> No offence, but this is why the plural of anecdote is not data. Your sample bias here is extreme - people who were idealogically inclined to take acid daily as an almost pseudo-religious sacrament are not a representative sample.
Case studies are a major source of data in science. People exposed to extreme radiation doses from accidents play a significant role in our understanding of the consequences of said exposure and ways to treat it.
If the people I knew were able to handle that kind of LSD use then I would have a different opinion because LSD would be safer. Is 1% the use 1% as dangerous? Now that’s where you need controlled studies because the effect size diminishes, but feed 50 rats 1 gram of substance Y and they all die and that’s strong evidence on it’s own no need for statistical analysis.
> feed 50 rats 1 gram of substance Y and they all die and that’s strong evidence on it’s own no need for statistical analysis.
Your theoretical study there tells you about substance toxicity, but it tells you very little about the real-world dangers of substance Y to rats.
Would rats choose to consume substance Y at all? How often and how compulsively? What are the effects on rats at the levels they choose to consume, rather than the effects on them at levels the scientists chose to dose them with? If most rats would choose to eat 1mg of Y, at most once a year, and that level doesn't seem to harm them, then that 50mg test has shown you nothing at all.
Substance Z doesn't kill at 50mg, in fact it's safe up to 1000mg, is it less harmful? Well that depends. If the rats compulsively gobble every bit of it they can find and then drop dead after a month, we can say that Z is much worse.
Imagine you worked in an alcoholic rehab centre. You have first hand evidence all around you of the horrific damage alcohol wreaks on its users. Many of them have liver problems, are aged by the booze and their tough lives. Some are battling cancers caused by the drink, others have severe mental impairments. Your impression of drinkers is likely very skewed.
That's what you're doing with your sample of LSD users. With alcohol you're explicitly excluding the equivalent group in your assessment - "You can reasonably safely consume a lot of alcohol without issue, as in 1-3 beers a night for a decade" - when we can be confident that the group of alcohol addicts is larger as a proportion and suffers from all sorts of issues you're happy to ignore.
This way of thinking is so common I'm considering naming it the "it's just a beer" fallacy. Most people who have a beer or two don't become alcoholics. Most people who drop acid once or twice don't become your dads friends. To gauge relative harm we need to figure out what 'most' means and what the consequences are across users. You don't get to select a moderate drinker and an extreme outlier acid-head and say "see! this one is worse!". That's not a valid comparison.
Case studies are a major source of data in science. People exposed to extreme radiation doses from accidents play a significant role in our understanding of the consequences of said exposure and ways to treat it.
If the people I knew were able to handle that kind of LSD use then I would have a different opinion because LSD would be safer. Is 1% the use 1% as dangerous? Now that’s where you need controlled studies because the effect size diminishes, but feed 50 rats 1 gram of substance Y and they all die and that’s strong evidence on it’s own no need for statistical analysis.