Hacker News Reader

Comment by dmauro | Hacker News Reader

The author was referring to contrarian anecdotes in reply to statistical data, not someone's opinion.

- He is only arguing that there is universal truth in proxy by arguing that the scientific method is more valuable than anecdotal evidence. If you don't agree with that, you probably disagree with most of HN (pure conjecture, does anyone else here think anecdotal evidence is more valuable than statistical analysis?)

- I agree with you on this point. I don't think we should downvote comments because a fun discussion is what the comments are for, not to try to prove or disprove a study.

- Statistical analysis is not a specialized form of anecdote. That's a stretch.

View on HN

> He is only arguing that there is universal truth in proxy by arguing that the scientific method is more valuable than anecdotal evidence.

The strawman here is in equating (published) statistical analysis with the scientific method. Of course the scientific method is more valuable, but that's not necessarily relevant.

Please have a look at http://xkcd.com/882/ if you haven't already - what this comic describes is a very valid statistical analysis, according to the "scientific method", (only neglecting base rates like 99% of published papers do).

This is (unfortunately) very commonly practiced in the life sciences, including medicine -- sometimes knowingly but mostly unknowingly. Bad reporting not required for a horrible, long lasting effect on the future.

Scientists' use of statistics is often problematic, but that doesn't mean that its valid to counter statistics with anecdote unless you have some larger argument.

Right. But it is also not valid to bring up statistics unless you can properly qualify their relevance, which is almost never the case. This requirement sets much higher bar for anecdotes than published statistics, when the latter rarely deserves that high bar.

As a result, most arguments about science are invalid from a scientific-method point of view. But the claims brought up -- including anecdotes -- are often interesting and informative.

does anyone else here think anecdotal evidence is more valuable than statistical analysis?

To a certain degree I do. I would go even so far and say that the provider of the contrarian perspective is making a scientific contribution, by pointing out the lacking external validity of the original study. I personally believe that the more interesting phenomenons in science are those incidents, when "things" are acting different than expected. Contrarian anecdotes are most often the best starting places for these phenomenons.

Reply on news.ycombinator.com