with Spencer Greenberg
the podcast about ideas that matter

Episode 076: How to measure impact, and why we may have all been doing it wrong (with Michael Plant)

Enjoying the episode? Want to listen later? Subscribe on any of these apps or stores to be notified when we release new episodes:

October 25, 2021

Researchers in the Effective Altruism movement often view their work through a utilitarian lens, so why haven't they traditionally paid much attention to the psychological research into subjective wellbeing (i.e., people's self-reported levels of happiness, life satisfaction, feelings of purpose and meaning in life, etc.)? Are such subjective measures reliable and accurate? Or rather, which such measures are the most reliable and accurate? What are the pros and cons of using QALYs and DALYs to quantify wellbeing? Why is there sometimes a disconnect between the projected level of subjective wellbeing of a health condition and its actual level (e.g., some people can learn to manage and cope with "major" diseases, but some people with "minor" conditions like depression or anxiety might be in a constant state of agony)? What are some new and promising approaches to quantifying wellbeing? The EA movement typically uses the criteria of scale, neglectedness, and tractability for prioritizing cause areas; is that framework still relevant and useful? How do those criteria apply on a personal level? And how do those criteria taken together differ conceptually from cost-effectiveness? How effective are psychological interventions at improving subjective wellbeing? How well do such interventions work in different cultures? How can subjective wellbeing measures be improved? How can philosophers help us do good better?

Michael Plant is the Founder and Director of the Happier Lives Institute, a non-profit research institute that searches for the most cost-effective ways to increase global well-being. Michael is also a Research Fellow at the Wellbeing Research Centre, Oxford. He has a PhD in Philosophy from Oxford, and his thesis, entitled Doing Good Badly? Philosophical Issues Related to Effective Altruism, was supervised by Peter Singer and Hilary Greaves.

JOSH: Hello, and welcome to Clearer Thinking with Spencer Greenberg, the podcast about ideas that matter. I'm Josh Castle, the producer of the podcast and I'm so glad you've joined us today. In this episode, Spencer speaks with Michael Plant about research in the effective altruism community, metrics of subjective well-being, and evaluating impact and prioritizing interventions.

SPENCER: Michael, welcome.

MICHAEL: Hi, it's me.

SPENCER: So there are a lot of different people that tell us there are different ways to do good in the world. And the effective altruism community has a particular take on this. And one thing, I think, is really interesting about your work is that in many ways, you're coming from an effective altruist perspective, asking how do we do the most good, but your lens on this is quite different than most effective altruists. And I think you have some really interesting ideas that can make the conversation more nuanced, more complex about what it means to do the most good. So I'm excited to dig into that with you.

MICHAEL: Alright.

SPENCER: So just want to layout just very basically, what is your approach to thinking about how to do more good?

MICHAEL: Yeah. So it's interesting that you say, I'm taking a controversial of a different approach from other effective altruism, I think that's half true and half not true. So, I mean, where my side, start with a background thinking, okay, that's probably the most good, what do we mean by good, I tend to think of the good in terms of happiness. And if you might think that's the only thing which matters, even if you don't think that's the only thing which matters gonna be up there on your list.

SPENCER: So when you say happiness, I assume you mean both the positive and the negative, like reducing suffering and improving well-being.

MICHAEL: So what happiness I mean, you're happy if you're experiencing overall positive contradicts, if you feel good. So this is a fairly obvious place to start and work on. The new stuff I've been doing at the effective altruism came from, is that I was aware of all of this background research in social science, which was trying to measure happiness. It has been going on for decades. And I thought this is great like this is obviously the kind of thing we should be using to try and figure out how to do the most good. And people like Peter Singer, and Will MacAskill in their 2015 books on effective altruism had nods to this. Will MacAskill said you can use these health metrics, QALYs and DALYs, as this sort of staple for doing cost-effectiveness comparisons, like really we'd like to be using, at least we'd like to have some measure of happiness. And I'm really mapping this research. So you know, why don't we use it. And then Peter Singer says the same thing, kind of, has this nod to subjective well-being. So what I mean by that, subjective well-being, is this sort of term of art in social science. So better well-being refers to their self-reported measures of happiness, and life satisfaction, and sometimes, meaning as well. One of the staple measures, there's life satisfaction, which is just found simply by asking zero to ten. Overall, how satisfied are you with life nowadays? And I thought this stuff is really valuable. It's actually quite a lot of information on it. It just seems puzzling that this wasn't the kind of thing that effective altruism had been using. I thought, let's make use of this.

SPENCER: Got it. So social scientists developed this measure, which says, basically, rate yourself on a scale of one to ten. How satisfied are you with your life overall? Or, you know, how well is your life going overall? And there's kind of like a lot of variants, these questions. There's also like, the ladder question, which is, you know if you imagine ten being the best possible life, and one being the worst possible life, like where do you fall? Right? So there are a lot of different variations on this. And I think what some people have argued, is that well, are we really measuring the right thing if we just asked people to rate themselves on this kind of one to ten scale?

MICHAEL: Yeah. So this is a point of departure for lots of people, you know, you tell them that you can measure people's feelings, they just look at you strangely, as if this is just not possible. But yeah, so I looked at this quite deeply into my Ph.D. a few years ago. And the problem is you have this objective thing we're trying to capture. So someone's happiness, or their life satisfaction or their sense of meaning in life. And like, obviously, this isn't an objective thing. So you can't go out and like probably, you can't be really sure, you have to check that your measures are working in a different way, that background theory and philosophy of science is called Construct Validation. Basic idea is to test if your measure of happiness is working, you need to go out there in the world and see how it responds and see if it responds in a believable way. So you might expect that if your measure of happiness was exceeding in measuring happiness, then it'd be associated with more smiling, with greater income, with a better state of health, with being in a relationship, with living near water, with good air quality, will be negatively associated things like suicide, people leaving their jobs and their relationships. And it turns out when you check to see how life satisfaction and happiness scores behave actually, they do behave in basically the way you would expect them. And that's what gives you confidence that we think we're measuring happiness. And in fact, we probably are just by asking people how things are going.

SPENCER: Right. And so I think you want to argue that if we look at the things we'd expect these measures were correlated with, they indeed are correlated with these things. How much confidence can this give us that it's actually measuring the right thing? You know, by saying it has the right correlations?

MICHAEL: Yeah. I mean, ultimately, there isn't a sort of a single mechanical test you can do to assess whether you're measuring something is valid, which is to say, that actually captures what it's trying to capture. Ultimately, it's kind of an evaluation. And, you know, this probably familiar idea to your listeners to think of this is sort of a Bayesian way is the what your kind of view of how you think, what you think in the real world is associated with happiness. But some, like some resources, central core predictions, you think that people who are less happy are going to commit suicide or more likely to commit suicide and people who are richer, if you think you're going to be happier, so you've got some really basic things, like, look, this is definitely how happiness works. And then you go out and you've got measures and stuff, and you see what your associations are. And then if the measure gets the core stuff, right, you think, okay, like, we're probably buying this, let's go and test things that we're not so sure about. So, you know, maybe you wouldn't have a view the relationship between, like, it's one of the most famous, but for many controversial findings in happiness research is the so-called Easterlin Paradox, which is that over the course of time, it seems like countries or at least the rich countries aren't, on average, getting happier and more satisfied. But at any one point, richer people are happier than poorer people. And richer countries are happier than poorer countries. There's this tension here. And you might think, well, obviously, as the world is getting more economically developed, like things must be getting better, you know, before iPhones, their teams, and we've got Wi-Fi, and we've got airplanes like of course, society is getting better and better. So this must be false. This can't be the case that these happiness measures are working. And so maybe you want to throw out your happiness measure if it's not giving you this prediction that you think makes sense. But if you already looked at is kind of the big sweep of giving you sensible results everywhere else, then you get this new controversial result. And they go, maybe we learn something, rather than we're just gonna assume that I'll mess it didn't work, after all.

