Hacker News new | past | comments | ask | show | jobs | submit login

It's been shown plenty of other places, including by OKCupid, which why I qualified with "kind of thing." It's common knowledge at this point.



There's lots of common 'knowledge' in social psychology including studies done by serious reputable researchers that turns out to not be so. This is a big enough issue that it's termed the "replication crisis". So, yes, maybe. Maybe not.


Being shown by a lot of studies means its been replicated a lot.


only the ones with valid statistics count

some guy dug up a bunch of studies done on abductions by satanists in the 1990s. remember that, from 60 minutes?

the punchline was there had never actually been such an abduction, but there were 30+ studies.

and that's replicated a lot.

the quality of the replication matters.


Alright, keep that goal post moving then. In the meantime, I'll go with the best we got.


There's no goalpost moving. It's the same thing I said originally.

You shouldn't be swayed by inappropriately small sample sizes. Your response was "well what if I had a lot of them?" My answer was "still no."

.

> In the meantime, I'll go with the best we got.

This isn't even close to the best we've got.


My original claim is that something like that was true and what someone else was referring to and I gave you a lazy link, and you complained about the guys terrible methodology and replication. I told you that there are dozens of studies and data analyses (some are on huge data sets [1]) out there and your response is that oh no they have to be quality. That's a goalpost move from "this is bad" to "all those (that I haven't even seen) are bad."

> This isn't even close to the best we've got.

If you've got that then show me and I'll have a look. Until then I'm going with that studies I've seen that all seem to say roughly the same thing (despite widely varying sample sizes and quality of methodology)

1. https://medium.com/@worstonlinedater/tinder-experiments-ii-g...


> Until then I'm going with that studies I've seen that all seem to say roughly the same thing (despite widely varying sample sizes and quality of methodology)

There are none with acceptable sizes. You're just talking.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: