{"items": [{"author": "Hollis", "source_link": "https://www.facebook.com/jefftk/posts/691379003812?comment_id=691412576532", "anchor": "fb-691412576532", "service": "fb", "text": "I share that same concern. I've found that those sorts of studies have value for about a month (in gamifying happiness and making me want to do things to earn higher numbers) but that the effect wears off pretty fast and that it then becomes an erratic process so subjective that its error bars are larger than the 1-10 range.", "timestamp": "1412876344"}, {"author": "cbr", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl4zl1a", "anchor": "r-cl4zl1a", "service": "r", "text": "Followup to supplementing memory with experience sampling (2013).\n", "timestamp": 1412882984}, {"author": "IFuckingLoveSundays", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl52dxd", "anchor": "r-cl52dxd", "service": "r", "text": "This is pretty insightful, I liked the discussion on sampling time where he begins to doubt if tracking happiness over a long period of time actually makes sense. Some comment stated that a shorter time (month?) allows for quicker reaction and improvement. \n", "timestamp": 1412888092}, {"author": "Beth", "source_link": "https://www.facebook.com/jefftk/posts/691379003812?comment_id=691458579342", "anchor": "fb-691458579342", "service": "fb", "text": "I tried this briefly but changed replaced the numeric ratings with adjectives after a few days in, when I realized that there is no way to quantify feeling happy and angry at the same time.", "timestamp": "1412899032"}, {"author": "andero", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl5g7uc", "anchor": "r-cl5g7uc", "service": "r", "text": "Neat. I tracked my happiness for about three years before I figured I had learnt what I needed to and stopped.\n\n

Here is the spreadsheet of my data with pretty charts and graphs (It is Google so you will need to download the file to see it properly) (I think, let me know if the link is broken!)\n\n

METHODS: I used a retrospective daily recording. I would wake up in the morning, recall my previous day, and write down a series of numbers and letters before getting out of bed: High-Score, Reason for High-Score, Low-Score, Reason for Low-Score, Overall (not an average). The "Reasons" were limited to these tags: Meditation, Class (I was in school at the beginning), Social, Alone, Exercise, Drugs (including alcohol/caffeine), Food, Biological (headaches, etc), and Job. I was also able to double-tag (for example, a great lunch with friends could be SF for Social Food, or if the food was better than the company, FS)\n\n

Retrospective reports have limitations, but measuring something in the heat of things also has limitations, and random-daytime probes have limitations. Every method has limitations, but the thing to note is that the measurements were pretty consistent (I missed a few days in there, like OP, and should have added an extra tag for belated entries, but the vast majority were done the morning of the following day). This started as a general 1-10 scale, but I expanded it as needed. Rather than saying "well, I put a 6 here last time so, 6 again!" I went with what I felt, and I do have the introspection to get a decent read on my emotional state. Of course this is my idiosyncratic arbitrary scale; I would say writing this I am about an 8 happy (I am pretty happy right now) but maybe OP would rate this same feeling as a 6 or 7 and reserve a 10 for something else1.\n\n

RESULTS: See for yourself!\n\n\nDate of the month did not matter\nDay of the week did not matter\nEye-balling the monthly graph reveals an oscillating curve with a period of about 6-months (seasonal?)\nLow-scores remained relatively constant over the three years\nOverall-scores increased over the three years2\nHigh-scores increased over the three years2\nHigh-scores were usually social or alone time\nLow-scores were usually alone or social time\n\n\n

I did not know the proper statistical methods for analysing the reasons back when I did this so these come with a pile of salt (but also seemed face-value accurate to me):\n\n\nMeditation, Food, and Exercise typically resulted in stable higher scores\nSocial-time was usually a stable medium-high score while alone-time was less stable and lower scoring\nBiology was a pretty stable bad time (I had a headache-condition)\nJob was also a bad time, but less stable\n\n\n

