>>26609
> I just want to make the point that you can get by just fine on smaller samples as long as they are representative and useful for what you're trying to study. Sometimes it's done out of necessity, like a longitudinal study of rape victims or something.
I believe no scientist with an at least decent understanding of stochastics would content himself with or settle for a small amount of data unless there are financial and/or availability restrictions (there usually are). The rationale for any kind of statistics lies in the law of large numbers and the central limit theorem, respectively. Both deal with limits (n approaching infinity, where n can be interpreted as the number of tries or samples), so the larger the actual (finite) number of samples is, the more justified is any faith in your data being representative. In practice, this is connected with outliers: If you have only a small number of samples, outliers may distort your observed distribution severely, they may not be identifiable as outliers. So removing what might be outliers can mean ignoring the wrong portion of your data, depending on what you consider an 'outlier'. For instance, if you take samples far away from the arithmetic mean as 'outliers', it may happen that one single actual outlier shifts the arithmetic mean far enough away from the arithmetic mean of the remaining samples that some of them are far enough away from the distorted mean to be considered 'outliers', too. On the other hand, if you have a large number of samples, outliers will hardly affect or distort the distribution. They'd have to be far too many to be considered outliers anymore.
> I saw that he decided to skip the peer-review
Yes, he actually decided to publish in a magazine without peer review, which claims, however, to be peer-reviewed ('Although the Archives’ editor claims that “all articles submitted to the journal are reviewed in a rigorous way,” […]'). He also writes: 'To find out how many of those publishers are keeping their promise of doing rigorous peer review, I submitted ridiculously flawed papers and counted how many rejected them. (Answer: fewer than half.)' So the criticism of peer reviews being overrated stays justifiable.
> Researchers are biased in what they choose to research, how they research it, and how they interpret it.
Yes. Therefore, take research results with a pinch of salt.
> I'd like if they could just report things accurately. More often than not they totally misrepresent things.
Also true. I doubt it's intentional though.
> Not-fucking is just one thing out of a long list of things we don't want kids to do. Driving is another.
So kids shouldn't be driving cars, but shouldn't be not fucking either?
>>26615
> the truth isn't as important as what might happen if people believe the truth
Let's rephrase this: Even a truth that isn't dangerous itself might become dangerous if people actually know/believe it. It's been a common theme in science fiction, e.g. Men in Black, Stargate, or Battlestar Galactica. You always have to think twice what could happen if you tell someone the truth (or not even the truth; consider how much damage lies have caused throughout history). I'm not saying the truth should be hidden, I'm saying you have to be careful how you release it.
> Not nearly the same thing. People dramatize it a lot worse than it was.
Sure, but there were still people who were strongly against Darwin's evolution theory. There are even today people who don't believe it. The difference is that those people are a minority today, while they used to be a majority back then – of course. We don't live in a theocracy today, either, so the situations are indeed comparable; that is, comparable with the situation when Darwin's theory hadn't yet gathered as many people.
> racial characteristics in biology […] an environment where science doesn't do it's best
Some governments (those of countries which have biological weapons, I think, like Israel, the US, or India) do spend money on research into differences between races that could be exploited to design weapons that only kill one single specific race. I don't know either whether those scientists do their best, but I bet they do.
> Because when you get a study that confirms that narrative, people aren't going to look that deeply at potential methodological flaws
> Science isn't perfect… it's good, on the whole, but it's less good the more you ignore the ways it can go wrong.
True. But that's how people are.