By dkl9, written 2025-083, revised 2025-083 (0 revisions)
From 2024-190 thru 2025-051, I recorded selected numbers about my daily life. Originally, those numbers were just about my experiences with music, for an experiment on the effects of music. At several points I added more variables to track, and was tracking quite a few by 2024-340. I did all this figuring that I'd do statistics on all those records and learn something about how I live. I put off such analysis until now.
The entire study was observational: I let myself do whatever seemed proper in the moment, or whatever I was tempted to do, and the records were only relevant after it happened. Early on, I had a random boolean generator largely control whether I listened to music, but I stopped randomising that before adding any more variables to track. All statistics here will show correlations, but leave it ambiguous which variables affect which. If I claim here that one thing causes another, that's just sith I confidently assumed how things work.
Many variables, I recorded by the obvious "remember what it was and put it in the spreadsheet the next day". I used fancier (hopefully more reliable) methods for a few variables:
My mood was expressed on a scale of -1 to 1, and averaged for each day. My earworm-state was one of silent (N), deliberately-picked (D), recently-heard (R), or some other song (O), totalled into a count for each state, for each day. Both measurements are biased, in that I could only keep track of them when I was at my laptop.
Auto-collected data was timestamped to UTC. I live in the eastern US. My dataset thus groups into the next day what, to me, was the late evening of each day.
I kept track of whether I listened to any substantial amount of music, whatever its kind, each day. On days with music (there were 84), my mood averaged as 0.14, compared to the 0.12 from days without music (there were 103): I seem to be a tad happier when listening to music, whichever way the causality. My earworm-state was 22% N/3% D/23% R/52% O on days with music, and 25% N/12% D/10% R/54% O on days without music. It appears listening to music mostly makes my mind play what I heard recently more, and what I consciously pick less. Any effect on earworms is confounded by how I opted to record that the song in my mind is the song that plays before me from the outside world, which condition I usually described as R.
I worry so much about avoiding and controlling earworms, but does that even make me happier? Improvisedly, the mood associated with an earworm-state is the average mood across all days, weighted by the number of samples from each day at which I held that earworm-state. This turns out as 0.14 for N, 0.19 for D, 0.15 for R, and 0.13 for O. I'm notably happier on days when I tended to (state D) have picked the song that plays in my mind. The higher value associated with R may come from R's use to label the time when I actively listen to music.
How about music's effect on (or, less likely, effect from) my cognitive performance? I use my average time per prompt reviewed in Memoire as a proxy — lower is better. I listened to music on 83 days with substantial Memoire review, with average review time 5.3 s, treated as one data point per day, sans weighting. On 37 other days, when I didn't listen to much music, my reviews took an average of 5.5 s. Often, I'd listen to music while I reviewed notes in Memoire, and it seems that helped me recall a tad faster.
A similar weighted average as for mood seems to show that I recalled faster (5.2 s) with earworm type D or R than with types N or O (5.3 s).
I kept track of roughly how much I ate, in "plates per day". "Plate" is a vague unit, but I tried to have it mean roughly the same amount over time. It appears that I usually (Q1 to Q3) ate 5 to 7 plates per day (170 days).
Where both recorded, the amount I ate correlated positively with my mood the same day by r = 0.08. The correlation is stronger with mood the following day (r = 0.1), and weaker with mood the preceding day (r = 0.06). Average mood (across days), as a function of how much I ate, peaked as 0.13 at 6 plates.
When recorded, I usually (Q1 to Q3) "slept" (was contiguously in bed) for 7.7 to 10.5 hours. I usually got up between 08:11 and 10:40 — a good Wakeful Club would've helped — and went to bed between 23:32 and 00:47. How long I was in bed was much better explained (r = 0.8) by when I got up than (r = -0.44) by when I went to bed. I went to bed later the next night after I got up later each morning, by r = 0.48.
Where both recorded (90 days), how long I "sleep" had what rounds to zero correlation with my mood the following day. It correlates negatively with my mood the preceding day, by a weak r = -0.03. Naively, that means I sleep more after I'm sad, but the amount I sleep doesn't affect how I feel the next day.
Does sleep affect my earworms? I calculated ratios N/(D+R+O), D/(R+O), and R/O for each day. How long I sleep the night before correlated with those ratios by r = -0.06, -0.1, and -0.01, respectively. This seems to mean that more sleep leads me to have earworms more often, and leads more of my earworms to be accidentally "picked".
How much I ate in one day correlates negatively (r = -0.08) with how long I slept the night before, but positively (r = 0.16) with how long I slept the following night. But there's a likely confounder: long sleep correlates with getting up late correlates with skipping breakfast and, sometimes, lunch correlates with eating less. As it turns out, when I got up each day correlated negatively (r = -0.13) with how much I ate, as did when I went to bed the following night (r = -0.02).
Does sleep help me recall better (here, faster)? I'm told it should, but sleeping longer one night led me to recall slower (r = 0.02) the following day.
Sometimes, I meditate, shortly after I get up. When I did (65 days), I recorded how long I did so, in minutes, and otherwise (100 days) recorded 0. When I meditated, it was usually (Q1 to Q3) for 7 to 15 minutes.
Where both recorded, my mood was, on average, 0.1 when I didn't meditate (86 days), and 0.15 when I did (52 days). Confusingly, among days when I meditated, how long I meditated correlated negatively (r = -0.14) with my mood. On the one hand, perhaps on many days with long meditation, I sampled mood only once or a couple times, messing up measurements. On the other hand, -0.14 seems a bit much to come from just that kind of mistake. If these relations are causal, to be happiest, I should meditate each day, but only briefly.
My earworm-state was 24% N/3% D/19% R/53% O on days without meditation, and 23% N/2% D/22% R/53% O on days with meditation, i.e. basically the same. Recalling notes took an average of 5.3 s without meditation, and 5.4 s (worse) with.
I recorded a boolean variable whose meaning I shall keep private, except that it's bad when true, and expected it to correlate with sleep. When the variable was true (9 days), I had slept for an average of 10.6 hours. When it was false (92 days), the average was 9.1 hours. I got up, on average, at 10:38 on days with this variable true, and at 9:29 on days with it false.
Memoire recall time correlated negatively with mood by r = -0.1. That is, I was happier on days when I recalled notes faster.
For each of several types of useful activities, I recorded how many times I did it within the day (78 days). From those, I calculate a "score" for each day: a sum of those counts, weighted by the relative importance and difficulty of each task. The distribution of scores was very skewed right. They negatively correlated with mood by r = -0.11.
Half of these results are wrong sith my mood-records are sparse and a bit dishonest. Another half are wrong sith my computer's concept of a day misaligns with my practical concept of a day. These halves overlap, so some results remain valid.
I should probably listen to music, to be happier and recall faster. The direct effect on earworms is limited and acceptable. I should sleep a bit less, preferably by getting up earlier. It seems that would reduce earworms, help keep a private variable false, and matter little for mood and memory. I should probably meditate, but we knew that already.