4 ms·

I'm not misunderstanding. The word link literally means a connection or relationship. There is neither of those two things in this data, except that their 1.3

I'm not misunderstanding. The word link literally means a connection or relationship. There is neither of those two things in this data, except that their 1.3% of "abusive" smokers had heart attacks versus 0.5% of non-smokers. You cannot draw a link from that alone, but they are stating there is a link. You can state there is a link between diet and heart disease, because there's a proven causal relationship. Not so in this case. Also, the variable was independent because they study stated that they only included "abusive" smokers, implying they excluded casual or occasional smokers. They split the data on that variable. That's a selection process.

hirundo 5y ago
You are describing epidemiology, which is not an experimental science, because it does not have the luxury of independent variables. And you are correct that it is weaker at establishing links than a controlled experiment. So it's mostly a resort for when an RCT is impractical or as a way to select which RCTs to pursue.
- 52-6F-62 5y ago
  I didn't say it was experimental science. I said their use of the word link implies a causal relationship because that is precisely what the word link means. If they are not describing a link, and instead are describing a correlation then they should do so. Calling their findings a "link" is much more final and implies causation. It's not an arcane definition I'm pointing to here—it's the only definition for the word. There is no link. There is an apparent correlation based on their filtered dataset. Epidemiology isn't meant to create problems, it's meant to resolve them. They are borderline fabricating a problem with their misleading study. Maybe next time they should indicate a "possible link" since that is all they actually have. When studies make hard-nosed, misleading conclusions about health subjects that are presently on the political fence I get suspicious.

hirundo 5y ago

You are describing epidemiology, which is not an experimental science, because it does not have the luxury of independent variables. And you are correct that it is weaker at establishing links than a controlled experiment. So it's mostly a resort for when an RCT is impractical or as a way to select which RCTs to pursue.

52-6F-62 5y ago
I didn't say it was experimental science. I said their use of the word link implies a causal relationship because that is precisely what the word link means. If they are not describing a link, and instead are describing a correlation then they should do so. Calling their findings a "link" is much more final and implies causation. It's not an arcane definition I'm pointing to here—it's the only definition for the word. There is no link. There is an apparent correlation based on their filtered dataset. Epidemiology isn't meant to create problems, it's meant to resolve them. They are borderline fabricating a problem with their misleading study. Maybe next time they should indicate a "possible link" since that is all they actually have. When studies make hard-nosed, misleading conclusions about health subjects that are presently on the political fence I get suspicious.