Support this blog!
If you enjoy this kind of data-driven educational content, click here to check out my Patreon page and support future work (membership is $2 per month!) Membership will come with a number of perks (see page for details.) I’d like to keep providing content that includes original surveys, but it isn’t cheap!
Note: Each section will have a Section Summary at the top. If you want the highlights without the details, you can read those and move on!
Gratitude is healthy.
Section Summary
Gratitude-inducing practices (e.g., gratitude journaling) yield mental and physical health benefits. However, it is important to find practices that work for you (and if your inability to keep doing it makes you feel guilty, then just don’t worry about it.)
Gratitude is associated with greater well-being, life-satisfaction, physical and emotional health, and ability to manage stress (read more about this here.) Of course, demonstrating an association, e.g., “people who have x quality are healthier/more successful/happier/etc.” does not mean x quality is the reason why — the world is full of associations (just look at my other posts!) — causal mechanisms tend to be harder to dig up. Thankfully, psychologists have worked on ways to see if increasing gratitude might lead to some of these other positive outcomes. Typical interventions that have proven to increase gratitude include:
Gratitude journaling: chronicling the things one is grateful for each day (or “counting blessings). A variation on this is the “Three Good Things” exercise in which people write down 3 good things that happened and identify what caused them.
You don’t need to buy an expensive journal for this, but this is a nice one by Kurzgesagt.
Gratitude letter: writing a letter (or message) to someone who they wish to thank.
Gratitude visit: reading a gratitude letter (or just expressing gratitude) in-person to the individual they’d like to thank.
These methods of increasing gratitude can have beneficial emotional and behavioral effects, including reducing anxiety, increasing life-satisfaction and positive mood as well as increasing prosocial behaviors. This doesn’t mean that everybody should go out and buy a journal (I’m terrible at journaling…) — these approaches aren’t one size fits all and the effects on wellbeing are not very big. That being said, picking whichever method of practicing gratitude you are most comfortable with is likely to at least add a bright spot to your day.
Are we more grateful on Thanksgiving?
Section Summary
Yep. There’s an estimated 6.2% increase in gratitude on Thanksgiving relative to regular weekdays, which is similar to the effects of other gratitude interventions reported in the psychological literature.
“Are people more thankful on Thanksgiving?” may seem like a ridiculous question to ask, but this is a place for ridiculous questions! Moreover, instead of just assuming “of course they are!” this kind of study can put some sort of measurement to this question (albeit, a still pretty imprecise one...) There are numerous ways to go about answering whether gratitude increases on Thanksgiving, with pros and cons to them all. I chose a simplistic approach, asking people to “Please indicate how strongly you feel the following emotions at the moment“, with 17 emotion words that they would rate from 1: Not at all to 5: Very Much. The 3 emotion word of interest for this post are “Thankfulness”, “Gratitude”, and “Appreciativeness”. I sent out this survey on Nov. 22 (day before Thanksgiving), 23 (Thanksgiving), 27 (Monday after), 29 (a week from the first survey), and 30 (a week from Thanksgiving) — meaning the design is observational (i.e., we can’t rule out alternate explanations for differences.)
Thankfulness, Gratefulness, and appreciativeness
Given the name of the holiday, looking at how thankful people reported feeling seems like a good place to start. Judging by the figure below, it appears that there is more thankfulness on Thanksgiving relative to other days (I felt comfortable just comparing Thanksgiving to the aggregate of all of the other days due to their similarity.) Breaking down the different response options, we see that there is an estimated 8.1 percentage-point increase in people choosing “Very much” for thankfulness on Thanksgiving as compared to the average non-holiday (we also see a corresponding reduction in people choosing “A little”, “Not much”, and “Not at all”, with no real difference in people choosing “Somewhat”.)
Figure 1. Estimated percentage of people choosing each Thankfulness category.
How to interpret this graph: This is a series of stacked bar charts. For each day, we have a percentage of people who chose one of the five response options. If you add up the percentages within each day it should equal roughly 100% (though there may be some errors due to rounding.)
For stats nerds: these estimates were calculated via ordinal Bayesian regression (with a cumulative link function). We also ran an additional model with just Thanksgiving vs. not Thanksgiving — the overall effect of Thanksgiving was b = 0.37 95% CI [-0.04, 0.81], probability of direction = 0.96. The estimated increase in people choosing “Very much” was 8.1 pp 95% CI [-.01, 0.18],
The results for gratitude ratings looked very similar to thankfulness, with an 8.7 percentage point increase in selecting “Very much” for gratitude (which makes sense, as people who rated thankfulness highly also tended to do so for gratitude (r(373) = .87). As for appreciativeness ratings, there was a slightly smaller, 5.3 percentage point bump on Thanksgiving (which, according to conventional norms wouldn’t be considered “significant”. It also had a slightly weaker, though still strong, association with thankfulness; r(373) = .76.) All of these results held even when accounting for happiness ratings, which means that it is unlikely that these effects were just because people were feeling positive in general.
How does this compare to the scientific literature?
