The IQ test is a test, people good at taking tests are good at taking tests. This doesnt rise to the level of a theory of intelligence, nor evidence that IQ measures it.
A vast amount of "mere statistical" research is pseudoscience; entire fields. It requires a lot of experience in hard scientific methodology, and an actual education in statistics. A lot of experimental sciences attract people interested in the experiments who just plug data into models and report the results.
The vast majority of such "research" is pseudoscience, and it's one massive house of cards. Most "heritability" research, population genetics research, psychometrics, psychology, nutrition, etc. is pseudoscience.
The whole type of statistics that has basically been invented by the social sciences is the basis for massive amounts of pseudoscience. Factor analysis, propensity scores, correlation coefs, etc. -- it's just an analysis of correlation.
Such is the poverty of thinking in these "social sciecnes" that they arent trained enough to realise that a model of correlations in effects isnt a model of their causes, not even nearly so. And this is esp. sevre in the case of complex systems like humans and animals.
All of the methods of IQ research are simple models of correlations in quiz answers. They do not rise to the level of a theory of anything, and their use is routinely circular and unethical.
You're just repeating yourself and then making hand-waving attacks on scientists. This is the tier of "argument" we hear from anti-vaxxers. Surely you can do better than vague ad hominems implying either incompetence or malfeasance? Such claims require evidence.
Yes there is a replicability crisis in much of the social sciences. Yes cargo cult science[1] has been a real problem and continues to be. Nevertheless there isn't a replication problem for psychometry. Countless researchers have dedicated their careers to showing that there is, and yet they have all failed to do so. I agree though there is room for improvement. IQ is only a model and we're measuring a proxy for intelligence, not intelligence itself. Still, it's a pretty good proxy, just as grip strength is a pretty good proxy for overall physical strength.
There is also high quality replicable research in all those fields you disparage. For example, generally speaking, metabolic ward nutrition research is of high quality.
There can't be a "replication" problem for IQ. There is no theory of intelligence provided, there are no independent valid and reliable measures of it, and there has been no work to show IQ measures it.
The argument is just, "I rekon intelligence has something to do with quizes like this, and look, there are correlations in people's answers!"
That's the premise of the entire field. It doesnt have a theory of intelligence, and any even plausible scientifically-grounded theory of intelligence would pretty clearly invalidate IQ as a measure of it.
HN isnt a place to give a lecture course on statistics, causal modelling and the scientific method. But the vast majority of people I'm aware of with serious work in those areas, regard this entire field as pseudoscience. It is only researchers within this field who believe otherwise, and they arent professionals.
> There can't be a "replication" problem for IQ. There is no theory of intelligence provided, there are no independent valid and reliable measures of it, and there has been no work to show IQ measures it.
The argument is just, "I rekon intelligence has something to do with quizes like this, and look, there are correlations in people's answers!"
Is your claim now that there is no such thing as intelligence? If we accept that premise then sure the whole field is bunk, because you can't measure something that doesn't exist. However I consider such a claim as ludicrous as saying that there is no such thing as strength, so if that's your position then I don't think I have much to learn from you.
Meanwhile, the g factor[1] is a theory of intelligence that basically says people who are good at one kind of cognitive task will also be good at other kinds of cognitive tasks. Is that entirely satisfactory? I've already said there's room for improvement. Nevertheless we have to start somewhere, and given the complexity of the array of phenotypes we associate with intelligence the g factor isn't a bad start.
> any even plausible scientifically-grounded theory of intelligence would pretty clearly invalidate IQ as a measure of it.
This sounds like an easy claim to support. Since you are enthusiastically anti-psychometry I assume you've done the research. What do you mean by "plausible scientifically grounded?" Would you please give a supporting example?
> HN isnt a place to give a lecture course on statistics, causal modelling and the scientific method. But the vast majority of people I'm aware of with serious work in those areas, regard this entire field as pseudoscience. It is only researchers within this field who believe otherwise, and they arent professionals.
HN is a place for satisfying intellectual curiosity. You would be contributing to the site's reason for existing by giving an abbreviated lecture actually showing that you have a point rather than just appealing to unknown authorities.
A vast amount of "mere statistical" research is pseudoscience; entire fields. It requires a lot of experience in hard scientific methodology, and an actual education in statistics. A lot of experimental sciences attract people interested in the experiments who just plug data into models and report the results.
The vast majority of such "research" is pseudoscience, and it's one massive house of cards. Most "heritability" research, population genetics research, psychometrics, psychology, nutrition, etc. is pseudoscience.
The whole type of statistics that has basically been invented by the social sciences is the basis for massive amounts of pseudoscience. Factor analysis, propensity scores, correlation coefs, etc. -- it's just an analysis of correlation.
Such is the poverty of thinking in these "social sciecnes" that they arent trained enough to realise that a model of correlations in effects isnt a model of their causes, not even nearly so. And this is esp. sevre in the case of complex systems like humans and animals.
All of the methods of IQ research are simple models of correlations in quiz answers. They do not rise to the level of a theory of anything, and their use is routinely circular and unethical.