SPENCER: How much does it matter which measure we use? Right? Like if we vary the wording a little bit, you know, switching from life satisfaction measure to let's say, a ladder questions. Or if we ask people, how meaningful their life is versus how happy they are or how satisfied they are? Do we tend to get the same answers, and it doesn't really depend that much on the wording.

MICHAEL: We don't know nearly as much about this as we would like, or certainly, I would like. So if you look kind of broadly speaking, the same sort of thing, that happiness as a fact, life satisfaction, happiness, like how you feel at the moment, life satisfaction is the judgment of life overall, these are different things, the narrative self for the experience itself.

SPENCER: Right. And happiness might be measured by pinging people at random points and saying, How do you feel right now?

MICHAEL: Different ways to measure happiness? So you might ask you like exactly this sort of pinging method called Experience Sampling Method, you might ask someone to break their yesterday down into a series of scenes, like a set of scenes from a movie, and write all of those for Day Reconstruction Method, you might ask people, you know, how happy did you feel yesterday, so that these kinds of variety, but like, broadly, the same sorts of things increase happiness as life satisfaction, but some things are more important for one than the other. So mental health seems to have a bigger impact on happiness, and life satisfaction and income seem to have a bigger impact on life satisfaction than health. So these sorts of things. So if you're putting on your EA hat, thinking like, is this gonna make a difference? The answer is, well, what like, we're not quite sure. If you're doing of fast, effective altruist thing you want to work out what your priorities are. So it's literally kind of a question I've been wondering about for a while because we get these differences like, actually, is our ranking of what we think is overall, the most important thing to do? Is it going to turn on which measure my thought, like, maybe, maybe not, we have really excellent information on this?. So it's something of an open question and something we want to keep in mind. But what I'm going to be sort of getting on to, hopefully, later when we talk about them, is because new research we've done comparing the mental health interventions like therapy as a cash transfer, we actually don't find the measure we use makes very much difference. And actually, that's was a surprise to me. It remains open as to exactly what we're going to get.

SPENCER: Right. Just stepping back for a moment. I think what you're saying is that if our goal is to make the world happier, meaning both kind sof alleviating suffering and making people feel better, then we should actually try to figure out how much interventions change happiness, not just use these proxy measures like health, or income, or things like this, but actually try to go to the core of the happiness itself. And you know, this is a fraud exercise. It's difficult to do. But I think you're arguing that it turns out just asking people how they feel it's a pretty good measure. It's not perfect. There are other ways to do it. But it seems like it tends to come to similar conclusions, regardless of the kind of which method you use to do it.

MICHAEL: Yeah. So I basically agree with most of that, yes, just sort of, it's a bit puzzling that I mean, you don't have to think that happiness is the only thing which matters. But you probably do think that either happiness or life satisfaction, probably a capturing like a really a big part of what does matter, even if they're not the whole story. And I think it was probably the case that those efforts to try and do cost-effectiveness in terms of objective well-being are pretty new, you probably couldn't have done them, you know, more than ten years ago, there probably wasn't like actually enough information out there. But now there is so you might have these reservations about the measurement piece, like, do we really succeed in measuring happiness? Well, seems like we do, the measure seems to check out. So like, let's kind of go-go and do it and see what we find. And just to kind of like press on the importance of this, the way that effective altruism and others have tended to do cost-effectiveness is just you look at either health, sort of standard things like QALYs and DALYs, and these basic health metrics, or you look at income. Neither of these is what ultimately matters. And they're also like, it's not obvious how you compare them, though, you know, how many years have doubled income is, as good as, a year of healthy life. In the end, this is your trial, this is an important decision. Unless you've got this principled way of comparing these things. And the same sort of unit, it's arbitrary. We want to try and avoid arbitrariness and it's an extremely important question, we're trying to figure out how to do the most good we want we've got a sensible coherent approach rather than basically just to be guessing.

SPENCER: Well, I think another way to frame this is that every measure we have, how much good we're doing is a proxy, in a sense. None of them are measuring the thing we actually care about. So whether you're using QALYs and DALYs, or health or income, or a celebrating life satisfaction scale, they're all proxies, right? So then the question is just which is the best proxy for what we care about? I think you'd argue that asking people how happy they are is actually a closer proxy than some of these other things. But I'd like to dig into QALYs and DALYs a little bit. Can you tell us how are those measured? And what do you see as the flaws of those?

MICHAEL: Yeah, so the QALYs and DALYs, these standardized health metrics used by the NHS in England, used by the global burden of disease, evaluate how much health is lost. And what happens in both cases, is you get some members of the public, and you tell them about different states of health. And then you ask them to tell you through, basically things like time trade-off, where you see how bad you think things are. So if you could have five years off, let me think of an example. So what you might do is, you might say, Look, you can have 10 years with blindness, or you can have some other number of years of a healthy life, like for you what would be the equivalent number of years of a healthy life, that's as good as having blindness for 10 years. And so it's by getting people to engage in these comparisons, that you can then put units on different states of health, you know, you got some sort of blind, there's some score for depression, some sort of all these other things. And then you know, you can pull out of your statistical hat, this idea of this combined measure of how good or bad something is, with time. So you have this sort of quality-adjusted life here. So you know, one QALY is one year of healthy life. And then hey, presto, you've got this kind of shared unit of healthiness. And if you know how much it costs you to do some various, you know, get out there in the field and do some health interventions, handout pills, provide surgeries, whatever, then you've got this kind of way of comparing the cost-effectiveness of different things in terms of health. And that's pretty good. That's a big step forward. But there are obviously two problems lurking with this. So one is that even if you think this is totally accurate, as a measure of health, as health captures well-being, we care about things besides health. But then you have this question about, well, how good is one quality-adjusted life here, compared to some improvement, I could make someone's income, or I could enhance their education or something else? Oh, you know, how good is one quality-adjusted life year compared to reducing crime in an area and at this point, you don't look, sort of, run out of rope, you have to do something else to work out how you're going to trade those things off, and, and you want to have some sort of sensible way of doing it. The second concern that you're gonna have about all of these standardized health metrics is that what people are doing is that are guessing how bad a certain condition is? So asked him how much? No, you're sort of sitting here in a comfortable armchair. And this person's giving you the survey and you're wondering, like, really, you know, how bad is it to be blind or to be depressed, and people don't have good access to this information. And so the other kind of motivating piece is that there can be a difference between what we expect affects other people's happiness, and then what really does affect our lives as we live them. And so because there are these various differences, psychologists call them failures of affective forecasting affects the New York emotional state, there are biases, psychologists identify between what we expect to affect our lives and what does. So there's going to be this difference that actually, people might not be getting the right sort of answers about how bad these different health states are. And that motivates trying to actually, you know, find out how people with different health states feel like ask them at the time as they have them, and use that information rather than use guesses to other people.

