00	122
   02:30	92
   03:00	76
   03:30	48
   04:00	27
   04:30+	63

   854 nights, April 2024 to September 2026, in half-hour steps. Orange bars are the nights that met the goal.

   • 02:00
     median bedtime
   • 5%
     of nights before 23:30
   • 8.4 h
     average sleep

   And it never changed. In no month did my median bedtime get earlier than 00:30.

   Median bedtime, month by month. Every month's median sits between 00:30 and 03:00. It never reached the goal.

   label	Median bedtime
   Apr 24	00:30
   May 24	01:30
   Jun 24	02:07
   Jul 24	01:45
   Aug 24	01:45
   Sep 24	01:00
   Oct 24	02:00
   Nov 24	01:45
   Dec 24	01:53
   Jan 25	02:15
   Feb 25	02:00
   Mar 25	02:30
   Apr 25	01:30
   May 25	01:30
   Jun 25	02:00
   Jul 25	03:00
   Aug 25	01:45
   Sep 25	01:30
   Oct 25	00:45
   Nov 25	01:15
   Dec 25	03:00
   Jan 26	03:00
   Feb 26	02:37
   Mar 26	02:30
   Apr 26	02:37
   May 26	01:45
   Jun 26	02:07
   Jul 26	01:30
   Aug 26	01:45

   Every month's median sits between 00:30 and 03:00. It never reached the goal.

   So did the late nights cost me anything? I compared each night's bedtime with how much of my daily habit list I did on the day that followed and, as a check, on the day before the night.

   • The day before the night
   • The day after the night

   Habits done, by bedtime. Average share of my daily habits done, by bedtime in one-hour groups labelled by their start. The day before is a control: if late nights only followed bad days, it would slope too.

   label	The day before the night	The day after the night
   <23:30	69%	63%
   23:30	71%	69%
   00:30	69%	72%
   01:30	71%	71%
   02:30	70%	70%
   03:30+	69%	69%

   Average share of my daily habits done, by bedtime in one-hour groups labelled by their start. The day before is a control: if late nights only followed bad days, it would slope too.

   There is no slope. The correlation between bedtime and the next day's completion is essentially zero (p = 0.67), and just as flat for the day before. The one bar that stands out is the early one, and it points the wrong way.

   • +0.02
     correlation, bedtime vs next day
   • 63% vs 71%
     habits done after an early night vs any other
   • 10.5 h vs 8.3 h
     sleep on those early nights vs others

   My guess is that those 41 early nights were tired nights and the long sleep pushed the whole day later, but I can't prove that and it is a small sample. "No evidence it helped" is the fair reading, not "it hurt".

   So I retired it. The habit mostly failed, but it did its job: it was the test. Sleeping late wasn't what was holding my consistency back. I had picked that up from the advice around me and never checked it.

   Every day91 entries, 14% done

2. Retired

   Forms filling

   Jan 2024

   I started this with my family as an expenses tracker, filling in a form each time we spent something. Apparently it only lasted a week as an idea, and the log agrees: it began on 17 January 2024 and was over by the 24th.

   Every day4 entries, 50% done

   Mon

   Tue

   Wed

   Thu

   Fri

   Sat

   Sun

   17 Jan

   18

   19

   20

   21

   22

   23

   24