Fifteen minutes of writing nobody will read, followed by the meta-analysis saying it barely works. The most recommended technique on the thread, held to the same standard as everything else.
10 May 2027
Third of the room’s picks. “Write it down” was the single most common suggestion in the thread, which is exactly why it is getting the full treatment rather than a nod. Still provisional, like the rest of the back half.
Before the session#
Bring something to write with and on. Paper if you can. Nothing else.
In the session#
Fifteen minutes, writing, no stopping.
The instruction is the one Pennebaker’s studies use. Write about something that is genuinely bothering you, and write about what happened and how you feel about it. Do not stop to fix a sentence. Spelling does not matter. Nobody reads this. Nobody collects it. At the end you can keep it, bin it, or burn it somewhere safer than campus.
If fifteen minutes on something real is too much today, write fifteen minutes on something trivial instead. The mechanics are the point and you do not owe anyone the subject matter. Passing entirely is also fine and needs no reason.
Then we talk about the experience, not the content. Was it a relief, a grind, or did it make things louder? Did you write toward something or in circles? Did fifteen minutes feel long?
Then the uncomfortable part#
Everyone recommends this. The evidence is much thinner than the recommendations suggest.
Frattaroli (2006, Psychological Bulletin) ran the largest meta-analysis of experimental disclosure. 146 randomised studies. The pooled effect was positive and significant at r = .075, which is roughly d = 0.15. That is small. Small enough that it sits below most of the other things on this syllabus, including a ten-minute nap.
Two earlier meta-analyses had reported larger effects. They used fixed-effects models, which assume every study is measuring one identical underlying thing. Frattaroli used random effects across a bigger pool and the number came down. That is a correction, not a scandal, but it is the sort of correction that never reaches the blog posts.
It has not improved since. A 2023 meta-analysis in Asian populations found no significant difference between expressive writing and control at first follow-up, g = 0.05, or at second, g = 0.04.
What the meta-analysis does find is a pile of moderators, which is another way of saying it depends. On who, on what about, on for how long, on how bad things were to start with.
Discussion#
- You have just done the thing and you have just been told it barely registers. Which of those do you believe about your own fifteen minutes?
- This is the third technique this course has reported as smaller than its reputation, after the sugar and the phone break. What do those three have in common in how they got popular?
- A small average effect across 146 studies is compatible with it working well for some people and doing nothing for most. Is that a reason to try it or a reason to stop recommending it to everyone?
- Is there a version of this where the writing is not the active ingredient and the fifteen uninterrupted minutes are?
Afterwards#
Nothing due. Week 10 is deliberately empty of deadlines, as Assessment explains. If the writing helped, it is a log entry. If it did nothing, it is a better one.
The spec
What you bring to the session. Written so a reader can tell whether it has been met without asking you.
- you have written for a set period without stopping, and can say what the experience was like independently of whether it helped
- you can state roughly how large the measured effect of expressive writing is, and why that sits awkwardly with how often it is recommended
Teaching team
- Idris Fenn, tutor