SPENCER: Yeah, it seems really tricky to ask just regular people, you know, how much is one year of healthy life worth in terms of years of buying life or years of disabled life and with different disabilities, etc. I don't understand how people would know that unless they have experienced those states. Are their attempts to do this with people who, let's say, had gone blind, and then they actually have the ability to compare, and then you can ask them?

MICHAEL: Yeah, so I'm going to confess at this point, I'm not a QALY historian, there have been different methods have been done to sometimes with people who have the condition, sometimes you ask doctors about how bad they say these things are? And yet you find differences. We know how bad the difference being like if you ask depressed people, how bad to have depression, and the other people, how bad is it (depression), and then people who don't know about it, I mean, in general, people who don't know about conditions think they're there to see these differences between when you ask the individual self. And then when you ask other people to imagine. If you really want to be finding out how some health condition affects someone's life as they live it, you actually don't want to necessarily draw their attention to it. So if I asked you, you know, you're depressed, how bad is your depression, then like you're really thinking about it. So in fact, if you want to know what the kind of the true effects are, tease it out through sort of sneaky randomized control trials and sort of statistical inferences. So you just ask people how their lives are going. And then you work out how much these things really affect their lives, rather than how much they imagine they affect their lives when you think about them.

SPENCER: And so what's the difference between QALYs and DALYs? They're like, pretty similar, but to some sort of subtleties, is that right?

MICHAEL: Yeah. So one of them is sort of positive measure. Actually a DALY is like a reverse QALY. So QALY is a year of healthy life, but a DALY IS a year of unhealthy life. There are some differences. When you ask people about QALYs, you give them sort of different health states. So you might talk about like, Okay, this is your mobility with this condition. This is your level of pain. This is your level of mental anxiety. And then you ask them to do the trade-off. And with the DALYs, you just ask people how bad is this condition. And you give them a description of the condition, but for the purposes of trying to accurately measure well-being, the challenge is that you have to ask people to imagine these things or report on them as they happen. And really, you want to find out from the while, how much these things really affect people's lives as they live there, rather than how much they expect they would.

SPENCER: So how are people using these measures now – QALYs and DALYs? And then I'm interested to know what would you like to see them doing with his life satisfaction measure instead, like what happens when you replace the QALYs and DALYs measures with these other measures?

MICHAEL: So this context is talking about, one is the effective altruism world. The second is the kind of the broader health world. So in the effective altruism world, people started when sort of effective altruism was kind of getting rolling, almost 10 years ago, started by using QALYs and DALYs, and have now largely moved away from it. So peopl have all used them for a while and then stopped and has since then, I've replaced them with a different method, which I think is problematic and it's kind of its own way. What people have been doing for a number of years is they have sort of two main outcomes they look. The one is doubling someone's consumption for a year. And then the other one is saving the life of an under-five-year-old child and they've got this health one saving years of life and then there's this economic one, and then from those years, work out how to kind of compare those, how to put those in the same unit. So you can compare improving life and saving lives, they pulled their staff members. So, you know, how many years of double consumption do you think are as valuable as saving the life under-five-year-old child? Is it ten years or double consumption or a hundred years or five years? And for a number of years that the figure given that we're using is basically around hundred to one. The challenge with this is that you've got these few things, saving the life of a child and then increasing this measure of wealth. But then, you know, how do you compare this to other sorts of things? So let's say you're looking at reducing pay or alleviating depression, well, surely you don't want to measure the value of these things in terms of the impact they have on people's income, like, what's the value of stopping someone from being, you know, the value is not just the extent to which they can go and they can get out back to back to work and earn more like value is actually improving their well-being. And that's like, not captured by the economic peace. So you still need this, although get moved away from these kinds of conventional health measures, they're still the same challenge of need like you want a sensible way to compare all these different sorts of things.

SPENCER: Do you know why they moved away from QALYs and similar purchases?

MICHAEL: Yeah. My understanding is that they use DALYs for a while. But then they found there were some methodological errors and how they were constructed and talked to the academics. And they found out that the basic, although actually, these measures aren't that robust. And like, I'm mind reading a bit here. I think it's that in the end, they realized that they were ultimately just focusing on saving lives, and then improving economic prosperity. So actually, the focus on other sorts of improving quality of life, as measured through health didn't end up being so wasn't so so important. And actually, I'm not sure what method GiveWell do we compare comparing improvements in health. Well, people are alive to their improvements in income. So I mean, I know I would like to know the answer to that question. I'm actually not sure what their internal machinery is.

SPENCER: Okay, what about in the traditional health world? How are methods like QALYs used now?

MICHAEL: Yeah, so one flagship thing is the global burden of disease. And when you get all these different problems, and you try and work out, you know, how much health is lost from them. And really, what I would like to do there is I would like to see something like a global burden of happiness. I'm interested in health, because to the extent that it impacts happiness, this is basically never going to happen because of internal institutional pressures. But I would just rather that we measure the impact of health states on happiness, or maybe life satisfaction rather than QALYs and DALYs, and then we can then compare improvements in health to other sorts of policy outcomes, like reducing crime or reducing unemployment, that sort of thing.


SPENCER: Right. So this gives us a few different ways to try to decide what interventions to support to try to improve the world, right? You can use methods like QALYs to try to say, you know, what's the best intervention is going to improve them, QALYs is the most you can use methods like GiveWell is using to say, okay, which interventions are going to save the most five-year-old lives per dollar or double consumption the most per dollar of investment? Or you can use these kinds of new approaches you're advocating, which are using measures like life satisfaction. So how do our results change? If we use measures like life satisfaction, what actually points us to these other measures and don't?

MICHAEL: I'm glad you asked that question. I mean, that's really the point of the enterprise to find out. So yeah, as I said, efforts to think what is our global priorities, if we put on a subjective well-being hat, we look at self-reported happiness and satisfaction, people haven't really tried to do this. So attempts to do cost-effectiveness in terms of subjective well-being are basically new for the last ten or so years. So people are now are trying to get this going in policy-making, there are a few attempts to cost up different policies. And the UK is a sort of a world leader in this, even though like nothing is yet actually happening in government, but people are trying to work out how it would be done and get the ball rolling. And the reason I set up the Happier Lives Institute was that there's like, no one in the effective altruist world was doing this. And so I thought, well, I should, you know, probably have a go, if I don't push this looks like it probably won't happen. So yeah, attempts to do this are like on you. And like I said, there are these differences between what we expect improves our happiness, and then what actually does these differences in effective forecasting. So what this motivates is like having a look and seeing if it turns out that there really is new stuff. And so where the effective forecasting research indicates that when we think about other people's happiness or future happiness, we don't take duration into account. So we don't really account for the fact that we get used to some things and not to others. So one of the things which come out of that research is that health conditions turn out to be probably not as bad as we can be expected they are. So if you know, if you walk with a limb, then like you get used to it, you withdraw your attention, you stop thinking about it. Whereas things like, you know, mental anxiety and physical pain, you just like there's obviously keep being a problem. There's aren't things that go away. And so whether that prompt is the thing well, like these aren't the kind of things which effective altruism to have talked about before. So these are the sort of things that do stick out as being a real drag on happiness. So why don't we take a look and see if these things do look like priorities when we think about things a different way?

