Does gratitude practice actually work? The evidence, honestly graded

From the founder

Educational content, not medical advice. Gratitude practice is a well-being tool, not a treatment. If you're dealing with depression or persistent low mood, that deserves real care from a professional; a journal is a supplement to help, not a substitute for it.

I'll admit my bias up front: "gratitude journal" sounds like something embroidered on a throw pillow. My professional instincts run toward interventions with confidence intervals. So when I started building an evening routine for myself and kept running into gratitude practice in the literature, my first move was to try to dismiss it. I couldn't quite, but I couldn't endorse the popular version either. It's one of the most-studied practices in positive psychology, and the honest reading of the evidence is more interesting, and a good deal more deflating, than either the hype or the eye-roll.

Here's that honest reading: what the trials actually show, which of the famous results didn't survive replication, how much shrinks once you compare gratitude against something rather than nothing, and the version of the practice that survives contact with a skeptic.

The foundational experiments, and what actually held

The paper that launched the field ran three studies, and it's usually quoted as though it ran one. In the first, undergraduates were randomly assigned to write weekly about things they were grateful for, about hassles and annoyances, or about neutral events. The gratitude group came out ahead on well-being and optimism.^[1] That study also produced the two most repeated statistics in the entire gratitude canon, that grateful writers reported fewer physical symptoms and exercised more, and I'm not going to repeat them as findings, because the same paper's other two studies looked for both and found neither. Positive affect was the one result that held up across all three.

There's a subtler problem with the design that the field took years to name. The comparison group was writing about hassles, which is a mild bad-mood induction rather than a neutral control, so "gratitude beats hassles" might partly mean "listing your irritations makes you feel worse." When a Spanish team re-ran the study with the baseline and follow-up measures the original lacked, the results replicated only partially, and the positive-affect gap looked more like decline in the hassles group than improvement in the gratitude group.^[2]

The second landmark study is the source of the famous "three good things" exercise: write three things that went well today and why. Several hundred people did one week of assigned exercises, and three-good-things was one of only two conditions still showing benefit six months later.^[3] That's the version you've heard. Here's the part that rarely travels with it. The authors reported that people who kept doing the exercise voluntarily were the happiest, and the durable effect is concentrated in that self-selected group, which makes "one week of writing bought six months of happiness" an overstatement of what was shown. Participants were also recruited from the website for the lead author's own book, so the sample arrived pre-sold on the idea.

Worse for the clean story, two independent teams later re-ran that design with a positive placebo, asking people to write about early memories instead. One found the placebo produced happiness gains as large and as durable as the real exercises, with no advantage on depression.^[4] The other put it bluntly: every condition improved happiness and reduced depression, and the positive-psychology exercises were statistically indistinguishable from the placebo.^[5] The replication record isn't uniformly negative, but the two studies built specifically to test whether these exercises do something particular could not show that they do.

The honest grading: what the meta-analyses say

Now the skeptic's turn, because single studies flatter every intervention. When researchers pooled the gratitude trials, the picture that emerged was small, and possibly not specific to gratitude. One meta-analysis found benefits for well-being that shrank sharply when gratitude was compared against other active, structured exercises rather than against nothing at all; its authors described the result as weak evidence and wrote that gratitude interventions may operate primarily through placebo effects.^[6] A second, focused on depression and anxiety, found a small effect on depression, no significant effect on anxiety, and against realistic comparison groups an effect its authors called trivial.^[7] The largest synthesis yet, 145 papers across 28 countries and nearly 25,000 people, lands in the same place: a small increase in well-being.^[8]

That last clause about active comparisons is the single most important fact in this literature, and it's almost never included in articles like this one. Most of what makes gratitude journaling look impressive is that researchers compared it to doing nothing. Measured against any other structured two-minute writing exercise, most of the advantage disappears.

I find this deflating and still worth acting on, which I realize is an unsatisfying place to land. A free practice with a two-minute dose, no side effects, and small effects that may not be specific to gratitude is a reasonable thing to do and a terrible thing to sell. It's the claims of transformation that should trigger your skepticism, and, to be fair to you, a fair amount of skepticism about the practice itself is warranted too.

The surprising part: the sleep connection

This is the thread that moved gratitude from "fine, whatever" to a permanent slot in my own evening, because it connects to the energy story this whole site is about. Researchers studying gratitude as a trait found that more grateful people slept better, and when they traced the mechanism it ran through pre-sleep cognitions: what your mind is chewing on as you lie in the dark.^[9] Grateful people rehearse different material at lights-out, less threat and rumination and more of what's actually fine, and that difference statistically accounted for the better sleep. Grade it carefully, though: it's a survey taken at a single moment, so it can't tell you which way the arrow runs, and the arrow demonstrably runs both ways, since sleeping badly is itself very good at draining your gratitude.

