How To Keep A Gratitude Journal, And What It Does

The top of this search is mostly notebooks for sale. Beautifully made ones, some of them, and one of them is literally titled "Backed by Science" on the product page. So it seems worth writing down what the science actually says, because I read the papers this week and the picture is more interesting than either the sellers or the sceptics let on.

Here's the short version. A gratitude journal, in the form the research tested, means writing down up to five things you're grateful for, at a fixed interval, with a short explanation attached to each. The original 2003 experiment used a weekly interval over ten weeks, not a daily one. The measured benefits are real but small, and the largest meta-analysis on gratitude interventions and mood found a pooled effect of g equals -0.29 on depression and anxiety symptoms, with the authors recommending that anyone actually trying to treat those symptoms use something with stronger evidence. It's a cheap habit with a modest return. That's a fine thing to be.

The Protocol From The Study Everyone Cites

Every page on this topic traces back to one paper, usually without saying what was in it.

Robert Emmons and Michael McCullough published "Counting Blessings Versus Burdens" in the Journal of Personality and Social Psychology in 2003, volume 84, pages 377 to 389 (PDF via Greater Good). Study 1 took 201 undergraduates in a health psychology class, ended up with 192 after dropouts, and split them into three conditions with roughly 65 people each.

The gratitude group got this instruction, word for word.

"There are many things in our lives, both large and small, that we might be grateful about. Think back over the past week and write down on the lines below up to five things in your life that you are grateful or thankful for."

Note "up to five" and note "the past week." Participants filled in a packet of ten weekly reports. The second group listed hassles, defined for them as "irritants, things that annoy or bother you." The third listed neutral events. So the comparison wasn't gratitude against nothing, it was gratitude against deliberately dwelling on annoyances and against a neutral listing task.

The Greater Good Science Center at Berkeley, which is where a lot of the popular guidance descends from, now recommends a somewhat different protocol, fifteen minutes a day at least three times a week for a minimum of two weeks. That's a considerably heavier dose than the original weekly listing exercise, and I think it's worth knowing that the famous study and the famous instructions aren't the same thing.

What Study One Actually Found, Including The Null

This is the part that gets skipped, and it's the reason I'd trust the finding more rather than less.

The gratitude group did rate their lives as a whole more favourably than the other two groups, and reported better expectations for the coming week. They reported fewer symptoms of physical illness than either comparison group. And they spent nearly an hour and a half more per week exercising than the hassles group, which is an odd result that nobody has ever satisfactorily explained to me.

Then there's this sentence from the results. "However, participants in the gratitude condition did not differ significantly from participants in the hassles or events condition on either the positive or negative affect factors."

In the weekly version, listing your blessings did not significantly move measured positive or negative emotion against the comparison groups. The authors' own abstract puts it as the gratitude groups showing heightened wellbeing "across several, though not all, of the outcome measures," with the effect on positive affect being "the most robust finding" across all three studies taken together. Study 2 ran daily rather than weekly and Study 3 used people with neuromuscular disease, and the picture across them is genuinely mixed rather than uniformly positive.

I find that far more convincing than the version where everything worked. Papers where everything works are usually papers where something went wrong.

What Twenty Years Of Trials Added Up To

Now the aggregate, which is where honesty gets expensive.

David Cregg and Jennifer Cheavens at Ohio State pooled 27 studies with 3,675 participants in the Journal of Happiness Studies in 2021, in a paper whose title asks outright whether gratitude interventions work as self-help (paper here). The effect on symptoms of depression and anxiety came out at g equals -0.29 at post-test, standard error 0.06, and -0.23 at follow-up. Small, in the conventional reading of that scale.

Two other things in there matter more than the headline number.

First, moderation analyses showed effect sizes were larger in studies using waitlist controls rather than active control conditions. Which is to say, gratitude journaling looked better when compared against doing nothing than when compared against doing some other structured thing. That gap is where a lot of the placebo question lives, and the authors flag prior reviews raising exactly that concern.

Second, their closing recommendation. They write that because the effects are relatively modest, they recommend people seeking to reduce depression and anxiety symptoms engage in interventions with stronger evidence for those symptoms. That's the authors of the biggest gratitude meta-analysis telling you it isn't the treatment. Not one gratitude journal listing I looked at mentions it.

None of this makes the practice worthless. A small effect from a habit that costs three minutes and no money is a perfectly reasonable trade. It just isn't the thing that fixes a depressive episode, and anyone selling it that way is selling something.

The Protocol Variables, And Which Ones Have Support

Six things you can change about how you do this. Here's what's actually behind each.

The "why" clause deserves its own note. The durable version of this practice in the literature is Seligman and colleagues' three good things exercise, published in American Psychologist in 2005, where participants wrote three things that went well each day for one week and also wrote an explanation of why each happened. Berkeley's write-up of that protocol reports participants showing increased happiness immediately and at one week, one month, three months and six months later. One week of practice, effects reported half a year out. That's the strongest single result in this area and the causal explanation was part of the instruction, not an optional extra.

So if you're going to do one thing differently from the generic advice, add the because.

The Specificity Ladder

Every guide says be specific and then gives you examples that aren't. Here's the ladder I'd actually use, from the level most people write at up to the level that seems to be doing the work.

Level 1, the category. "My family." "My health." "Coffee." This is what nearly everyone writes by week two and it's the level at which the habit dies, because you're producing a word list and you know it.

Level 2, the instance. "My sister called on Tuesday." Better. There's an event.

Level 3, the instance plus the detail. "My sister called on Tuesday and stayed on for forty minutes even though she was clearly cooking."

Level 4, the instance plus the because. "My sister called on Tuesday and stayed on for forty minutes even though she was clearly cooking. That happened because she'd noticed I went quiet in the group chat last week and decided to do something about it."

Level 5, the subtraction. "If she hadn't called I'd have gone another fortnight without telling anyone about the work thing." Berkeley's guidance calls this trying subtraction rather than addition, imagining the absence rather than counting the presence.

Prompts that push you up the ladder rather than sideways along it:

• What did someone do this week that they didn't have to do?
• What went right that I'd have blamed someone for if it had gone wrong?
• What's currently working that I'd notice immediately if it stopped?
• Who made a decision that benefited me and probably cost them something?
• What am I taking for granted this month that I wasn't a year ago?
• What almost went badly and didn't, and why not?

That last one is my favourite because it catches the near misses, and near misses are the category of good fortune that leaves absolutely no trace unless you write it down.

If you want the plain version without the theory, there's a simpler gratitude list approach, and a full walkthrough at gratitude journaling for beginners.

When It Backfires

The interesting thing is that the people who ran the original study saw this coming, and said so in the same paper.

Emmons and McCullough wrote that gratitude "has an obligatory aspect," that to be grateful is to accept the position of a recipient and to feel indebted and dependent on others, and that most people experience indebtedness as an unpleasant and aversive state. Their words are that making people aware of what they have to be grateful for "might increase their recognition of the need to reciprocate, and people may resent these obligations and even report strong negative feelings toward their benefactors."

That's in the introduction of the founding paper. Twenty-three years of wellness content has been built on top of it without repeating it once.

The practical read. If your gratitude list is populated by people you feel you owe, or the exercise reliably produces guilt rather than warmth, that's a known mechanism and not a personal failing. Same if the list keeps reminding you of what you have while you're in the middle of losing something, which is its own particular flavour of unhelpful. Grief has a different practice attached to it and I'd point you to writing through grief rather than to a blessings list.

I'd also say, and this is judgement rather than evidence, that gratitude journaling seems worst suited to the exact moment people reach for it. Nobody starts one on a good week.

Which is the point at which I should say the obvious thing. A notebook is not treatment. If what you're carrying is heavier than a weekly list can hold, or the exercise keeps producing guilt and flatness rather than anything warmer, that's worth taking to a qualified professional rather than to a better template, and if you're in immediate crisis then a crisis line is the right call and this article is not.

The Notebook Question

Do you need to buy one? No.

The moderator analysis in the big expressive writing meta-analysis found handwriting versus typing came out at r equals .004 across studies, which is nothing at all. What did move was privacy, with people writing alone in a private room showing larger effects than those writing alongside others, and writing at home outperforming controlled settings.

Buy the nice notebook if the nice notebook is what gets you to do it. That's a real reason. Just don't believe the object is the mechanism, and if you're weighing a printed guided journal against a text file, the honest difference is structure and pacing rather than efficacy. The five minute journal format is the most popular commercial take on this and it's essentially the protocol above with a fixed template.

What I'd Actually Do

Weekly, not daily, on a set day. Up to five things. One sentence of because attached to each. Fifteen minutes if I have it, four if I don't.

And I'd write down the near misses, because in my own work the thing I've consistently failed to notice is the times something didn't go wrong. I've shipped seven paid toolkits, I've had a site sit at average position 47 in search results for months, and I've got a music channel with seven subscribers, so I've had a lot of practice at cataloguing what isn't working. The counting is easy in that direction. It's the other direction that needs a form.

Nine weeks would be a reasonable trial before deciding, which is roughly what the original study ran. If nothing's happened by then, nothing's going to, and there are other kinds of entry worth your fifteen minutes. The full guide to journal prompts covers what those look like and why the wording changes what they do.