SPENCER: Right. So even though it's early days in your research, and tracing out all the implications of this way of doing things is I could lead to, you know, a lot of different conclusions, at least one seems, tentatively to be the case, is that mental health might be underestimated by these other measures.

MICHAEL: Yeah. So I mean, it might be the case that we're having this conversation again in a few years' time. And basically, it turns out that like nothing has, in fact, today, when we use this different and better method for thinking about having an impact, but like, you know, it's certainly worth like you're trying to see if we can dig some goal and try and find things and take a look. I am at the project in which I'm engaged in (inaudible), when you think of things in terms of happiness, you do, in fact, use these, this new method, like, do we find some new stuff? Yeah, that's the see, when we get to.

SPENCER: I think about the idea that different health problems could have vastly different implications for how happy someone is, right? So for example, someone could have a chronic fatigue condition that doesn't seem very serious, but actually literally, like, not able to enjoy anything, right? And then like, she could be devastating. Whereas someone else might have a more serious-sounding condition, but it's well managed, and actually doesn't like affect their happiness very much. And so once you kind of look at the psychological aspect of how much is this actually affecting the moment to moment lived experience that can be seen, I'm surprised inclusions on that.

MICHAEL: One of the kind of components of how QALYs are constructed, is looking at how whether people are able to engage in usual activities. And actually, this turns out whether you're able to engage in usual activities, and also effects on your mobility have like quite limited impact when you assess the health conditions in terms of happiness. But actually, these do quite a lot of the work in giving us these different QALYs for. When you think in terms of subjective well-being you want to think about, okay, which things are actually continuing to make us feel bad and aren't going to go away, and other things we can do about those. And when we look at those things, do those actually look to be more cost-effective than our current (inaudible), that seems to be the place to go?

SPENCER: Yeah, it's interesting to think about the category of bad things in our life that we don't get used to, versus those that we do get used to, because there are so many things that we can get used to over time, and they no longer bother us very much. You know, maybe we maybe don't, we don't fully adapt, but you know, we feel significantly better about them once we get used to them. And then there are other things there just because equals suffering all the time. And, you know, these basic, unpleasant mental states seem to be one example, as you pointed out, as anxiety feels like anxiety, right? Maybe you can get better at the kind of living with it. You can get better at, say, doing normal activities with anxiety, but it still kind of feels like anxiety, or depression would be another example. It just sort of like intrinsically feels bad, and you can't really adapt to it.

MICHAEL: Yeah, you're right.

SPENCER: Alright, so tell us about some of your attempts to actually quantify the effectiveness of different interventions with this new way of looking at it.

MICHAEL: Right. So when I got thinking about this, during my Ph.D. thesis, I thought, well, okay, maybe we need to go back to the drawing board on what our global priorities are, and I don't really talk about it so much anymore. But there was this kind of idea that was part of effective altruism. Likewise, this was years ago that if you want to work out how to do that most good, you need to do this thing called cause prioritization. What do you assess problems in terms of scale, neglectedness, tractability? And this is some sort of mechanism you can use prior to actually digging in and assessing things that are cost-effective. So you kind of look at things from kind of a bird's eye view, and then you dig into it. And I spend about nine fairly unhappy months, during my Ph.D., like trying to work out how to do this, maybe it wasn't as long as nine months. But in fact, like, I tried to apply this method, it just didn't make any sense. Like, what, what do you mean, you can assess like poverty, like, you know, in broad terms as a whole, with these supposedly juristic? Without looking at the individual things you could do? A dreamt up this kind of different approach, but you know, how can we kind of categorize different sorts of problems and relate them? But yeah, in the end, it was pretty hopeless. And I thought, Well, exactly the way to make progress. It just takes some things, the matches match what actions you can take, like giving money to a charity, and then try and work out how cos- effective they are, and try and put these in units. And one of the things I did was I did this sort of back of the load calculation, looking at Strong Minds, which is a mental health charity, which provides inter-personal group therapy to women and children in Sub-Saharan Africa. And then I do this in happiness unit, I've got, I've got some different bits of information. And then compare this to the kind of GiveWell top fave, the cash transfer,s and deworming. And when I ran head to head, or actually, these mental health interventions, which have so far been overlooked, looked pretty compelling. And at least they were kind of in the same ballpark, it wasn't just totally obvious, they were way less effective, and we had kind of really good reasons for overlooking them. And so it was kind of having done this initial analysis thinking like, Oh, this is promising, like, maybe we found a little bit of we were sort of panning for gold and like, maybe find a little bit okay, like, that's what I think this motivates us to go and do a lot more seriously and see if there's real stuff, it was this, this line of thinking which motivated through research, which now I and my team have been doing at Happier Lives.

SPENCER: So going back for a second, I was really interested in what you said about this approach of looking at a scale, neglectedness and tractability, for cause prioritization because this is something you hear about so much, right? People say, oh, scale, you know, you want it to be potentially really large interventions. So you know, if it works, you want to build up a really large room for people. Neglectedness, well, you know, if a lot of people are already doing a thing, it's probably not as effective. Do you also start doing it, because maybe they've already got all the low-hanging fruit? Do you want to kind of neglected intervention? And then finally, tractability? You know, there are some things that would be great, but we just don't know how to do them. You know, we just don't have the technological capacity or the know-how, and if nobody knows how to do it, there's no point in working on it. Yeah, it'd be great to cure all cancers with, you know, one treatment, but nobody seems to be that close to doing that, right? So this framework seems to guide a lot of people's thinking on this, but it sounds like you really struggle to apply. So I just want to hear more about that. Like, do you now think that this is just not a useful framework? Or you know, I want to hear your critiques in more detail.

