Hello Guzey. Yes, this paper has caused quite a stir. It’s very hard to understand (at least as a non economist) what the paper is saying as it’s filled with jargon and formulae, and the argument seems to turn on statistical considerations that are outside my scope of expertise. I had to ask a couple of economists to explain it to me.
As I now understand it, the authors’ main objection is to the use of 3-point scales. What you can infer from such scales depends on what you think the underlying distribution of the data is that’s being allocated into those three categories. If you make very different assumptions (e.g. utility is unbounded and the points on the scale are not ‘equal-interval’, i.e. the same distance apart) you can reverse the results. Such a reminder is useful and it’s important to examine these assumptions. That said, their argument is not specifically about happiness measures, but about the use of scales with only a few points, so it’s somewhat confusing that that’s where they’ve leveled their critique. It’s increasing thought that 3-point measures are unreliable and the modern literature predominantly uses 10-point scales. The authors don’t test the robustness of their claims against 10-point scales (where, I gather, they would be less plausible).
I don’t think I’ve update my views on the basis of this article. I still don’t fully get what the paper is doing, so I’m relying on those I consider my epistemic superiors in this domain—the economists I’ve spoken to—and they don’t think this poses a problem (relies on odd assumptions, doesn’t apply elsewhere, etc.).
I think that the evidence you present in section 4, e.g. that people interpret scales as equal-interval and that immigrants have similar SWB, could be a good response to this paper, though, because it suggests that we can interpret the discrete life satisfaction scale as cardinal and just aggregate it instead.
The theoretical results don’t depend on the scale being 3-point. Their argument deals directly with the assumed underlying normal distributions and transforms them into log-normal distributions with the order of the expected values reversed, so it doesn’t matter how you’ve estimated the parameters of the normal distributions or if you’ve even done it at all.
In the case of life satisfaction scales, is there any empirical evidence we could use to decide the form of the underlying continuous distribution?
They do suggest that you could “use objective measures to calibrate cardinalizations of happiness”, e.g. with incidence of mental illness, or frequencies of moods, as the authors have done something similar here https://www.nber.org/papers/w19243.
Well, I confess I don’t fully understand the paper and a further social scientist I’ve since spoken to had a different take on what the paper said altogether. I’ll try to bring this up with a few more people.
Hello Guzey. Yes, this paper has caused quite a stir. It’s very hard to understand (at least as a non economist) what the paper is saying as it’s filled with jargon and formulae, and the argument seems to turn on statistical considerations that are outside my scope of expertise. I had to ask a couple of economists to explain it to me.
As I now understand it, the authors’ main objection is to the use of 3-point scales. What you can infer from such scales depends on what you think the underlying distribution of the data is that’s being allocated into those three categories. If you make very different assumptions (e.g. utility is unbounded and the points on the scale are not ‘equal-interval’, i.e. the same distance apart) you can reverse the results. Such a reminder is useful and it’s important to examine these assumptions. That said, their argument is not specifically about happiness measures, but about the use of scales with only a few points, so it’s somewhat confusing that that’s where they’ve leveled their critique. It’s increasing thought that 3-point measures are unreliable and the modern literature predominantly uses 10-point scales. The authors don’t test the robustness of their claims against 10-point scales (where, I gather, they would be less plausible).
I don’t think I’ve update my views on the basis of this article. I still don’t fully get what the paper is doing, so I’m relying on those I consider my epistemic superiors in this domain—the economists I’ve spoken to—and they don’t think this poses a problem (relies on odd assumptions, doesn’t apply elsewhere, etc.).
I think that the evidence you present in section 4, e.g. that people interpret scales as equal-interval and that immigrants have similar SWB, could be a good response to this paper, though, because it suggests that we can interpret the discrete life satisfaction scale as cardinal and just aggregate it instead.
The theoretical results don’t depend on the scale being 3-point. Their argument deals directly with the assumed underlying normal distributions and transforms them into log-normal distributions with the order of the expected values reversed, so it doesn’t matter how you’ve estimated the parameters of the normal distributions or if you’ve even done it at all.
In the case of life satisfaction scales, is there any empirical evidence we could use to decide the form of the underlying continuous distribution?
They do suggest that you could “use objective measures to calibrate cardinalizations of happiness”, e.g. with incidence of mental illness, or frequencies of moods, as the authors have done something similar here https://www.nber.org/papers/w19243.
Well, I confess I don’t fully understand the paper and a further social scientist I’ve since spoken to had a different take on what the paper said altogether. I’ll try to bring this up with a few more people.