Psychology “incompatible with hypothesis-driven theoretical science” | Hacker News Reader

Psychology “incompatible with hypothesis-driven theoretical science”

madinamerica.com

36 pointsbranko_d16 comments

Not that it's not interesting, and perhaps there's even a kernel of truth to it -- who ever thought of psychology as an exact science? -- but note that this website, madinamerica.com, is dedicated to fighting psychiatry in all its forms. They very much cherry-pick studies like this to report on; don't expect a study proving the opposite would show up there.

When ingesting news, even if you're willing to accept outlying or extreme points of view into your mind, you should at least be aware of the agenda of the news source.

I don’t think the idea is psychology is, or should be, and exact science. But failing to replicate for 60% isn’t an issue of being exact or imprecise, it’s just bad or wrong or fraudulent.

I don’t the counter to cited studies showing problems is criticizing the web site presenting them, but making substantial claims to refute. Or ideas for improving.

Some of the ideas called out- preregistration, open data, more systematic reviews- are good and should be embraced by psychologists who want to have a reputable science. Not an exact science.

Psychology bears the brunt of the criticism for lack of reproducibility, but here’s the secret: the “hard” sciences aren’t much better.

The claim would be they’re irreproducible for other reasons than psychology studies are, but how could we possibly know?

Many years ago I had the displeasure of attempting to teach statistical inference to psych students. My background is applied maths. Psychs tend to over-sample (so combined with step-wise regression can support _any_ hypothesis), over-test (so type II error rate is totally out of control) and mangle data, rejecting data that doesn't fit the hypothesis - I could go on.

I belive that the academic psychologists are desperate to be considered scientists and not voodoo practitioners. That's why their research is so loose.

But is that the only way to do Psychology? Does for example William James or Freud or Jung use statistics for their analysis? I doubt.

I also had a Professor of Psychology, the Dean, tell me that "mathematicians don't understand statistics".

Can confirm, am mathematician.

It's not like I can't understand it, but it never attracted me enough to put in the work. (And I am quite put off by questions which require statistics to settle, which, yeah, I know, it's most of science.)

I'm an engineer, and while I found the rest of math doable but boring, I always found statistics so fascinating because it's so useful to understand the world.

> belive that the academic psychologists are desperate to be considered scientists

Well yes, they kinda have to, because if you aren't a scientist other people (like mathematicians) will look down at you. And if you cant show fancy formulas and statistics to show you proved things you are running into trouble getting funding, because you arent considered a science and thus not worth any money. Just look at the humanities, who are having an even harder time getting funding.

Yes!! I think it is to justify grant applications. Trouble is they're using stats as weasel words and the people who approve the grants have no idea.

> Just look at the humanities, who are having an even harder time getting funding.

Many people who do research in humanities are keen on spreading their left-wing political agenda. Couldn't this also be a huge part of the story why many people don't take humanities seriously?

If education tends to generate left-leaning ideology, maybe the people not taking them seriously are either just (a) right-leaning (b) uneducated or (c) both.

It's NOT SCIENCE if this isn't part of it.

Repeatability is essential. Without it, you are just guessing and making shit up. That may well be the nature of psychology in some (or even most) cases. Then it's simply not a science at all or ever. Accept that and move on.

I have a friend who, as a psychology researcher at a major university, did a meta-analysis of the reproducibility of psychology research and found much the same as this article. Results were generally not reproducible and there is way too much confirmation bias.

Something else he found was that the tenure track policies for most universities encourage poor study design by rewarding the quantity of research published rather than the quality of the research. That resulted in research departments doing 10 studies with 10 participants rather than 1 study with 100 participants, given a certain amount of funding available.

I think they suffer from the problem that the brain is very complex, and it's impossible to boil it down to simple - and testable - rules.

Rule-based, understandable, AI has mostly been replaced by deep learning, which is more difficult to understand.

It has useful properties, but we don't know why, they're emergent.

There's a reason it's called "the hardest science."

Reply on news.ycombinator.com