MICHAEL: I think this approach is confusing, and we should retire it. So to think about doing the most good, like one thing we do understand, and it's not at all controversial is expected value risk. We're trying to do the thing, which is in which expectation has the most value. Like that's how we think about doing the most good. Now we've got this other thing, which is to assess causes in terms of scale, neglectedness and tractability. Okay, now, this is one thing we definitely understand. Okay, expected value, you put like units and values and costs on things. And then you could draw some conclusions. So what's the relationship between these things like the scale, neglectedness, and tractability? If it doesn't turn out, we're giving you answers in terms of expected value, like, obviously, you're doing it wrong. There's something mysterious, and there's sort of what this idea is, what's going on the neglectedness and tractability is supposed to be how it applied, like changed a bit over the course of time. So in sort of circa 2050, this would, I say, when the Will MacAskill's Doing Good Best Ever, there's supposed to be sort of heuristics. The idea is that, like, look, if you know nothing about a problem, other than how big it gives you some sign that it's like, yeah, there's something important there, or if you know nothing about a problem other than how practical it is, then, well, at this point, it starts to become a bit unclear, like, what do you mean by tractable? Like if you don't mean cost-effectiveness, then like, what do you mean? And yeah, and then you know, and then we've neglectedness, so maybe the fact that lots of people are doing something is a really good reason to go and get stuck it so maybe lots of people concerned about climate change has loads of resources going so maybe you can like to nudge it in the right direction, you can have a small impact on lots of resources. Maybe the fact that something is neglected means that no one else has been paying attention. Like the point is, it depends like none of these things are actually useful by themselves, you can think of them together. But when you think of them together, like all you're really doing is taking some action like you've made to a charity, take some career, do some piece of advocacy. And then in fact, you're actually just like implicitly comparing the cost-effectiveness of basically the different sorts of things you can actually do. So this sort of idea that you think you can assess how cost-effective it is to get stuck into some problem, as distinct from how cost-effective it is to like, take any action that solves that problem is just sort of confused, like the only sensible way to compare causes is to compare like particular things you can do to make progress on. And that just, and at this point, this idea that we had this sort of initial, you know, first cup method, these kinds of neglectedness and tractability, which is different from and prior to looking at the expected value of certain actions, this just sort of starts to disintegrate. But there really were these two different things.

SPENCER: That's really interesting. Now that you say that, it makes me think about the idea that I actually don't apply this approach of scale, tractability and neglectedness. However, upon reflection, I think there's a more personal version of something like this that I apply. And to me, it makes a lot more sense to apply it personally. So what I mean by that, let's say you're thinking about doing a project, right, you want to know that you have some competitive edge at doing that project, right? Like, if it's completely outside of your own skillset, then it doesn't necessarily make sense for you to work on. So there's a form of tractability, which is sort of like tractability, given what you're good at your skillset your competitive advantages, right? On the neglectedness front, you know, it doesn't seem that important, to me whether a broad category is neglected, what matters is that if you're thinking of doing a particular project, you don't want there to already exist other projects that are doing what you're trying to do, as well as you think you could do it, right? And needs to be neglected in the sense that you want to think you, you can add unique value to the world that's not already being done on that kind of specific project you're planning on implementing. And then finally, on the scale front, the way I think about that, as ideally, you want it to be the case that if your project succeeds, it will matter. So there's at least scale in terms of, you know, if everything goes really well, this will be something we're having done, you know, there'll be some kind of the amount of value out of the world will be, you know, non-negligible. And so that kind of personal version of this, where you're thinking about your own abilities, is tractable with regarding to your own skills, is it neglected in the sense that somebody's already done this project better than I can do it? And, you know, is it worth doing? So is there enough scale there to have it be justified even working on this project? That's how I think about it.

MICHAEL: Yeah, so I basically don't really disagree with that. I mean, maybe a helpful analogy is to imagine you're trying to work out the volume of something. And you've got like width, height, and depth. And if someone says to you, okay, well, we really want to work out, like, which object has the most volume, so we're going to use height as heuristics. So generally, like, the taller something is, the greater volume it's gonna have, okay, that's fine. But like, actually, you don't know about the volume of your object continuing for all of the three things together. But that's the kind of the message rather than thinking that goes like this tractability, like different independent heuristics, they just like three bits of information, you have to combine until you stop that any once you start the three of them together, then you have at least implicitly thought like, here's an action I can take. And like, here's how good it's going to be. So like, now you've got your width, depth, and height. So you've got this, like, you know, what the volume of your object is, you've guessed the volume of your object. So the point is that if someone said to you like, okay, we're trying to find this, you know, the biggest shape, and we're going to use height as a heuristic. He said, well, you can't just do that you actually just use it? Yes. Just want to look at now with height and depth, and then you've got it. If you thought that this cause privatization approach was something you did like, before you're assessing the effectiveness, then you weren't like if you thought we sort of in the abstract going to judge the scale, neglectendess and tractability of poverty without looking at like any particular thing we can do to try and alleviate poverty, then like, how are you going to do that? But then if you have, you know, you've got some particular thing in mind, and you know, you thought about that, then you've just done the work of finding out the expected value of something if you've got your width, depth, and height and just have the volume.

SPENCER: Right. And I think once you've honed in on a very specific project, this makes a lot more sense, talked about it. And I think part of what you're saying is that if you're just doing a very broad intervention, you can't really evaluate these things, right? It's sort of like you can't really say what is the volume of containers or what containers, right? But once you've honed in on a really specific project, tractability has to do with your ability to execute this thing successfully, which is related to cost-effectiveness. Neglectedness has to do with, are you going to actually add a causal contribution or other people already going to do the thing anyway, so you're not really going to add any value because already being done. And scale is sort of about the total size of the opportunity, right? So can you do it at a large enough scale that at the total benefit, if you succeed as large? Then those are the three things that start to add up to just sort of like how much good you're doing in a sense, right?

MICHAEL: So one way to kind of put pressure on it is to imagine someone comes up and says, Okay, we're trying to work out what's good. Let's say we're trying to give our money away. we've assessed all these different options, and we've gotten their cost-effectiveness numbers. And then you turn around and say, Oh, well, but they don't we really need to work out their scale, neglectedness, and tractability. You say, no, like, we've got the information we need, like, what's this other thing we're trying to do like in the end, we're just like trying to get, like, looking at actions we take and see, and expectation how valuable they are. And my pitch is that rather than people trying to like tangle themselves up about exactly what the scale, neglectenesss and tractability framework is, and how to apply it, it's just like, go and look at some stuff, and then try and you know, try and estimate the impact of those things. And I think it's telling that this idea of the scale, neglectedness and tractability as the way you try and work out how to how to do good in the world has been floating around in the ether. And it's there's been all these sorts of conversations about what it is, and it's still basically pretty unclear. So it's like, it can't be that helpful as a tool if people don't really know what it is and how to use it. And maybe one thing I'll add is that at one point, as a few years ago, there was maybe 2018. I think like Owen Cotton-Barrett suggested this way of formalizing it and then neglectedness and tractability as he follows it just in the end, ends up being cost-effectiveness. So then there's clearly no different is that you're looking at, like, what's the best thing you can do and religious person problem, okay, and then you just estimate cost-effectiveness. And then, at this point, it's just kind of clear, there's no different you can't kind of the abstract, assess the world's problems without looking at solutions you can that are available for them.

SPENCER: Okay, so you have this approach for evaluating how to do well based on subjective well-being, can you tell us about what are some of the specific analyses you've done so far? And where do they start to point in terms of how to do more good?

MICHAEL: The work I'd done, myself basically motivated, like doing this whole thing, like much more seriously, and apparently, I'm a philosopher, not an empirical researcher. So the question is, like, what's the where should we start? And I thought the place I thought was by looking at a mental health intervention, psychotherapy, in a low-income country context, compare that to the classic effective altruism favorites, as popularized by GiveWell, effective altruism world hadn't really done and have a serious look at mental health in terms of subjective well-being, and then to run this against the current leading candidates, and then to see if it made a difference. And so I thought, like, that was kind of the first apples-to-apples comparison we should do. That's now what we've done.

SPENCER: Cool. Yeah. Well, do you want to walk us through how you approached that and what your findings were?