The most interesting thing I learnt was from late October 2011: I recorded my emotional valence through a particularly rough breakup. Pretty suddenly, a girl I loved deeply broke up with me. This happened to be near the lowering states brought about that oscillating curve. November in particular shows vastly dropped low-scores, and those couple of months show perhaps slightly blunted high- and overall-scores, but it was nice to see that two months was all it took. Two months and the scores are indistinguishable from the pre-breakup scores. It was really nice to put a number on "my breakup healing time". \n\n

It was a fun experiment in self-tracking, but I know that when I started I would have to collect at least 6 months of data to see anything at all. The more I collected the more I saw, but then I felt like I reached a point of diminishing returns. It would be neat to track it for a decade, but daily stuff is too cumbersome for me to bother. I am tracking my personality changes for the long term (7 years in and pretty neat results so far!)\n\n

1 I personally feel like this is a problem with self-report scales, especially when using a finite scale. For example, OP says the scale used is 0-9, but OP never rates anything higher than an 8 or lower than a 4, so it seems to me that OP's scale is really a 4-8 scale, which is only a 5-point scale. If one wishes to capture the nuances of happiness, well, I know I needed a bigger scale to get the precision. My lowest score is a -4 and my highest score is a 13 so I covered an 18 point spread. Ultimately inter-scale comparisons are impossible. I could appear happier and more emotionally unstable than OP, but the opposite could be just as true since we are not using the same scale.\n\n

2 This increase in overall- and high-scores could be due to some kind of score-creep in reporting bias. The fact that my low-scores remained constant brings this explanation into question, however. Also, subjectively I did feel that I got happier over those three years so it seems very in-line with what I recall about that time in my life.\n", "timestamp": 1412917996}, {"author": "cbr", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl5ktga", "anchor": "r-cl5ktga", "service": "r", "text": "→ \n

High-scores were usually social or alone time, low-scores were usually alone or social time\n\n\n

I'm tempted to ask "what else is there?" but I see you have other categories (meditation, class, exercise) and so some experiences are neither "social" nor "alone".\n\n\n

Two months and the scores are indistinguishable from the pre-breakup scores.\n\n\n

I think "I stopped feeling bad and got back to normal" is probably the right interpretation, but how can you tell the difference between this and "I acclimated to my new range"? I mean, it's not plausible that a breakup would make you less happy for the rest of your life, but maybe it took longer to actually recover? Or less long, if there was some bias in the form of expecting to write down things more similar to what you'd been writing down lately.\n\n\n

For example, OP says the scale used is 0-9, but OP never rates anything higher than an 8 or lower than a 4, so it seems to me that OP's scale is really a 4-8 scale, which is only a 5-point scale.\n\n\n

The problem is I can easily imagine feeling more or less happy than 4/8. I expect those states to happen to me at times, so I want to have a wide enough range that I can capture them. I'm also not confident I could distinguish between smaller gradations than I'm using; 5-6-7 aren't actually very far apart in terms of internal experience, I'm a pretty stable person.\n", "timestamp": 1412940275}, {"author": "andero", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl5t49b", "anchor": "r-cl5t49b", "service": "r", "text": "→ → \n

some experiences are neither "social" nor "alone"\n\n\n

Yeah, it was more about the source of the emotion. I could be exercising alone and the aloneness has nothing to do with it, or I could be done exercising and showering and feel fantastic because of the exercise: in either case exercise is the reason. Alternatively, I could be home alone feeling terrible because of a negative social experience so social would be the reason even though I was alone.\n\n\n

how can you tell the difference between this and "I acclimated to my new range"\n\n\n

I guess because I lived it. If I understand you correctly, that would mean I got used to feeling bad and "bad" became my new normal. But then I would just feel bad so I would expect to see a drop in scores that became a stable drop. On the other hand, the scores go back to normal and that is in line with my experience of eventually getting over the relationship and moving on with my life.\n\n\n

if there was some bias in the form of expecting to write down things more similar to what you'd been writing down lately.\n\n\n

I do not really understand this. Why would I a priori expect the number to be similar day to day? It seems to me that I just wrote down whatever struck me in the moment as the correct number. The numbers change too so even post hoc it does not seem that there is a bias in trying to write similar numbers. I am probably just not clear on this idea of yours.\n\n\n

The problem is I can easily imagine feeling more or less happy than 4/8\n\n\n

In my view, that imagination is useless on your scale if you never use those numbers. Perhaps you had a very stable year, but it is also plausible that you had a sort of artificial ceiling/floor effect. I think the range of emotions available to a human are too vast to capture on such a small scale. It is like looking at the world population in terms of percentages; there are 7 billion people so ten-thousand deaths is a 0% drop in population.\n\n

To elaborate, I think if the emotional range was actually limited to 1-10 then all my scores would have been 5. The depths of pain possible are immense much as the heights of ecstasy are towering. The breakup was shitty, but I could imagine being drawn and quartered after being tortured for three weeks, having my finger-nails and teeth pulled out one by one and 2/3 of the skin on my body flayed off; with a finite scale, that experience would probably be a 1 or a 2 out of 10 since I can imagine worse. If that is near the bottom of the scale, my breakup was a 5. On the other hand, I could imagine mystical self-transcending feelings of unity, peace, and love, and remaining in that state all afternoon, which I guess would be about a 9 out of 10, because there is probably better. If that is near the top of the scale, the most caring and tender moment with friends while eating the most fantastic meal is a 5, maybe a 6.\n\n\n

I'm also not confident I could distinguish between smaller gradations than I'm using\n\n\n

Perhaps you would find more interesting results if you adopted a retrospective approach. Like you said, your phone does not probe you when you are at your lowest point of tending to a sobbing baby at 3am. If you just get probes during the day when you are engaged in mundane tasks it makes sense that you would have moderate numbers and stability. You are unlikely to get a probe just at the height of orgasm or just after you stub your toe, but if these were the highlight and lowlight of your day, you could remember them, as well as a gist/overall for the rest of the day. Also, I do meditate and apparently meditation experience predicts introspective accuracy.\n", "timestamp": 1412960160}, {"author": "cbr", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl5z0vm", "anchor": "r-cl5z0vm", "service": "r", "text": "→ → → \n

I could be exercising alone and the aloneness has nothing to do with it, or I could be done exercising and showering and feel fantastic because of the exercise\n\n\n

Makes sense. I guess there's some potential for thinking you feel good because you've been exercising but actually it was the time spent outside or something, but "why do I think I feel good" is probably good enough to optimize on.\n\n\n

If I understand you correctly, that would mean I got used to feeling bad and "bad" became my new normal.\n\n\n

Right. Imagine you lost most of your energy permanently, through something like chronic lyme disease. At first this would really suck, because each day feels like a regular day that you start exhausted. But then over time you get used to the new state and start reporting higher numbers. (Which is what we expect: most research on this finds that people have a "happiness set point" and after any sort of event they tend to drift back to reporting a similar level of happiness they reported before.) There are two reasons you might now be reporting higher levels of happiness: (1) you actually do feel almost as good as you did before you got lyme or (2) you've gotten used to the situation and it doesn't actively make you sad anymore. Ideally you could experience days in alternation, one day with lyme and one without, and then you could make a good comparison, but the longer you've been living with lyme the more remote your higher-energy days become and the harder it is to compare your current state to your pre-lyme state.\n\n\n

Why would I a priori expect the number to be similar day to day? It seems to me that I just wrote down whatever struck me in the moment as the correct number. The numbers change too so even post hoc it does not seem that there is a bias in trying to write similar numbers.\n\n\n

People often have vaguely mood-like things that last many days. For example you found oscillations at the 6-month level, sometimes I have a few weeks when I feel generally uninspired and bored, etc. When you've been feeling up/down lately the numbers you will be recording will as a whole be on the high/low side. If someone asked you "predict how you will be feeling tomorrow" you would predict high/low based on how you'd been feeling lately. So then when it comes time to ask yourself "how am I feeling now" you're going to be biased by the answer you would have expected yourself to give. For example, when I'm at work I generally am reasonably content but not super happy and usually put "6". But then if I'm unusually enjoying my work and get pinged I feel like "this is a situation in which I would usually put '6', because I'm working" and have to override that, really introspect, see how I'm feeling, and then put down, say, "7". And I might fail to do that, or just be pulled down towards my expectation so that I round the other direction from where I would if I didn't have the "I know I'm at work and when I'm here I tend to put '6'" bias.\n\n\n

Perhaps you had a very stable year\n\n\n

So, first off, I'm a very stable person. I tend not to get very happy or sad and I tend to generally feel on the happy side.\n\n\n

I think if the emotional range was actually limited to 1-10 then all my scores would have been 5. The depths of pain possible are immense much as the heights of ecstasy are towering.\n\n\n

My scale isn't linear. I want to have the most range available in the parts I'm likely to use. So the difference between 5 and 6 is small, while 6 to 7 is larger, 7 to 8 is even larger, and 8 to 9 is really extremely large. I do have some moments outside of the 4-8 range, but they're so brief that I'm really not likely to get pinged during one of them. Even 8 basically only happens when I get pinged at the top of some experience that is normally very good but in this circumstance is extremely good. (Spending time with friends I rarely see, and right after someone says something especially funny and I'm just starting to laugh I get pinged.)\n\n\n

You are unlikely to get a probe just at the height of orgasm or just after you stub your toe, but if these were the highlight and lowlight of your day, you could remember them, as well as a gist/overall for the rest of the day.\n\n\n

These I already capture through the normal human means of "remembering important things". What I want experience sampling for is to get access to a view of the past that doesn't weight extreme experiences more highly. Your remembered past, either generally or as written highlights and lowlights, is like a movie: full of the important stuff but skipping over all the mundane details. And to a large extent that's what I care about. But I also want to be able to see how much of my life was spent waiting for the bus.\n", "timestamp": 1412971177}, {"author": "andero", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl64n3j", "anchor": "r-cl64n3j", "service": "r", "text": "→ → → → \n

Makes sense. I guess there's some potential for thinking you feel good because you've been exercising but actually it was the time spent outside or something, but "why do I think I feel good" is probably good enough to optimize on.\n\n\n

As I said, any self-report methodology has this problem. This is strictly unavoidable. Even if we were to use an objective bio-marker for happiness, we would first need to develop objective bio-markers by correlating them to self-report, and we are back to square-one.\n\n\n

Right. Imagine you lost most of your energy permanently, through something like chronic lyme disease...\n\n\n

I actually had a chronic pain problem called "New Daily Persistent Headache" that lasted about three years. I had a headache 100% of the time I was awake (and sometimes in my dreams). I can confidently say that I did not "get used to it". Sure, people tend to have a "happiness set point", and I was happy during much of the time when I had headaches, but I was also in pain and was not as happy as I would have been had I been pain-free, and introspection revealed this to me trivially. Though I did start to forget what it was like to not have headaches, I never forgot that it was vastly superior to having a headache. All I can say is that my experience does not align with your proposed explanation.\n\n\n\n

Why would I a priori expect the number to be similar day to day?\n\n\n

...If someone asked you "predict how you will be feeling tomorrow" you would predict high/low based on how you'd been feeling lately. So then when it comes time to ask yourself "how am I feeling now" you're going to be biased by the answer you would have expected yourself to give.\n\n\n

Nope, I am going to introspect in that moment and come up with a number based on how I am feeling. Why would I give a number for now that is based on a number I predicted in the past? That seems extremely counter-intuitive to me since now is available to me right now and all I need do is introspect.\n\n

Consider the weather: If someone asked you "predict the weather tomorrow" you would predict high/low based on how the weather has been lately. So then when it comes time to ask yourself "how is the weather now" you're going to be biased by the answer you would have expected yourself to give. Nonsense! You are going to look outside or read a thermometer. If you predicted rain and it is sunny, you are not going to be biased by your prediction, you are going to report sunny skies.\n\n\n

For example, when I'm at work I generally am reasonably content but not super happy and usually put "6". But then if I'm unusually enjoying my work and get pinged I feel like "this is a situation in which I would usually put '6', because I'm working" and have to override that, really introspect, see how I'm feeling, and then put down, say, "7". And I might fail to do that, or just be pulled down towards my expectation so that I round the other direction from where I would if I didn't have the "I know I'm at work and when I'm here I tend to put '6'" bias.\n\n\n

That is totally counter-intuitive to me. I just - whomph - introspect on a moment's notice. I guess I am just a more accurate introspector than you? Like I linked, meditation can apparently do that.\n\n\n

So, first off, I'm a very stable person. I tend not to get very happy or sad and I tend to generally feel on the happy side.\n\n\n

Yup, me too. (Breakup is an obvious exception but generally I am very stable, calm, and content, and even moreso now than at the time of collecting this data)\n\n\n

My scale isn't linear ... the difference between 5 and 6 is small, while 6 to 7 is larger, 7 to 8 is even larger, and 8 to 9 is really extremely large. ... 8 basically only happens when I get pinged at the top of some experience that is normally very good but in this circumstance is extremely good.\n\n\n

What can you learn from your data when all you get back is a whole bunch of 6s? This is a double-trouble for you since even if you are actually having a 7 in a regularly 6 situation, your personal bias is to go based on what you thought you would feel (a 6) rather than introspect on what you actually feel in that moment (a 7). It sounds to me like you are extremely limiting your range, additionally so since it is non-linear in a vague way.\n\n

My question for you becomes 'why do you want it to be easier to write a 6'? You are consciously biasing your scale toward being a 6. I think this makes your data a lot less useful. For example, what if you wanted to see a movie and were checking out some reviews, but every movie you saw was a 6/10? This information is almost useless for making a decision.\n\n\n

These I already capture through the normal human means of "remembering important things". What I want experience sampling for is to get access to a view of the past that doesn't weight extreme experiences more highly. Your remembered past, either generally or as written highlights and lowlights, is like a movie: full of the important stuff but skipping over all the mundane details. And to a large extent that's what I care about. But I also want to be able to see how much of my life was spent waiting for the bus.\n\n\n

Oh... okay... I guess if you are trying to sample mundane moments and you are biased toward reporting mundane scores it makes sense that you are going to get mundane data...\n\n

I do not mean to be a bummer, but I think I totally do not understand what you were trying to quantify. My big question for you would be: What good is your data to you now that you can see it clearly? What can you learn about yourself from your data? Can the data you collected answer the questions you had when you set out? What were those questions? Now that you have been going for a year, are you sure you are using the right methods to answer those questions?\n", "timestamp": 1412983236}, {"author": "cbr", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cllfay4", "anchor": "r-cllfay4", "service": "r", "text": "→ → → → → \n

introspection revealed this to me trivially ... all I need do is introspect ... I just - whomph - introspect on a moment's notice.\n\n\n

It sounds like we have very different experiences with introspection. Maybe it's the meditation helping you, maybe it's just something I'm kind of bad at. If I try to think about how I'm feeling it usually goes like "Hmm, I guess I feel fine? Pretty good, this is nice, I don't know?"\n\n\n

It sounds to me like you are extremely limiting your range, additionally so since it is non-linear in a vague way. ... why do you want it to be easier to write a 6\n\n\n

The non-linear scale expands my range dramatically! Imagine I had a linear scale from 1-10. I could be both way less and way more happy than I am now, which means I would probably put down "5" all the time. I'm not sure I have ever had moments in my life I would rate above or below "5" on a linear 1-10 scale. So to let there be finer distinctions in the part of my range I spend the most time in, I need a non-linear scale. Another way to do this would be to have a 1-10000 scale, and use numbers like 500, 2000, 4000, 4998, 4999, 5001, 5002, 6000, 8000, 9500, but that seems awkward.\n", "timestamp": 1414439639}, {"author": "andero", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cllnjus", "anchor": "r-cllnjus", "service": "r", "text": "→ → → → → → \n

