1. Rosser sets up a false dichotomy that defines away the interesting question. He is correct that the happiness curve is a statistical projection of the effect on subjective wellbeing of aging, isolated from other variables. This is a point I emphasize repeatedly in the book (not, as Rosser says, in “very slightly note[d]…caveats”). How individuals feel about their lives at any given time depends on many things (see, e.g., pages 69-75 of the book), and aging is just one of those things.
The book’s thesis is not that age, by itself, predicts happiness, but that age has an independent and significant effect, one which can operate in perverse and counterintuitive ways and is especially noticeable and puzzling to those who, like me, do not experience big fluctuations in the other variables. One need not choose between looking at unadjusted, as-experienced (“raw”) happiness and understanding the ceteris-paribus (“adjusted”) effect of aging. To the contrary, the only interesting thing to do is to understand how the variables interact; and the only way to do that is to explore lived lives and subjective experiences, as well as data, which is what I do.
2. The charge of cherry-picking is better aimed at Rosser than at me. He cites a single study, by Steptoe, Deaton, and Stone, who are very formidable scholars indeed (Stone’s work on stress is featured in my book), but who consider unadjusted (i.e., raw) happiness, as opposed to the independent effect of aging—a perfectly valid approach, but one that asks and answers a different question. For those who are interested in this important distinction and want a more thorough discussion of it, Blanchflower and Oswald explain it in more detail in this 2009 paper (PDF) and this 2017 paper (PDF).
Several other papers have taken issue with the U-curve finding. I chose not to include them because there are few of them; they are mostly old; they mostly are not about the phenomenon I’m concerned with (the independent effect of time on happiness, rather than unadjusted, average happiness levels); and the body of evidence supporting the U-shaped effect of time is now so large and variegated as to place an insurmountable burden on someone who would say the phenomenon does not exist or is not significant.
3. I don’t really understand Rosser’s complaint about country comparisons. I would have thought that including evidence that the happiness-age relationship varies across countries, and that no one pattern is universal, would be a mark in my favor, from a skeptic’s point of view. Again, this is not an “admission” (much less a “distortion”); it’s an important part of my argument, which is that what’s going on is complex and multivariate, and talking about, say, an inevitable midlife “crisis” gets the whole story wrong.
Be that as it may, if Rosser wants to assert that national or other variations and exceptions disprove the U-shaped relationship, taking potshots won’t suffice. He needs to contend with a well developed and sophisticated body of research (such as the 2017 paper mentioned above, which uses seven recent data sets, covering 51 countries and 1.3 million randomly sampled people, and finds the U-shaped pattern in all seven adjusted data sets as well as in five of the raw data sets).
Of course, science never closes its books. If my work stimulates Prof. Rosser or other talented scholars to pose new questions and challenges, that will make me happy.