MICHAEL: So like, I was saying we could have a really thorough comparison of psychotherapy, to cash transfers, both implemented in low-income context, in terms of better well-being, and also some measures of mental health. So this was kind of whenever a good answer to the question I've been itching to get a serious answer for a number of years. So we do a meta-analysis for each of these cash transfers. This meta-analysis was joint work between Happier Lives Institute researcher John McGuire, with two collaborators from Oxford, Caspar Kaiser, and Andreas Mortensen. And so these are the guys that have done the work and I just have kind of the privilege of telling you about it. So this meta-analysis has actually become an academic paper is now forthcoming in Nature Human Behavior. (inaudible), why did a separate meta-analysis of psychotherapy, we looked at these kinds of justice, broad interventions? And then we also looked at particular charities that seem like the best and fast way of implementing and what we find is that psychotherapy is about 12 times more cost-effective than monthly cash transfers to this giving people about $20 to $40. And when we look at the best in class charities, we find that Strong Minds which does interpersonal group therapy, for women and children in Sub-Saharan Africa, it's about 12 times more cost-effective than Give Directly give her a favor by sort of $1,000 lump sum against people in a low-income country context, which is equivalent to about a year's household income. So I'm really excited about this because what we've found is that this mental health, charity, Strong Minds, as well as, this category of interventions like psychotherapy, it's basically up there with the intervention charities, which better address already think are the most effective thing. So on the GiveWell's owns internal estimates, they think that the most effective life-improving interventions are these deworming ones that don't have intestinal worms that they're treated and then they earn more or less life and sort of buy on GiveWell own scheme, which is done in terms of improvements in income, these deworming interventions are about 10 to 20 times better than cash transfers. And we find that Strong Minds is about four times better than then cash transfers. So basically, it's in the same area, which is pretty big, because we hadn't been taken seriously as an option.

SPENCER: So that's a really cool finding, I just want to push on something a little bit, which is in my experience, looking into these things in detail, you realize there's just massive uncertainty. And so I'm wondering, does that apply here too? Like, are we really talking about some huge confidence interval? And so you know, when you say 12 times better than psychotherapy in these kinds of group psychotherapy applications in Africa, that they're taught on certain cash transfers, I assume we should think of that as like with very high uncertainty bars on that?

MICHAEL: Yeah. So there is some uncertainty over these, but we try to account for these. So made me sort of a place that we think there are three searches important in a novel in three main ways. The first way is that this is the first attempt to do this comparison using effective well-being. And in terms of some sort of some subjective units, rather than just helpful for wealth. And so there have been meta-analyses of therapy in terms of mental health, but not of cash transfer, so you know, but so new for cash transfers, it's new for comparison. The other thing that we do is we were estimating the like the total effects of the intervention, which isn't something that meta-analyses do. So not just for this initial facts, the facts over time, and then we use the evidence base for the cost, as well. And because we're able to, like in broad terms, look at the different types of interventions. And so we've got our 40 studies for each type of intervention, the population of about 100,000 people, cash transfers, 30,000, like therapy, so we're able to kind of look at these different interventions in the round and compare them. And this allows us to be much more confident about really what's going to go on to make better predictions. And so it's not just like we're sort of relying on one study, and the one study of some intervention is carrying the weight for how much good we think we're doing. And then the third piece is that rather than just using point estimates, the different sorts of things, we use Monte Carlo simulations to account for uncertainty. So we take the confidence intervals, and then we run these, and then the Monte Carlo simulation like sample these and so what you end up doing is that you can be much more confident about where you get to in the end, rather than thinking that, you know, if you put one number is too high in the wrong cell, then you're gonna end up being skewed was supposed to change the way. We think we're able to feel better to do more. But to account for uncertainty in a more sophisticated way, then, people have tended to be so far.

SPENCER: So the Monte Carlo simulation idea, am I correct in saying that what you're doing basically is saying, Okay, our final estimate depends on various factors, each of those factors, has some uncertainty, so we want to actually model the uncertainty in each of those factors. And then when those factors get added together, multiplied together, however, they combined into a final estimate, we're able to take into account how the uncertainty in each of the factors affects the uncertainty in the final output.

MICHAEL: That's exactly right.

SPENCER: We have to find those kinds of confidence intervals or whatever you produce they're trying to see.

MICHAEL: Okay, yeah. So what we find for the intervention service is kind of proto studies on this clarity. This is like a broad set of studies and like therapy programs, or cash transfers. And so we central thing we find is that in terms of these units of subjective well-being, it's 12 times more cost-effective psychotherapy than these cash transfers. And the confidence is about three times better, and then 40 times better.

SPENCER: Okay, so that's not actually as massive as I sort of expected in terms of the confidence interval. Of course, with any of these analyses, there's always like, the uncertainty that's captured in the model. And then there's kind of model uncertainty like that the model itself is flawed and that way. So you know, my view on these things is you always have to kind of broaden the confidence interval more than you think. But generally speaking, it's very hard to quantify that. So sort of just a meta-consideration on these things. So can you tell us a bit about how did you actually perform this analysis?

MICHAEL: Yeah, so I've given you the headlines. Oh, but haven't asked to explain what we did. So for each intervention, we looked at about 40 studies from low-income, country contexts. And for the cash transfers where we did a systematic search. So we're pretty confident this, like exhaustive, the literature, and we looked at lump sum, cash transfers, as well as monthly stream cash transfers. And then for psychotherapy, we did something a bit similar. So getting we found about 40 studies. We weren't exhaustive here. We didn't have a systematic process, but we basically thought we sort of hit diminishing marginal returns and so we think that's pretty representative. Yeah, so the psychotherapy, we looked at studies that were either delivered by non-specialists or were delivered in groups. And the reason we did that is that in low-income countries country context, they're actually not very many mental health specialists. But turns out that this may be such a problem, as the research shows that task shifting, which is when you get non-specialists that kind of lay help work, rather than highly trained a specialist of the liver, like therapy, it basically turns out to be like, as effective. And actually, what we did find was that the specialists were about 15% more effective than the non-specialist. But so they have this, what's the difference isn't huge. And then we looked at, we looked at groups, rather than individuals. And actually, what we found is that the groups were more effective than the individuals, which is a bit surprising, but that seems to fit with the rest of the literature. And kind of a further thing for us to try and think about is, you know, so it's kind of the high-income countries, we tend to deliver therapy individually, but it seems to be more effective delivery groups, at least in low-income country contracts. So we're not really sure what's going on with that. But anyway, that's kind of enough that it seems like the most promising version of the intervention is going to be late health workers and groups, and that's gonna be the most cost-effective version if we're thinking about having the impact.

SPENCER: Presumably, doing these group methods and doing it with people who have less training is going to make it much cheaper to implement.

