5 ms·

The challenge with the link fights is often about the difference between data and conclusions. For example, if I claim that "marriage has a positive impact on

The challenge with the link fights is often about the difference between data and conclusions. For example, if I claim that "marriage has a positive impact on men" then cite a bunch of data showing that married men live longer and make more money on average etc... Have I truly proven my conclusion that marriage benefits men? Or does the same data show that healthy men with jobs are more likely to get married? Yet, I will stubbornly say that citation makes the conclusion unequivocal. 9/10 times the disagreement isn't data/citations but conclusions or criteria.

duxup 6y ago
Agreed. What the article actually means / we should think about it is a big part of it. For a while there were some accounts on reddit (and even HN) posting the same collection of links with regards to a particular topic. The links even came in the same order each time ;) It always came up with regard to a particular nation and a specific topic. (for the record they're long since gone, at least with those links) Anyway one of the articles was a US university professor who said something about some other countries actions and so forth. To the astroturfers that was supposed to mean something, X. What it really meant was a university professor said that if the data supplied by that country is true that it means that they maybe did Y. The article was was miles and miles away from X. Granted those were astroturfers, but I see the same issues with folks who honestly believe a rando quote means something significant. I often find myself asking people about their links "What do you feel that article means?"
TeMPOraL 6y ago
> 9/10 times the disagreement isn't data/citations but conclusions or criteria. Or assumptions behind the cited data, and effect size. What I've seen very often (particularly around sketchy nutrition discussions about "natural" things and "evil chemicals") is that quoted research seems to support a given conclusion, but under assumptions that do not apply to the argument at all, or significance is so low as to make the result useless. Some study showing that a bunch of mice got cancer after getting injected with more artificial sweetener than you'd drink in a year does not show that artificial sweetener causes cancer in men. A study that says you get 0.003% greater chance of cancer from eating red meat every day has an effect size so small it's probably bullshit, but even if, 0.003% increase is not something worth caring about at all.
brabel 6y ago
I've seen even people who will attempt to discredit the data you've shown with kind of reasonable points like "the data is biased too, because it was collected by people who have the same bias". Well, yeah, that's possible, I guess... but if that's the case, then no discussion is possible at all and we have to rely on the feelings and intuitions of the people involved... and we all know how reliable that is.
jimbokun 6y ago
You are getting close to just saying it's impossible to know anything and giving up. The point of citing studies is to have some kind of empirical starting point for discussion, beyond just anecdata or ideology. Not as a declaration of absolute truth. > 9/10 times the disagreement isn't data/citations but conclusions or criteria. If it's not a good faith discussion, it's best to avoid those interactions entirely, or end them as quickly as possible to not waste your time.
hinkley 6y ago
Data interpretation is pretty hard and we get it wrong so often. I'm such a buzzkill when people show a graph that does not say what they think it says.

duxup 6y ago

Agreed. What the article actually means / we should think about it is a big part of it. For a while there were some accounts on reddit (and even HN) posting the same collection of links with regards to a particular topic. The links even came in the same order each time ;) It always came up with regard to a particular nation and a specific topic. (for the record they're long since gone, at least with those links) Anyway one of the articles was a US university professor who said something about some other countries actions and so forth. To the astroturfers that was supposed to mean something, X. What it really meant was a university professor said that if the data supplied by that country is true that it means that they maybe did Y. The article was was miles and miles away from X. Granted those were astroturfers, but I see the same issues with folks who honestly believe a rando quote means something significant. I often find myself asking people about their links "What do you feel that article means?"

TeMPOraL 6y ago

> 9/10 times the disagreement isn't data/citations but conclusions or criteria. Or assumptions behind the cited data, and effect size. What I've seen very often (particularly around sketchy nutrition discussions about "natural" things and "evil chemicals") is that quoted research seems to support a given conclusion, but under assumptions that do not apply to the argument at all, or significance is so low as to make the result useless. Some study showing that a bunch of mice got cancer after getting injected with more artificial sweetener than you'd drink in a year does not show that artificial sweetener causes cancer in men. A study that says you get 0.003% greater chance of cancer from eating red meat every day has an effect size so small it's probably bullshit, but even if, 0.003% increase is not something worth caring about at all.

brabel 6y ago

I've seen even people who will attempt to discredit the data you've shown with kind of reasonable points like "the data is biased too, because it was collected by people who have the same bias". Well, yeah, that's possible, I guess... but if that's the case, then no discussion is possible at all and we have to rely on the feelings and intuitions of the people involved... and we all know how reliable that is.

jimbokun 6y ago

You are getting close to just saying it's impossible to know anything and giving up. The point of citing studies is to have some kind of empirical starting point for discussion, beyond just anecdata or ideology. Not as a declaration of absolute truth. > 9/10 times the disagreement isn't data/citations but conclusions or criteria. If it's not a good faith discussion, it's best to avoid those interactions entirely, or end them as quickly as possible to not waste your time.

hinkley 6y ago

Data interpretation is pretty hard and we get it wrong so often. I'm such a buzzkill when people show a graph that does not say what they think it says.