For actual trial evidence there's one small randomized study: two weeks of a gratitude diary improved self-reported sleep quality alongside the mood gains, in about 40 people per arm, all young women, measured by questionnaire.^[10] Set against that, the current synthesis of 64 gratitude trials found that only insomnia-severity scores favored gratitude, while sleep-quality measures and the standard sleep-quality index showed no difference at all.^[11] And no gratitude study has ever moved sleep measured by a device rather than a questionnaire.

So I'd put it this way, having initially wanted to put it more strongly: there's a coherent mechanism, one supportive small trial, a correlation that runs both directions, and a meta-analysis that comes up empty on the main sleep measure. Gratitude earns its two minutes in the evening wind-down because the last channel your mind is tuned to before sleep plausibly matters and this is a cheap way to change it, not because anyone has shown it reliably improves sleep.

The mechanism, minus the part I can't support

The best-supported version of the mechanism is simply that bad weighs more than good. Across an enormous range of findings, negative events, emotions, and feedback have larger and longer-lasting effects on us than equivalent positive ones.^[12] One critical remark outlives five kind ones. A day with six things that went fine and one that went wrong gets filed under the one.

I want to flag what I'm not claiming, because I've made this mistake myself. The tempting version is that demanding work leaves you with a threat-tuned attention system, and that this is the price of being good at the job. The research doesn't support the specific mechanism. When researchers pooled the studies on threat-related attentional bias, they found a clear effect in anxious people and essentially none in everyone else.^[13] It's a marker of anxiety, not a general feature of how attention works, and the one study I found that tested whether a gratitude exercise shifts it came back null. I also can't point you to any evidence that scanning harder for what's wrong makes anyone a better clinician; if anything the vigilance literature suggests that narrow focus is bought by missing things elsewhere.

So the honest framing is smaller and still worth something. Gratitude practice is not about denying problems or manufacturing positivity. It's a couple of minutes that force a deliberate accounting of what went right, on the reasonable theory that what went right is systematically underweighted in memory. You're not reciting affirmations. You're correcting for the fact that bad is louder.

Minimum protocol

Here the research has something genuinely useful and slightly inconvenient to say, because the format turns out to matter more than the frequency.

Write in sentences, not bullets. The largest test of gratitude format, nearly a thousand people, compared letters and essays against lists. Longer-form writing outperformed lists on well-being, positive emotion, and gratitude itself, and structured lists did not differ from control conditions on any outcome.^[14] I have to be straight about what that implies for the version of this practice I like: a bare three-bullet list is the format that failed. If you write three items and stop, you may be doing the inert version. Write the reason too, in an actual sentence, or write one thing at slightly more length. That's the difference between logging and re-contacting.

Be specific. "Maya laughed at my terrible joke at dinner" does more than "my family," because specificity forces you back into the moment rather than gesturing at a category. Specificity is also what makes the sentence-writing above possible; you can't explain why a category mattered.

Evening placement is my inference, not a finding. No study has compared morning versus evening gratitude. I do it at night because the mechanism that has the best correlational support runs through what your mind is chewing on before sleep, so that's when a channel change should count for most. Reasonable, untested, and I'd rather label it than dress it up.

Don't force daily perfection. Trials have gotten results from weekly and two-week protocols, and missing nights doesn't break anything. Freshness beats streaks: the practice dies when it becomes rote recitation of the same three items, so reach for what was actually true today. The related research finding is that these practices work best in people who genuinely endorse and keep choosing them, which is a decent argument against turning this into another streak you defend.^[15]

Keep what you write. This one is untested and I'll say so; it's simply what I've found. A gratitude entry pays twice, once when you write it and again when you re-read it months later, because a record of what was fine on an ordinary Tuesday is exactly the evidence memory is worst at preserving.

And as always: no app required. A pen, two or three minutes, a few specific true things from today written out with the reason they mattered. It will feel slight, and I'm not going to tell you the trials say otherwise, because they mostly say the effect is small and may not be unique to gratitude. What I'll say instead is that it costs almost nothing, it has a plausible mechanism, the small effects are real, and it is one of the few things you can do at 10 p.m. that reliably makes the day you actually had easier to see. That's enough reason to keep doing it. It isn't a reason to sell it as anything more.

Reserve EMC is available on the App Store and Google Play. Your data stays on your device. This article is for education and isn't a substitute for care from your own clinician.