Imagine I had a linear scale from 1-10. I could be both way less and way more happy than I am now, which means I would probably put down "5" all the time.\n\n\n

It has been a while, but it sounds like you are repeating my criticism back to me:\n\n\n\n

To elaborate, I think if the emotional range was actually limited to 1-10 then all my scores would have been 5. The depths of pain possible are immense much as the heights of ecstasy are towering. The breakup was shitty, but I could imagine being drawn and quartered after being tortured for three weeks, having my finger-nails and teeth pulled out one by one and 2/3 of the skin on my body flayed off; with a finite scale, that experience would probably be a 1 or a 2 out of 10 since I can imagine worse. If that is near the bottom of the scale, my breakup was a 5. On the other hand, I could imagine mystical self-transcending feelings of unity, peace, and love, and remaining in that state all afternoon, which I guess would be about a 9 out of 10, because there is probably better. If that is near the top of the scale, the most caring and tender moment with friends while eating the most fantastic meal is a 5, maybe a 6.\n\n\n\n

Like I said, using a finite 1-10 scale is a terrible idea.\n\n\n

Another way to do this would be to have a 1-10000 scale, and use numbers like 500, 2000, 4000, 4998, 4999, 5001, 5002, 6000, 8000, 9500, but that seems awkward.\n\n\n

On a scale of 1-10,000 you are unlikely to experience 9500. Instead, how about only using a subset of the whole possible emotional scale because you will only experience a subset of possible emotional states. As your numbers show, you are more likely to experience around the 4000-6000 range, so rather than a 1-10,000 scale you could use a 4000-6000 scale and re-normalize that to be a 0-2000 scale or a (-1000) to (+1000) scale. You do not need such precision, though, especially since you do not have introspective access to such precision in the first place, so you cut it by a factor of 100 and you get a 0-20 scale. You re-normalize that to be a (-5) to (+15) scale, and this is my de facto scale. You start using numbers in the 1-10 range and expand as needed, but most of the numbers remain in the 1-10 space, yet you still get interesting variance between scores because you have the precision you need to measure the values of interest.\n\n

Think of it like "visible light" on the electromagnetic spectrum. Your eyes are not sensitive to radio waves or gamma rays, they are sensitive to a narrow band we call visible light. Sure, saying "visible light" cuts out a lot of the spectrum, but that is a lot of the spectrum your eyes were not going to detect anyway.\n\n\n

It sounds like we have very different experiences with introspection. Maybe it's the meditation helping you, maybe it's just something I'm kind of bad at. If I try to think about how I'm feeling it usually goes like "Hmm, I guess I feel fine? Pretty good, this is nice, I don't know?"\n\n\n

Maybe I am a highly sensitive person. Maybe it is the meditation (as research suggests). If you want to get more introspective prowess, you could try meditation. If you do not care to get more introspective prowess, so be it.\n", "timestamp": 1414454941}, {"author": "Jeff Kaufman", "source_link": "https://www.facebook.com/jefftk/posts/691379003812?comment_id=691607106692", "anchor": "fb-691607106692", "service": "fb", "text": "@Toby: For the most part, I think I'm putting '6' for \"happy, as usual\", '5'/'7' for \"a bit less/more happy than usual\", and '4'/'8' for \"significantly less/more happy than usual\". I'm not going for an interpersonally comparable score, though one that works for me over time would be helpful. Unfortunately I suspect my concept of \"level 6 happiness\" just isn't that stable.", "timestamp": "1412938086"}, {"author": "quantifiedself", "source_link": "http://www.reddit.com/r/QuantifiedSelf/comments/2isbgp#cl6c02p", "anchor": "r-cl6c02p", "service": "r", "text": "This is a great post. Have you thought of talking about it at a Quantified Self meetup?\n", "timestamp": 1413002666}]}