Coffee helps prevent COVID? - [Today's article] - Page 2

Want to talk espresso but not sure which forum? If so, this is the right one.
Capuchin Monk
Posts: 1192
Joined: 14 years ago

#11: Post by Capuchin Monk »

jpender wrote:I don't think that's fair. The correlations found aren't all equal in strength or sign. For example, tea consumption has a lesser apparent protective effect. And eating processed meats is correlated with a greater risk of COVID.
There are plenty of correlation isn't causation examples here. Some are really :lol: .

jpender
Posts: 3774
Joined: 11 years ago

#12: Post by jpender »

Well, of course.




But it's an extreme viewpoint to say that all correlation is just coincidence. That's more or less what tobacco companies tried to say for years and years. Proof is hard to establish and it makes sense to be cautious and skeptical. But that doesn't mean that it's all B.S.

Intuitively it makes sense that one's health impacts their immune response. And part of health comes from dietary choices. It's not an extreme stretch to believe that what we eat affects our health. And more and more lately it appears that coffee is generally a healthful beverage to consume. So... maybe it's true that it has a small effect on COVID susceptibility.

I actually find it far more interesting that having been breastfed as a baby is correlated with resistance to COVID.

Capuchin Monk
Posts: 1192
Joined: 14 years ago

#13: Post by Capuchin Monk »

jpender wrote:But it's an extreme viewpoint to say that all correlation is just coincidence.
I'll bring up "click bait" again. There are many stats the study groups can choose to compare / correlate. If they want to put something out online to attract readers and get paid for ads, they have plenty of options to work with. That's one of the downsides of internet info.

jpender
Posts: 3774
Joined: 11 years ago

#14: Post by jpender »

Capuchin Monk wrote:I'll bring up "click bait" again. There are many stats the study groups can choose to compare / correlate. If they want to put something out online to attract readers and get paid for ads, they have plenty of options to work with. That's one of the downsides of internet info.
It's really unfair to assume that because a subject might stimulate public interest that the motivations of the researchers are suspect. This was a paper published in a peer review journal not on an internet blog. The lead author has been part of studies of other possible COVID related factors. Why not judge the paper on its merits instead of whatever the name of the logical fallacy you're employing is?

TallDan
Posts: 315
Joined: 5 years ago

#15: Post by TallDan »

Capuchin Monk wrote:I'll bring up "click bait" again. There are many stats the study groups can choose to compare / correlate. If they want to put something out online to attract readers and get paid for ads, they have plenty of options to work with. That's one of the downsides of internet info.
Not just the internet. Have you ever seen the evening news shows on TV?

"Stay tuned, after the break we'll have a story about how coffee can help prevent COVID!"

"And also the weather forecast, which you could either look up in the next 30 seconds on your phone or watch 5 minutes of TV ads and hear someone read it to you 5 minutes"

User avatar
TomC
Team HB
Posts: 10527
Joined: 13 years ago

#16: Post by TomC »

These types of discussions rarely result in changing anyone's viewpoints or shedding much light on the topic. When the replies become repetitive or dogmatic, it's usually the sign of a downward spiral. I don't see much reason to keep such a thread open.
Join us and support Artisan Roasting Software=https://artisan-scope.org/donate/

User avatar
another_jim
Team HB
Posts: 13806
Joined: 19 years ago

#17: Post by another_jim »

The problem with research of this kind is that it requires a very deep dive into the details to see if it is valid or not. "Fishing expedition" research is a standing scandal in epidemiology and social science. A fishing expedition is taking a large data set and looking for correlations of any kind. Since there are thousands of possibilities, there will always be several correlations that are one in a thousand significant. Now pretend that the finding was your initial hypothesis, then apply significance tests based on that single hypothesis (and not that you've been doing combinatorics on thousands of variables), and you have an instanlty publishable finding based on statistical fraud.

I do not know if the coffee/covid correlation underlying this research is based on fishing expditions or not. The onus is squarely on the researchers, since everyone has been looking for correlates to Covid incidence. Given this level of data recycling, one in ten million significance levels are probably needed to make this valid.

The article's research tests whether the apparent correlation between coffee and covid is backed by any kind of inhibitory effect of coffee constituents on covid. Given there are about 1200 complex chemicals in coffee, is finding one or two that have an inhibitory effect, in vitro, with about 45% of patients, a fishing expedition or not?

I doubt even an epidemiologist with all the data on hand would know.
Jim Schulman
★ Helpful