Mood tracking patterns: what a month really shows

Four weeks of daily entries and you finally scroll back to look at the whole month. The shape is right there: weekends up, midweek sagging, one bad stretch near the start. It looks like a finding.

It's mostly a calendar.

Weekends lift almost everyone, which is what makes them useless to you

The weekend lift in mood is one of the best-documented patterns in mood data, and most of it is explained by who you spent the time with rather than by the day itself. In a 2015 PLOS ONE analysis, John Helliwell and Shun Wang worked through roughly 940,000 responses from working US adults collected between 2008 and 2012. Weekend effects turned up for all seven emotions they measured. Workplace trust, workplace relationships and the extra hours spent with family and friends together accounted for almost all of the weekend lift in happiness, laughter and enjoyment.

A pattern that holds across 940,000 people is not news about you. It is the background hum of having a job, and your tracker charts it first because days are the easiest thing to sort by.

Four weeks only gives you four Tuesdays

A month of daily logging gives you about four samples of each weekday, which is enough to notice something and nowhere near enough to believe it. One rough Tuesday — a short night, a dentist, a delayed train — moves your Tuesday average by a quarter. Two of them and the app draws a confident little dip under Tuesday that means nothing at all.

Weekday averages are the most fragile numbers in your log and the ones every chart puts on the first screen. The written entries underneath don't have that problem, because a sentence doesn't average with the sentence next to it.

Look for a repeated noun, not a repeated day

The pattern worth finding in a month of mood tracking is a word that keeps reappearing, not a day that keeps dipping. Three things survive a small sample:

• A repeated noun. The same name, the same project, the same room showing up on three of your five worst days. That is a problem with a handle on it.
• A run rather than a date. Four flat days in a row is a stretch you can go and ask questions about. One bad Wednesday is a Wednesday.
• The gaps. The days you skipped logging cluster too, and they tend to cluster somewhere honest.

None of this works if the month is a column of bare scores, which is the whole argument for writing a reason beside the number. Six words does it. In Grateful', our three-good-things journal, that note sits on the same entry as the day's three lines, so reading a month back is reading sentences instead of counting digits.

A pattern you can name is still not a pattern you can change, and reading your own logs back closely makes the mood louder for some people rather than clearer. If you already turn things over at night, do the read-back in daylight and leave the log alone the rest of the month.

Tonight, open the last four weeks and cover the dates. Read only the reasons, in any order, and see which word you have written more than twice.