Hacker News Reader

Comment by zeteo | Hacker News Reader

This reads like someone consciously decided "Hey, let's build the ultimate poster boy for bad statistical studies!".

1. Sample bias: "Why just study WEIRD [1] subjects? Let's do male Harvard graduates!" (Yes, half the study included inner city men, and one eighth of the duration featured women. It's still super heavily biased towards Harvard men.)

2. Correlation is not causation: "Hmm, health is correlated with relationship satisfaction. Could there be a common cause for both? Or maybe people like to hang out with healthier peers? No, the clear conclusion is that working on your relationships magically makes you healthier."

3. Inconsistent data collection: "Those '30s nincompoops were measuring skulls and handwriting. We'll stop doing that and take MRIs instead. But it's still the same study!"

[1] https://schott.blogs.nytimes.com/2010/07/14/weird/

View on HN

This sort of snarky comment that dispenses with the life work of others by means of internet clichés is the sort of thing for which pg coined the term "middlebrow dismissal". Perhaps you really do know better than specialists who worked in a field for decades, but if so, you're doing your superior knowledge an injustice by presenting it so basely. We're sincerely hoping to do better than that in this community, and would appreciate it if you'd join us in the effort.

More at https://news.ycombinator.com/item?id=14818590.

The parent comment raises some good points. The sample size here is too small and the article's conclusion has loads of confirmation bias, also displayed in majority of the comments here.

The best argument in science is repeating the study is it not?

> This sort of snarky comment that dispenses with the life work of others by means of internet clichés [...] specialists who worked in a field for decades

So if the work is so polished then it should be really easy to defend on its merits right? I find it curious that your main argument is instead about the work's authors.

That's not the argument at all. The argument is that the internet genre of the snarky dismissal is harmful in its own right. It subtly (and not-so-subtly) encourages many of the things we don't want here. So please stop posting like that.

If you have a substantive point to make, make it thoughtfully; otherwise please don't comment until you do.

Thank you, moderator.

"Psychiatrist George Vaillant, who joined the team as a researcher in 1966, led the study from 1972 until 2004. Trained as a psychoanalyst, Vaillant emphasized the role of relationships, and came to recognize the crucial role they played in people living long and pleasant lives."

My first thought was that he found what he was looking for.

I guess that's the danger in social sciences.

That would _never_ happen in the hard sciences!

This is an unfair and reflexive assessment. The study has clear and obvious limitations, and does not pretend otherwise.

You are also acting like the results of modern psychology studies are somehow superior or more useful. Where did you get that outlandish notion from exactly? Every such study I read about seems to go through a cycle of hyped up result (eg power posing) to TED talk to book deal to failure to replicate to finally working out these results are either irrelevant, invalid or just fake.

1. is an issue with external validity (i.e. does the findings generalize to other groups). But the comparison being made in the study is within a given group with similar socioeconomic status (all Harvard males). Assuming that they only vary in the quality of their personal relationships, any difference in happiness because of this could, assuming it is causal, be interpreted as a generalizable effect of relationship quality.

Have you ever tried recruiting subjects for a short-term social sciences study? It's not easy! Now do one for 80 years... Damn near impossible!

I don't see anything wrong with the subjects being predominantly male Harvard graduates.

A healthy and happy life means something different for an African farmer or for an Oxford professor. Obviously this study is concerned about the later. I also imagine that back in the day there were far more male graduates at Harvard.

Except that people are generalizing from that to all humans, starting with the subtitle of the article we're discussing.

Reply on news.ycombinator.com