Gratitude journaling: the research, honestly

Almost every gratitude app, including ours, tells you the practice is “backed by research.” Very few of them tell you what the research actually found, and none of the ones we checked mention the part where the authors rate their own evidence as low certainty.

This page is the long version. It names the papers, gives the numbers with their confidence intervals, and states plainly where the evidence is thin. The short version, if you only want one sentence: gratitude journaling has a real but small effect on wellbeing, and the quality of the evidence behind that finding is weaker than the confidence with which it gets repeated.

We make a gratitude journaling app, so we have an obvious interest in you believing this works. Weigh what follows accordingly — and note that the honest reading below is less flattering to us than the marketing version would be.

The two meta-analyses worth knowing about

Two separate research teams published meta-analyses of gratitude interventions in 2023. They used different methods and different outcome measures, and they broadly agree with each other, which is more informative than either one alone.

1. Kirca, Malouff and Meynadier (2023) — the effect size

Kirca and colleagues pooled 25 randomised controlled trials covering 6,745 participants and asked whether expressed gratitude interventions improved positive indicators of wellbeing: life satisfaction, positive affect, and happiness (1).

The result: a statistically significant effect, Hedges’ g = 0.22, 95% CI [0.11, 0.33], p < .001, measured against neutral comparison groups.

Two things about that number deserve more attention than they usually get.

First, 0.22 is a small effect. By the conventional rough benchmarks, 0.2 is small, 0.5 is moderate, 0.8 is large. This is at the low end of small. It is real — the confidence interval does not cross zero — but it is not transformative, and any page telling you gratitude journaling will change your life is not describing this literature.

Second, “relative to neutral comparison groups” is doing a lot of work. A neutral comparison is typically something like writing about your daily activities, or nothing at all. It is a much easier bar than an active control — another structured writing exercise of equal effort and plausibility. Effects measured against neutral controls are generally larger than effects measured against active ones. So 0.22 is closer to a ceiling than a floor.

One useful null finding from the same paper: intervention length did not significantly moderate the effect. Doing it for longer did not reliably produce more benefit. If you have been feeling guilty about a gratitude practice you kept for three weeks rather than three months, this is the paper that says the guilt is misplaced.

2. Diniz and colleagues (2023) — the effect, and the caveat that matters most

The second team ran a broader systematic review, identifying 64 randomised trials and including 33 in the meta-analysis, across children, adolescents, adults and older adults (2).

Their pooled results, reported as mean differences:

Outcome	Mean difference (95% CI)	Reported improvement
Life satisfaction	0.48 (0.21 to 0.75)	~6.9%
Mental health	0.29 (0.17 to 0.41)	~5.8%
Anxiety symptoms	−1.63 (−2.37 to −0.89)	~7.8% reduction
Depression symptoms	−1.86 (−2.89 to −0.83)	~6.9% reduction
Gratitude scores	—	~3.7–5.7% higher

Every one of those intervals excludes zero, so the direction is consistent. Improvements in the 6–8% range, across thousands of people, from an exercise that costs a few minutes a day.

And then the sentence that almost never travels with those numbers: the authors rate the certainty of the evidence as low to very low, because of high heterogeneity between studies and risk of bias within them.

That is the authors’ own assessment of their own result, and leaving it out is the single most common way this literature gets misrepresented. It does not mean the finding is wrong. It means the studies varied a great deal in design and quality, that publication and reporting biases are a live concern, and that the true effect could be meaningfully smaller — or, less likely, larger — than the pooled figure suggests.

The paper’s stated conclusion is that acts of gratitude can serve as a therapeutic complement alongside treatment for anxiety and depression. That is the authors’ framing of their clinical finding, not a description of what an app does. JotMood is self-care, not treatment, it is not a medical device, and nothing on this page should be read as a claim that a journaling app treats a condition.

Where the practice came from

The protocol most gratitude apps are descended from is Three Good Things, published in the research of Seligman, Steen, Park and Peterson (2005) and distributed free by UC Berkeley’s Greater Good in Action (3, 4).

The instructions are unglamorous, and worth reading because they are more specific than the folk version:

1. Give the event a title.
2. Write down what actually happened — where you were, what you did, what people said.
3. Include how it made you feel at the time, and how it feels now.
4. Say what you think caused it to happen.
5. Don’t worry about grammar or style.

Recommended dose: about ten minutes a day, for at least a week.

Two details are routinely lost in translation. The original asks what caused it — the causal step is part of the exercise, not decoration. And it asks for what happened, not what you are grateful for in the abstract. “My sister called” is the exercise working. “Family” is not.

What the honest summary looks like

Putting the two meta-analyses together:

• The effect is real and consistently positive in direction across thousands of participants.
• The effect is small — around g = 0.22 on wellbeing measures, roughly 6–8% shifts on symptom scales.
• The certainty of the evidence is low to very low, by the reviewers’ own rating.
• Longer is not reliably better. Duration did not moderate the effect.
• Most of the strongest-looking results are against neutral comparisons, not active ones.

Here is the thing that makes this worth doing anyway, and it is an argument about cost rather than about effect size. A small effect from an intervention that takes ten minutes a day, costs nothing, has no meaningful side effects and requires no clinician is a genuinely reasonable trade. Small effects are only disappointing when the price is high. The correct expectation is not transformation; it is good light — the room is unchanged, and you can see more of it.

Where it doesn’t work, and who should skip it

The literature and the practice both have edges worth naming.

In acute distress, forced gratitude can backfire. Being asked to list good things while in real crisis can read as instruction to dismiss what is actually happening. Nothing in these meta-analyses supports gratitude journaling as a response to acute distress, and this is the point at which a person, not an app, is the right answer.

Self-report is nearly all of the evidence. Wellbeing, life satisfaction and gratitude are measured with questionnaires, and people who have just spent a week writing about good things may answer questionnaires about good things differently. That is a known limitation of the field, not a flaw specific to these papers.

It can become another chore. If listing three things has turned into an obligation you resent, the intervention has stopped resembling the one that was tested. Skipping days is not failure — and it is the reason we don’t count them.

It is not a substitute for treatment. See the note above about what the Diniz conclusion does and does not say.

References

1. Kirca A, Malouff JM, Meynadier J. The Effect of Expressed Gratitude Interventions on Psychological Wellbeing: A Meta-Analysis of Randomised Controlled Studies. International Journal of Applied Positive Psychology, 2023. doi:10.1007/s41042-023-00086-6. — 25 RCTs, 6,745 participants; Hedges’ g = 0.22, 95% CI [0.11, 0.33].
2. Diniz G, Korkes L, Tristão LS, Pelegrini R, Bellodi PL, Bernardo WM. The effects of gratitude interventions: a systematic review and meta-analysis. einstein (São Paulo), 2023;21:eRW0371. PMID 37585888. — 64 trials identified, 33 meta-analysed; certainty of evidence rated low to very low.
3. Seligman MEP, Steen TA, Park N, Peterson C. Positive psychology progress: empirical validation of interventions. American Psychologist, 2005;60(5):410.
4. Greater Good in Action, UC Berkeley. Three Good Things. — The free, public protocol: ~10 minutes a day for at least one week.

If you want to read further

If you want the shorter, less technical version of this, does gratitude journaling actually work? covers the same ground in about a third of the length. If you want to do it rather than read about it, 40 gratitude journal prompts go deeper than “three good things,” and what to write when nothing happened is for the ordinary evenings where the practice usually breaks down.

If the wider question is which technique suits you, 12 journaling techniques explains what each named approach is actually for. And if streaks are the thing that has ended every journaling habit you have started, do streaks help or hurt? reports that literature the same way this page reports this one — including the trial that argues against our own design.

If you would rather have the prompt handed to you, JotMood is a gratitude journal app built on the five-minute version of this: one tap for a mood, three lines of gratitude, a monthly read of the patterns underneath, and no day counting anywhere.

JotMood is self-care, not therapy, and it does not diagnose or treat anything. If writing brings up more than you want to hold on your own, please talk to a qualified professional. In the US you can reach the 988 Suicide & Crisis Lifeline by calling or texting 988.