I found one meta-analysis (which is just a paper that compiles the results from lots of similar studies on the same topic in order to get a sense of the aggregate findings) on the effect of typical gratitude interventions on gratitude ratings. In this paper, they calculated an overall average increase between 3.4-5.7% (depending on the analysis) in gratitude ratings. For an in-kind comparison, I computed a “gratitude score”, which was just the average of thankfulness, gratitude, and appreciative ratings for each individual. The average estimated gratitude score on non-holidays was 3.57 (out of 5), and that increased by an estimated 0.22 points to 3.79 on Thanksgiving, which comes out to roughly a 6.2% relative increase. In other words, the effect of the holiday is pretty similar to (if not slightly larger than) the average effect found in the literature! I do think it is worth taking a pause here to consider whether you think that is an impressive effect size for a national holiday centered around thankfulness (that is, being on par with journaling) or not… It is also worth noting, a gratitude journal is a more practical approach to increasing gratitude relative to building a national holiday around it (though personally, I’ll take pie over journaling.)
If this study was a turkey, it would be a bit dry…
Section Summary
Some things to consider when drawing conclusions about this study: I couldn’t randomize people into celebrating Thanksgiving or not (so other factors could explain the results), I couldn’t follow the same people throughout the week or randomize who got a survey on which day (so perhaps the folks surveyed on Thanksgiving were just more grateful people), and I may be measuring a demand effect (people feel like they should report more gratitude even if they don’t feel it.)
My original question was, “Are people more thankful on Thanksgiving?“ Given that a gratitude exercise tends to be a Thanksgiving meal staple, the increase may not come as a surprise. However, there are a number of weaknesses with the approach I took that warrant consideration.
First, I surveyed different people on different days and wasn’t able to randomize who responded on which day. It is possible that people who do online surveys on Thanksgiving are simply a more grateful crowd. If that were the case, I’d probably expect there to be other differences aside from just the gratitude emotions. As you can see below, that isn’t really the case. The only solid differences are for thankfulness and gratitude, and it seems unlikely that I just happened across a group that showed this pattern of differences on Thanksgiving by chance (though I can never know for sure!)
Figure 2. Estimated difference in ratings between people on Thanksgiving and people on other days. (Note, the scale goes from 1 to 5, so the differences could potentially range from -5 (Thanksgiving rating is a 0 and no-holiday rating is a 5) to 5 (Thanksgiving rating is a 5 and no-holiday rating is a 0.)
How to interpret this graph: The dots are the best estimates for the average differences. The lines around the dots represent our uncertainty for those estimates — we have uncertainty because we only surveyed a small sample of people, so we need to reflect just how much information that sample can provide us in terms of the estimate of interest! The thick lines around the dots represent the interval by which there’s a 66% chance that the difference lies along that region. The thinner, wider intervals reflect a 90% chance that the difference lies along that region.
For stats nerds: these estimates were calculated via mixed-effects Bayesian regression (where we interacted holiday and emotion with participant as a random effect.)
Another issue is that there is no control group! I can’t randomly assign some people to celebrating Thanksgiving and others not! So there could be something I can’t account for driving these results. I wanted to try a quasi-experimental design called difference-in-differences, in which I surveyed people from countries who do not celebrate Thanksgiving (though I’d still have to hope that they are similarly grateful on non-holidays for that to work), but I wasn’t able to do this on the survey platform I used… c’est la vie.
One could make an argument that I’m just measuring one big demand effect. In other words, people aren’t actually more thankful, but they feel like they should report being more thankful because it’s Thanksgiving! I can’t really know if this is the case, but I do wonder if the same criticism could be leveraged at typical gratitude exercises, and if so, does it matter whether one is being genuine or just going through the motions?
Finally, I chose to ask people how they felt in the moment. That means I was capturing a thin slice of their experience on whichever day and time they took the survey. One could argue that I should have waited until the end of the day and asked them to reflect on their general feelings for that day. I think both approaches have strengths and weaknesses, but for this study, I felt that catching people in the moment might be less prone to the demand effects I mentioned earlier (as compared to asking them to reflect on Thanksgiving.) That being said, I could also be underestimating the effect of the day by taking this approach.
Conclusions
Taking all the caveats about study design into account, I’m comfortable going forward with a moderately strong belief that Thanksgiving leads to a small gratitude bump. Whether the size of the effect is surprisingly large or small depends on your prior belief (I’m sure some of you will be a bit disappointed that there isn’t more gratitude flowing and others will be surprised that there’s any effect at all.)
An important note
I would be remiss if I did not point out that there are plenty of reasons to feel angry (as opposed to grateful) during this holiday. The Thanksgiving myth may be all full bellies and friendships, but in reality, it is rooted in the devastation of native populations through disease and war, resulting from colonialism. So while I think a holiday that focuses on gratitude is probably a good thing, it is worth grappling with the specific history of this holiday and finding a way to build a Thanksgiving that can be truly inclusive of all Americans.
Like the article? Consider supporting for more like it!
Click here to check out my Patreon page!
Survey details
Sample size
375 participants (USA) from Cloud Research Crowdsourcing platform
Nov. 22: 48; Nov. 23: 97; Nov. 27: 95; Nov. 29: 47; Nov. 30: 88
Demographics
47% woman; 51% man; 1% Non-binary; < 1% Agender/prefer not to say
Median age = 39 (youngest = 18, oldest = 79)
Materials
Used the gratitude measurement from https://osf.io/preprints/psyarxiv/29ebh.