Happiness Logging: One Year In
I’ve been logging my happiness for a year now. [1] My phone notifies me at unpredictable intervals, and I respond with some tags. For example, if it pinged me now, I would enter “6 home bed computer blog”. I always have a numeric tag for my current happiness, and then additional tags for where I am, what I’m doing, and who I’m with. So: what’s working, what’s not?
When I first started rating my happiness on a 1-10 scale I didn’t feel like I was very good at it. At the time I thought I might get better with practice, but I think I’m actually getting worse at it. Instead of really thinking “how do I feel right now?” it’s really hard not to just think “in past situations like this I’ve put down ‘6’ so I should put down ‘6’ now”.
Being honest to myself like this can also make me less happy. Normally if I’m negative about something I try not to dwell on it. I don’t think about it, and soon I’m thinking about other things and not so negative. Logging that I’m unhappy makes me own up to being unhappy, which I think doesn’t help. Though it’s hard to know because any other sort of measurement would seem to have the same problem.
There’s also a sampling issue. I don’t have my phone ping me during the night, because I don’t want it to wake me up. Before having a kid this worked properly: I’d plug in my phone, which turns off pings, promptly fall asleep, wake up in the morning, unplug my phone. Now, though, my sleep is generally interrupted several times a night. Time spent waiting to see if the baby falls back asleep on her own, or soothing her back to sleep if she doesn’t, or lying awake at 4am because it’s hard to fall back asleep when you’ve had 7hr and just spent an hour walking around and bouncing the baby; none of these are counted. On the whole, these experiences are much less enjoyable than my average; if the baby started sleeping through the night such that none of these were needed anymore I wouldn’t see that as a loss at all. Which means my data is biased upward. I’m curious how happiness sampling studies have handled this; people with insomnia would be in a similar situation.
Another sampling issue is that I don’t always notice when I get a ping. For the brief period when I was wearing a smartwatch I was consistently noticing all my pings but now I’m back to where I sometimes miss the vibration. I usually fill out these pings retroactively if it’s only been a few minutes and I’m confident that I remember how I felt and what I was doing. I haven’t been tagging these pings separately, but now that I think of it I’m going to add an “r” tag for retroactive responses.
Responding to pings when other people are around can also be tricky. For a while there were some people who would try and peek and see what I was writing, and I wasn’t sure whether I should let them see. I ended up deciding that while having all the data eventally end up public was fine, filling it out in the moment needed to be private so I wouldn’t be swayed by wanting to indicate things to the people around me.
The app I’m using isn’t perfect, but it’s pretty good. Entering new tags is a little annoying, and every time I back up the pings it forgets my past tags. The manual backup step also led to some missing data—all of September 2014 and some of August—because my phone died. This logging data is the only thing on my phone that isn’t automatically backed up to the cloud, so when my phone died a few weeks ago I lost the last month of pings. [2] So now there’s a gap in the graph.
While I’m not that confident in my numeric reports, I’m much more confident in the other tags that indicate what I’m doing at various times. If I’m on the computer I very reliably tag ‘computer’, etc. I haven’t figured out what to do with this data yet, but it should be interesting for tracking behavior chages over time. One thing I remember doing is switching from wasting time on my computer to on my phone; let’s see what that looked like:
I don’t remember why the big drop in computer use at the end of February 2014 happened. I assumed at first it was having a baby, after which I spent a lot of time reading on my phone while she was curled up on me, but that wasn’t until a month later. I think this may have been when I realized that I didn’t hate the facebook app on my phone afterall? I’m not sure. The second drop in both phone- and computer-based timewasting, the temporary one in July 2014, was my being in England. My phone had internet but my computer usually didn’t. And there was generally much more interesting stuff going on around me than my phone.
Overall my experience with logging has made me put less trust in “how happy are you right now” surveys of happiness. Aside from the practical issues like logging unexpected night wake-time, I mostly don’t feel like the numbers I’m recording are very meaningful. I would rather spend more time in situations I label higher than lower on average, so there is some signal there, but I don’t actually have the introspection to accurately report to myself how I’m feeling.
I also posted this on my blog.
[1] First ping was 2013.10.08 06:31:41, a year ago yesterday.
[2] Well, it was more my fault than that. The phone was partly working and I did a factory reset to see if that would fix it (it didn’t) and I forgot to back up pings first.
I can confirm that doing this makes you less happy. Right now if you asked me whether I was happy, I’d say, sure, I’m happy. But if you asked me to make a 1-10 rating, I’d put it at maybe a 6. Well, that doesn’t seem so happy anymore. It seems lacking relative to where I could be. Do this for a few days and you realize you’re almost never at or near a 10. It’s grinding, and your subjective experience of your life becomes defined, not just measured, by the happiness score.
I used to rate my happiness and productivity on 1-10 scales at the end of each day, and this was my experience too. I’ve since dropped that part of my diary routine, instead focusing on just writing the three best things and three things to improve/do more/do next. I still have an idea if I look back later of trends of happiness and productivity, since I can see some good things are better than others, but I don’t have that feeling of disappointment in myself every time I don’t make an 8+ for both.
That said, the only way I can analyze this is looking back over it, I can’t input it to make informative graphs or the like.
What about alternate measures?
Binary comparisions like: “Would I rather: Experience the previous hour exactly as it occured a second time, or X for an hour?”. Where X might be : sleep, or some reference experience.
The abscence of negative mood. This isn’t the same as happiness, but something like boredom or anger could have a clearer floor, or lend itself to a “Present, a little, none” scale.
Or more specific questions. Here’s a PDF of an adjective based scale. An idea is to pick out only those adjectives that you already expect to vary.
I think ordering/ranking experiences would be more successful (in general) than trying to just give them scores.
An example of such a system: Every ping asks you to briefly describe the previous hour, and then shows you a list of every other ping you’ve written for the last week/month or so. You then put the description wherever it fits in the list, above everything that was less fun to experience, and below everything that was more fun.
In this way it’s very easy to notice happiness trends (whether or not you’re getting happier or sadder over time) without worrying about associating the same activity with the same score, even if it’s becoming less or more fun to do.
My worry in that case would be present conditions bleeding into the memory and evaluation of those earlier pings. For example, I’d expect that when you’re hungry your relative ranking of past ping moments is going to change to more heavily weigh moments when you were eating.
I’ve also been suspicious of happiness surveys for a similar reason. One theory I have is that a large portion of the variation in happiness set-point is just that different people have different tendencies in answering “rate in 1-10”-type questions. It would be interesting to test how much does happiness set-point correlates with questions such as “rate this essay from 1 to 10″. Another test for this theory that is far more like have actually been conducted already is to see how well happiness set-point correlates with neurological signals of happiness (the difficulty being here that the primary way to determine whether a neurological signal signals happiness is through self-report. Nonetheless, if the happiness set-point correlates with any neurological signal then it more likely that this signal plays a role in happiness than in inducing high number ratings).
Perhaps one way to improve the measurement would be to structure the question in terms of preference rather than direct measurement—something like “what is the least-good thing I’d rather be doing right now” (ie as a kind of upper bound to current happiness). But there’s also an issue of energy level etc.
This is a really cool idea. But even preference has issues. For example, I like contra dance (a kind of social dancing) a lot, and have a good time when I go. The feel in the moment is one of my favorite things. If you asked me, “would you rather be contra dancing” I would usually say yes. But if you look at my behavior, I don’t actually go that often anymore, even when I do have free time. How do you tell the difference between me irrationally underconsuming something I enjoy vs me overestimating how much I enjoy it in posed comparisions?
For certain formulations of this, that objection seems irrelevant. Imagine that instead of a 1-10 scale, you had a ranked list of activities (or sets of activities).
Like opportunity cost in economics.
Max L.
Have you found any interesting correlations?
I haven’t. I’d actually be a little more interested in trying some randomized experiments; with correlations it’s so hard to know if there’s really a common cause.
If this experiment in what makes you happy is a) not turning up interesting correlations, b) takes time out of your day to log and curate, and c) is actively making you unhappy, then why are you still doing it? It seems like the one singular result is that doing the research is bad for your personal well being. If I were you I’d cut it out of my life right now.
In general the more data I have the more useful it gets. I don’t think it makes me much less happy, and it doesn’t take very much time, so I’m willing to pay those small costs to keep the data analysis options open.
One thing worth mentioning about introspective accuracy is that the numbers people give are similar across cultures, across time for a single person, and they lit similar areas of fMRIs (Layard 2005). What about using a 1-5 scale?
To some extent I already am. I’ve never scored my happiness above 8 or below 4, though I can imagine being in states outside that range. This same preference for leaving room for more extreme future experiences would lead me to use a 1-5 scale as a 2-4 scale.
Interesting. Re: phone timewasting, I’ve found having a crappy phone with a slow data connection helpful for avoiding this. Also saves money. (I have a Kyocera Event on Virgin Mobile’s Beyond Talk plan, FWIW.)
I do enjoy timewasting, and I want to continue to do some amount of it. For example, I count reading facebook, lesswrong, people’s blog posts, etc in this category, and if I stopped doing that I would feel both less connected to my friends and less engaged with the world’s ideas. But the marginal returns diminish, as always, and sometimes I do it when I should be doing something else.
One idea for improving the quality of your ratings: instead of taking quantitative, numerical measurements, do qualitative observations (write a paragraph verbally describing how happy you’ve been lately). Then, after a while, look over your accumulated qualitative observations and try to cluster them or develop some kind of rubric or decision procedure that allows you to quickly code your happiness level. (For example, you might be able to do some kind of happiness binary search… if you know what your median level of happiness feels like, you could diff that against your current level of happiness and figure out which way the diff went, then after that diff yourself against the 25th/75th percentile level of happiness. That’d give you 4 buckets. Depending on the shape of your happiness distribution, you might find this percentile-based approach not all that useful, especially since the distribution of your happiness might change depending on the circumstances, etc.)
Part of my goal with tracking happiness is to better understand my in-the-moment experience of life as opposed to my remembered experience. This means any tracking probably needs to be quick and frequent.
I think part of the problem here, and why writing a qualitative description is a good idea, is that forcing yourself to “quantify” your happiness (an incredibly difficult thing to do even in the best of circumstances) is going to inherently lead to inaccuracies. I’d even consider making the argument that having to quantify your happiness could inadvertently lead to you actually being -less- happy. Imagine it—you feel fine one moment, then, as a result of a ping, you now have to evaluate exactly how you’re feeling and your day so far and turn it into a number. What if that number doesn’t come out the way you want it to? Now you’re unhappy.
That’s a lot harder to do with a qualitative observation, where you can quickly spitball how you feel at the immediate moment without in-depth contemplation.
Makes sense. I think this is still compatible with my suggestion though: you could write a paragraph describing how you feel in this moment. It would be slower for a while but if you followed the rest of my advice you could go back to quick tracking fairly soon.
I guess the problem is I’m not even willing to go through a single week where at random times I might need to write a paragraph on my phone? My current system is fast enough I can almost always immediately respond to a ping, and if it would be too socially awkward I can decide what I’m going to enter and then enter it in a minute or so when there’s a break.