MICHAEL: Yeah, that's the idea. So even if you're kind of a strong believer that specialists individual therapy is going to be more effective per person. It's pretty plausible that it's gonna be much more cost-effective to do the non-specialist group stuff. Yeah. So the point is, we've got we've all these different sorts of studies. So this gives us this, like, in both cases, this gives us this wide evidence base. And we can use this to estimate effects and the cause. So we have this information on the effect over time, and we see how that decays over time. And that's what we do to get the total effect, you know, kind of nothing big kind of magic there. For the cash transfers. We had outcome information on measures of happiness and life satisfaction, and standardized measures of mental health. But things like the depression questionnaire for psychotherapy, we actually didn't have direct subjective well-being measures, we didn't have happiness and life satisfaction, we did have standardized mental health questions. And so we were one of the things we were a bit worried about was look, if you look if you measure this different way, even get different results. And what we found was that in terms of the sort of the standard deviation improvement, like how effective these interventions were, for the cash transfers, where we had different information, we found that like, it didn't really make much difference, whether you were measuring it in terms of happiness, or life satisfaction, or standardized mental health, if you did in terms of life satisfaction, it was maybe like 20%, more impactful than the other two. But basically, it was kind of awash. So we didn't find enormous differences. And that was interesting and surprising. So what was meant is that we could just combine our results with these different things. So we can have a more kind of statistical power to do the analysis. So in the end, we just sort of preferred happiness and life satisfaction and mental health changes. We're just at the kind of comparable to do our analysis now.


SPENCER: So something I think might surprise people about this is when we're talking about cash transfers, we're talking about giving money to extremely poor people. And I think one's intuition would be that if you give money to extremely poor people, this is going to have a very large improvement in their life. Would you say that this result is more about the kind of surprising effectiveness of mental health interventions that they just work so well? Or is it more based on the idea that actually giving people cash doesn't seem to improve their well-being as much as you might think?

MICHAEL: So it actually is quite interesting and surprising what we found, actually, like, how do we get these different cost-effectiveness numbers? We started with these interventions and give this hypothetical intervention of this kind, like what are the effects costs, and then we looked at Strong Minds and give directly those that seemed like the best-in-class versions of those. So for Strong Minds, we find that the initial effect of this therapy is 9.9 standard deviations, okay? So that might not mean anything really by itself. But then in comparison, Give Directly we find that $1,000 cash transfer which is about yearly household income, is 9.25 standard deviations. So the kind of initial effect is about a quarter so we find is that the cash transfers last longer than there, which is a real surprise, it's certainly a surprise to me. So it looks like, we come to the conclusion that the effects of cash transfers probably disappear after about eight years. So therapy, it's about five. And when I came into this, I thought of cash transfers like OLIO, or the fact would have been gone off for a couple of years. But that isn't what we find, it turns out they last much longer. And then the last bit is just the just cost. And this is what we work out. The other big difference comes in so from mines can deliver therapy for about $150 per person, whereas a cash transfer, probably $1,000, cash transfers broadly about $1,100 to deliver. So therapy has the bigger initial effect, but cash transfers last longer. But the therapy is much cheaper. And that's how we end up getting to there, the 12 times bigger.

SPENCER: That's a great summary. So one topic I'm particularly interested in is the use of scalable digital interventions like mental health apps. You know, as you know, we make multiple mental health apps. We make MyIndies, we make UpLift, and I see a lot of promise in this kind of intervention because they can be delivered all over the world at a very low cost. And so I'm wondering, have you investigated this topic?

MICHAEL: No, but it's on the agenda. So let me tell you where I think we go from here. So maybe just the backup a little bit that the story is like, we're trying to figure out how to do the most good. And oh, look at that we can measure the impact we're having in this better way that we had, we can really find out what impacts people's happiness, their life satisfaction, don't just have to use these to these other sorts of things, which we thought it affects the well being but like, weren't actually, you know, good measures of well-being. And then we do some digging, we find out that like, we can do things in terms of subjective well-being, it doesn't make a difference, it does seem to indicate principles or priorities. Okay, so what's next? So I saw, I guess, in terms of our research pipeline, I mostly the applied side, I think of it in terms of micro-, mezzo- and macro- interventions. But I'll just kind of flag some of the things which are kind of think of promising. And in each case, like, things we want to look into all of these we think could be more cost-effective than Strong Mind. If put differently, we think all of these things could be 10 times more cost-effective than a cash transfer. And that's why Yes, look into it. So want to look at a different mental health charity with the friendship bench does something similar to Strong Minds, but it looks they might be able to deliver this much cheaper, we want to look at just as you were flagging mental health app, and this looks really promising. But, you know, it's kind of a bit hard to think about, not least, because some of these can be commercial. So you know, what's the kind of how do you think about the role of the philanthropist there. I see this really sort of proving our case, to the effective altruist world that you can send things baseball bat, and it doesn't make a difference. So it's kind of easiest to start with these apples-to-apples things. So also, we want to look at cataract blindness, fistula surgery, we have some Summer Research Fellows working over the summer, the name indicates that we found them providing families with cement flooring and a sort of dirt floor that looks pretty good. So all these are kind of, you know, apples-to-apples GiveWell micro-interventions, we want to like have a further look at and see the how they check out. And then these sort of mezzo-interventions where we were looking at specific object-level policies. And these set out with enough to be related to drugs one way or the other. So interested in researchers, the use of psychedelics for treating different kinds of mental health looks pretty promising. It's not clear exactly how useful additional money is, and what's the best thing to fund relatedly lots of people don't have access to opiates, in the four bits of the world. And those are pretty cheap to provide. So, you know, like, that's nothing to do there. And then I'm also interested in also on this, this drugs piece on whether changing the laws around recreational drug use would be impactful. And the sort of that the thought there is that there's quite a lot of misery in drug-producing countries, places like Colombia or Afghanistan, which relate to kind of the drug trade. And so I'm just basically curious to see, does that look just look credible as a priority? And certainly, something which is quite reflective? And then we have these kinds of macro-interventions. So what are the global levers, we can be pulling on to try and steer the whole world towards a higher quality of life? And one of the things that stick out here is I've tried to get well-being into public policy, like what's the best ways to do that? How service is gonna make? How can we get companies to make improve employee well-being, like the cloud a core part of what they do? So is there some way of revising, say the environmental, social, and that sort of government, investing priorities? And these sorts of things, actually, oddly enough, might look credible as long term as kinds of intervention, or at least maybe sort of medium-term thinking, okay, like, what can we do to set the world on the right path to caring about people's well-being and taking action on that in a more appropriate way? So yeah, I mean, so basically, we've got those, that's the applied bit of research. There are some other bits of kind of global for his work, which I'm interested in doing. But as you can see, there's basically plenty of plenty more stuff to be done, and we'll be able to look at and hopefully, we will find even better ways to be good.

SPENCER: Awesome, Michael. Well, before we wrap up, I have I want to do a rapid-fire round with you where I just asked you a bunch of quick questions. How does that sound? Alright, so I've heard that the country of Bhutan uses gross national happiness as a metric, what do you think about that?

MICHAEL: Although Bhutan is famously associated, known for it is gross national happiness, but where they're kind of national measure of happiness, is, what they're not doing is just asking people zero to 10. How happy are you? Well, how satisfied are you using that as the only thing, there are some that they've got about the thing as 19 different indicators? And some of them are about people's subjective well-being, but they've also got other sorts of things like community participation, education, health, and so on. So yeah, I mean, they're, they'd like moved away from taking GDP as the sole measure of progress. But they haven't gone wholesale suggests taking the measure of progress to be some single measure of happiness.

SPENCER: Got it. So what would you like the role of happiness measures to be in government? Or would you like to see the government pulling people and measuring their happiness on a scale of one to 10? And then what would you like them to do with that? If so?

MICHAEL: Yeah, I mean, I think we should be using happiness, that's especially the ultimate measure of how we fetch policy.

SPENCER: Okay, next question. What about cross-cultural comparisons of happiness measures, right? So people might say, as long as you're within a fixed population, and you're having people rate how happy they are, it's, you know, maybe it's reasonable to say if those numbers go up, that's a good thing. If they go down, that's a bad thing. But then, you know, if you're comparing people in Brazil, the people in Japan, the people in the United States, the people in Ukraine, can you really say that, you know, what seven means is really comparable across cultures?

MICHAEL: Yeah, so this is a question that people raise a lot can we compare our happiness scale, bit more technically, the problem is known as kind of cardinal comparability. When we go basically take these numbers, face value, feel a pretty switch, especially about it, I actually have done a lot of research into this, I've got a paper on the Happiness Institute website, which deals with this topic, and kind of explains what's going on. And I'll give the kind of that there are quite a few different moving parts in it. But the kind of the broad story, basically, is that we can put numbers on our feelings to say how good or bad we feel. And that's not really that different from just saying how we feel using words, if you say, like, oh, I'm really happy, or like, I'm quite happy, or I'm sad, we're used to using verbal labels that capture our feelings. And basically, we just do the same thing with numbers. And then because we communicate, we understand how to like how to kind of have common meanings for words. That's how we kind of get this, get this thing off the ground, really, like not that much more mysterious than just saying, you know, I feel great, or I feel lousy. Yeah, there's this question about international comparisons of happiness scales that people use the questions in the same way. And the study I find most compelling on this, by John Halliwell and colleagues from 2016. And they looked at immigrants moving from over 100 Different countries to Canada. And they found that regardless of the country of origin, the average levels and distributions of life satisfaction among the immigrants mimic those of Canadians, so their reports converge with each other, and the distinct Canadians. That indicates that, despite different cultural backgrounds, individuals have the same scales each other, and we might worry if this is really the result of immigrants adopting the cultural norms of any country, given that immigrants from different places have different levels of subjective well-being prior to arrival. The only way the pattern of immigrants split will be would mimic that of Canadians would be if the immigrants just fully adopted the cultural norms of their new home. And so if they were only partially adopted it. And the selective offerings wouldn't mimic those of Canadian and it seems unlikely that we would just expect all the immigrants just to arrive and then instantaneously change how they think about their life. Yeah, it looks like there's this kind of consistent scale to use across the world. And there are some other bits of evidence that I can mention, and I discussed in my paper.

SPENCER: So how can we make happiness measures better? You know, what do you see as the future of happiness measurement?

MICHAEL: Rather than making the best, I think we can use the best measures already are as I mentioned, at the start, like the most common way of measuring subjective well-being whilst people about their life satisfaction, and ultimately, I think I'm more interested in people's happiness than their less satisfaction, their moment-by-moment in their lives. But we actually don't collect lots of data using really detailed measures of happiness. So things like the expense sampling method, we're just going to pick someone at the moment. So there are a couple of projects which have done this. And they're sort of multiple now. And so, you know, we know quite a bit about how satisfied people are with their lives. But we don't know nearly as much as at least I would like to know about how people experience their lives actually their happiness. And so actually, it's kind of an open question is, you know, we did have lots of really good data on happiness, how different would that be, from what we know about what impacts people's lives.

SPENCER: So this approach of looking at how happy people are with their lives? How does this get reinterpreted through a longtermist lens? If you say, Well, what I care about is sort of the long-term future of humanity and not just the short-term.

MICHAEL: Yeah, so I think the longtermist case for caring about subjective well-being, and I'm not, I'm not sure kind of exactly how strongly this texts out. So he's maybe like a kind of a good comparison case, imagine you could go back to the 1930s when GDP was better, and you could either make GDP better, or you could just replace something else instead. And it's pretty, pretty clear that like, lots of 20th-century policymaking is really being driven by GDP is a key measure of social progress. So if we can have like better measures, and we can get them to implement, then we can expect that like in terms of how people focus on things and assess policies, and so on, look at a public base we have, we can just stay those all in a much better direction. And that's the kind of promising global leave, like, what do you need to do to really get average quality of life as like as high as possible, it's like, we're gonna need to make it objective and think seriously about it and get it, you know, stuck into terms and then get it implemented in terms of how we make our decisions. And you might think, like, if we can kind of get this done now and get it to bed it in the mists will keep rolling and have an impact, some decades or hundreds of years.

SPENCER: So my final question for you is about the fact that you're a philosopher, and a lot of people think about philosophers not doing really pragmatic stuff, right? They're thinking about abstractions. And yet you're trying to use your work in a really pragmatic way. And it seems to me that you're actually leveraging your strengths as a philosopher in order to do that. So I'm wondering, how can philosophers help us do good better?

MICHAEL: I think the role of philosophers is to think really hard about what matters, and then work out what we should do next, as a result. I mean, I imagine your listeners will, you know, be quite sympathetic, the idea that philosophers can actually do practical things, maths, you know, the experience we have with, with all of the philosophers, which can seem to run around the effective altruism. Well, yeah. And so in my case, the direction that took me in thinking, oh, look like there's this thing called happiness, which matters, like, can we really measure it? Like, okay, if we can, what happens next? Like, what, in fact, to the answer? So I think that you know, the role of philosophers is to kind of doggedly pursue what matters, and then to try and bring clear-headed analysis to work out what then follows. In my case, I started with philosophy, a lot of ended up in economics, and now I end up learning about things like mental health and, you know, neuroscience, and this is like, a really kind of the apple is rolling quite a long way, the tree, I think, Do I really have anything useful to contribute to these fields, I have just felt so out of my depth that they are I think philosophers are able to bring is just high level of rigorous analysis. So you know, maybe I don't have expertise about like, you know, particular treatments of mental health, but I'm able to speak to those people. And then focus conversation like, okay, what are the priorities thought, like, how do we afford bail to compare these sorts of options?

SPENCER: So if people are interested in learning more or taking action on these ideas, what should they do?

MICHAEL: Yeah. So you can visit our website, you can think about ways to donate your money. So you know, if you're inspired by this sort of stuff that maybe you were holding off on mental health things because you didn't think there was credible research. Well, hopefully, that's now going to be there. So we're actually recording this just the week before we put out these reports. By the time this goes live. That information will be out there. And Happier Lives Institute itself is we do have a need for quite a bit of funding to be able to continue our search and try and find more effective ways to do good. So if that's of appeal to you, then I would love to hear from you. And you can find me on Twitter at @MichaelDPlant. That's the kind of place you go to.

SPENCER: Michael this was great. Thanks so much for coming on

MICHAEL: Thank you very much, Spencer.





Click here to return to the list of all episodes.


Sign up to receive one helpful idea and one brand-new podcast episode each week!

Contact Us

We'd love to hear from you! To give us your feedback on the podcast, or to tell us about how the ideas from the podcast have impacted you, send us an email at:

Or connect with us on social media: