Talk:Intelligence quotient

Discussion at Talk:Nations and intelligence

 You are invited to join the discussion at Talk:Nations and intelligence. Generalrelative (talk) 17:58, 18 August 2020 (UTC)

Discussion at Talk:Heritability of IQ#Claims of "scientific consensus"

 You are invited to join the discussion at Talk:Heritability of IQ#Claims of "scientific consensus". Generalrelative (talk) 18:23, 25 August 2020 (UTC)

Style of article is both revealing and problematic

Nearly every section of this article that mentions a study that could suggest that 'g' is anything other than a perfectly inherited single entity is met with a curt closing counter point.

The style of writing in almost the entire article is basically this: "DISCUSSION POINT: such and such replicable respected study found that certain factors appear to influence, to some degree, scores. CURT COUNTER POINT: This is wrong because another study which is not replicable done by persons historically invested in g-factor research said it was wrong. END OF DISCUSSION NEXT SECTION."

The problem with this style of writing, apart from what it obviously reveals to anyone with "social" intelligence and "critical reasoning", is that each study and point is presented with equal weight. Which is very interesting when one considers the main and mathematically strongest criticism laid at the feet of strong 'g' proponents, which is that by arbitrarily manipulating the weights of certain factors, without changing the factors, one can arrive at a desired conclusion quite easily.

This article does not read like an encyclopedia article, it reads as a posturing defense of theories that strongly support a 'g' model of human intelligence. A lay person, after reading this article would be better prepared to argue why a 'g' view of intelligence is the accurate view of human intelligence rather than explain what 'g' or and intelligence quotient actually is. 68.7.65.150 (talk) 21:34, 25 September 2020 (UTC)

You're raising an important point. Could you be specific about which passages are biased in favor of the claims for a reified 'g' factor and how you'd propose to fix those passages? Thanks. NightHeron (talk) 21:51, 25 September 2020 (UTC)

And since you claim that "nearly every section of this article that mentions a study that could suggest that 'g' is anything other than a perfectly inherited single entity is met with a curt closing counter point", we would need you to point out how each of the "other studies" used for the "curt closing counterpoint" is not reliable. I also would like to see your evidence that g is "perfectly inherited". We need a lot more than your opinions. We need reliably sourced evidence. Sundayclose (talk) 22:00, 25 September 2020 (UTC)

While I don't think the IP editor's comment is fully accurate, I do think there's room for cleanup. On a quick look through, some particular areas:

• General factor (g) - Could use better sourcing and removal of WP:WEASEL use of "still accepted, in principle, by many". I'd also suggest we should move the three-level theory out of the 'g' section and leave it in the Cattell–Horn–Carroll theory section where it belongs (which is well written, apart from perhaps odd wording and needing sourcing on the 'g was earlier often subdivided into only Gf and Gc...' sentence). If this is cleaned up, I'm not convinced a direct rebuttal is needed if the final paragraph is cleaned up to indicate that Spearman's g, as he defined it at the time, is somewhat antiquated. Basically, direct the reader to the following sections.
• Reliability - ends with a critique of the scores (I updated this section, could use further improvement), and I think is reasonably fair.
• Validity as a measure of intelligence - I think this section is the best example of the original comment, as the end of a long list of critiques about whether IQ measures intelligence is a single WP:WEASEL sentence, suggesting the test "generally" has "sufficient statistical validity for many clinical purposes". Which purposes? What exceptions to that general rule? Are any of them directly related to IQ's measurement of intelligence (g or its sub-components), or is this sentence better suited to the Reliability section? Is there more or less validity for aggregate use on a cohort than individuals? I'm tempted to just move that sentence to the previous section and improve it from there, if not removing it entirely.
• Test bias or differential item functioning - Thoughts on moving this section underneath Reliability and validity and moving the "A 2005 study found that..." paragraph from the above section into this one?
• Flynn effect - Any reason not to include a quick summary of the Flynn effect in the Reliability section with a wl to the main article?
• Age - Suffers from some issues, IMO. Per the original comment, this is a section that reads clearly assuming that IQ = g. It's written that if IQ scores, despite all the variance issues listed in previous sections, can only exhibit age variance due to fluid/crystal intelligence. Which of course is silly, since IQ tautologically is normalized by setting 100 to the mean score for a given age. IMO, this section should describe the variance of IQ scores with age, not the variance of underlying general intelligence with age, but it currently reads as the latter. I also removed a 'however' which seemed to be trying to phrase the r correlates as 'good'. Should probably place meaningful context around these numbers, though. I'm an engineer that's reasonably comfortable with math and stats, but have no idea how to analyze these naked r-values.
• Health - Two cn tags from 2012, and fitting the criteria of the original comment. As above, I think it's worth trimming down to how health affects scores and how IQ tests are used in cognitive epidemiology, then direct readers to those specific articles.
• Crime - Wanted to point out this section as one that I found well balanced, and would suggest is a good example for the other sections. Shows historical links, but ends on modern critiques of causality.
• Group-IQ or the collective intelligence factor c - Doesn't seem to fit this article, as there's no Collective Intelligence Quotient test that I can find linked here. Let the main article be linked to from the articles on g et al.
• Group differences - Another section I think is written well, speaking specifically to IQ, not g while linking to articles where that association would be appropriate.

I made a handful of edits along my way. I think I'd like at least a little consensus (or at least, lack of disagreement) before I make some of the suggested changes. I also wouldn't claim to be qualified to make some of the edits. Bakkster Man (talk) 16:06, 29 September 2020 (UTC)

This seems to me to be a super thoughtful agenda for improving the article. Specifically with regard to folding "Test bias or differential item functioning" and "Flynn effect" into the "Reliability and validity" section, I'd been thinking along similar lines when doing a round of edits a few months back but didn't want to rock the boat too much all at once. Same goes for getting rid of (or straightening out) the weasel language in "General factor (g)" and "Validity". Since you've come to similar conclusions, that may be a good indication that the changes are indeed warranted. Generalrelative (talk) 19:53, 30 September 2020 (UTC)

I began some updates, starting with the history section. Mostly just removing duplicate information, actually. Even the weasel wording in the g section ended up being cited in the CHC section so that was nice and tidy. I also did the grouping of all the error sources, tagging the comments above which I can't easily remedy. I'm wondering if Sources of error or Accuracy and precision would perhaps be a better title for the section. On the other hand, I suppose reliability and validity generally are just being given more specific examples in the added sections. Also tagged the old health tags and removed group-IQ (the latter I recognize may be contentious). Bakkster Man (talk) 15:48, 1 October 2020 (UTC)

Looks great. I'd tend to agree with your "on the other hand," that issues related to accuracy, precision and sources of error are best presented as specific examples under the heading of reliability and validity. Generalrelative (talk) 22:04, 1 October 2020 (UTC)

Stephen Gould doesnt accept Iq tests!

"Some scientists have disputed the value of IQ as a measure of intelligence altogether. In The Mismeasure of Man (1981, expanded edition 1996), evolutionary biologist Stephen Jay Gould compared IQ testing with the now-discredited practice of determining intelligence via craniometry, arguing that both are based on the fallacy of reification, “our tendency to convert abstract concepts into entities”.[84] Gould's argument sparked a great deal of debate,[85][86] and the book is listed as one of Discover Magazine's "25 Greatest Science Books of All Time".[87]"

His personal opinion is not of any value to the subject, and should be removed!Cynthia BrownSmyth (talk) 08:04, 13 December 2020 (UTC)

Gould's argument was that IQ tests are not a valid measure of human intelligence broadly construed. This is entirely consistent with current scientific understanding. See for example the quote from Wayne Weiten in the same subsection: "IQ tests are valid measures of the kind of intelligence necessary to do well in academic work. But if the purpose is to assess intelligence in a broader sense, the validity of IQ tests is questionable." Note that Weiten's statement is WP:RS/WP:TERTIARY because it is from a recent, respected textbook, not an individual study, and therefore can be taken as representative of the field. Generalrelative (talk) 09:01, 13 December 2020 (UTC)

Discussion at Talk:G factor (psychometrics)

 You are invited to join the discussion at Talk:G factor (psychometrics). Generalrelative (talk) 18:17, 29 December 2020 (UTC)

"Online IQ Test Validity" listed at Redirects for discussion

A discussion is taking place to address the redirect Online IQ Test Validity. The discussion will occur at Wikipedia:Redirects for discussion/Log/2021 March 20#Online IQ Test Validity until a consensus is reached, and readers of this page are welcome to contribute to the discussion. Kokopelli7309 (talk) 14:59, 20 March 2021 (UTC)

Buffalo8's addition

I suggest that Buffalo8 discuss the statement they wish to add to the article here. That statement, as they've written it, is When averaging across all test batteries, samples of male subjects typically display an intelligence quotient advantage of four points over female subjects. As I stated in my edit summary, I believe that in the context presented this misleadingly suggests that evidence indicates a clear case for male superiority in intelligence when the author (Earl Hunt) is explicit that this is not the case. Indeed, here is what Hunt has to say about precisely this issue within the page range cited: If men have higher scores on some subtests, and women on other subtests, then depending on the weights assigned to each subtest you could produce a summary score that favored men over women or vice versa. And it is certainly true that if a test battery omits an important ability on which there are male-female differences, then the balance of men's and women's scores in an overall index will be different than it would have been had the omitted ability been evaluated. Generalrelative (talk) 07:13, 7 December 2021 (UTC)

As a side point, that figure of a 4 point difference is stated to be an outlier, present when comparing average male and female scores on two specific test batteries (WAIS-III and WAIS-R) and only in China and Japan; very far from what was stated in Buffalo8's edit. In the U.S. and Canada, male-female averages of those test batteries show a 2 to 3 point spread. But even still, the larger point is that the author considers this spread to be very plausibly an artifact of the types of ability measured by the test batteries in question rather than a real difference in general mental ability between the sexes. Generalrelative (talk) 07:22, 7 December 2021 (UTC)

In addition to the lack of context present in the Hunt source, the edit appears to be ignoring all research not mentioned/summarized in Hunt or conducted afterward, much of which finds no evidence of male/female sex differences in IQ. I am not 100% sure that a statement like Buffalo8's doesn't belong in the article, but I am confident that the WP:ONUS is on Buffalo8 to build consensus for the change instead of edit warring. Firefangledfeathers 13:20, 7 December 2021 (UTC)

I agree that the statement as written doesn't belong in the article. If anything similar is restored the nuances of interpretation and additional related research should be included; and it likely should get consensus here. I'm not accusing anyone, but on the surface this edit gives the appearance of either lack of knowledge of the research, or intentional POV. Again, that's not an accusation, but we need to be careful to avoid the appearance of bias or misunderstanding. Sundayclose (talk) 16:03, 7 December 2021 (UTC)

References

IQ Testing Reliability

"On aggregate, IQ tests exhibit high reliability, although test-takers may have varying scores when taking the same test on differing occasions, and may have varying scores when taking different IQ tests at the same age."

This statement is meaningless without also including the significance of the variability. The scores vary, but the degree of variation is not significant enough to merit being used to diminish the overall reliability of IQ testing, relative to all other forms of psychometric testing. Standard IQ testing is the most reliable form of psychometric testing that there is.107.195.106.201 (talk) 01:15, 20 January 2022 (UTC)

No, IQ is still suspect to errors. If your claim that it's the most reliable psychometric testing out there, then all other psychometric forms of testing are also suspect to errors in reliability and validity worse then IQ tests. 110.175.125.253 (talk) 14:16, 17 March 2022 (UTC)

Please use your registered account Vpha (talk · contribs · deleted contribs · logs · filter log · block user · block log) instead of using this Sydney IP. Thanks, Mathsci (talk) 19:59, 17 March 2022 (UTC)

Why I'm reverting edit citing Bouchard

First, the edit gives the misleading impression that Bouchard's estimate is generally accepted. The Minnesota Twin Studies have been criticized, and they're partially financed by the Pioneer Fund. Secondly, the 80% estimate goes back to Cyril Burt in mid-20th century or earlier, so it shouldn't be called "a recent estimate". NightHeron (talk) 09:32, 31 July 2022 (UTC)

A meta-analysis or review would be better, if one exists. BooleanQuackery (talk) 19:51, 31 July 2022 (UTC)

If it goes back to Cyril Burt, it should not even be called an "estimate". --Hob Gadling (talk) 10:04, 6 August 2022 (UTC)

Sorry, I didn't mean to imply that Burt's estimate was correct. Leon Kamin found extensive evidence that Burt faked his data, see The Science and Politics of I.Q.. NightHeron (talk) 10:16, 6 August 2022 (UTC)

Sorry, I didn't mean to imply that you implied that. --Hob Gadling (talk) 10:37, 6 August 2022 (UTC)

Short description improvement

Current short description: "Score derived from tests purported to measure individual differences in human intelligence."

My problems with it:

• Factually incorrect or ambiguous wording. The score is not derived from tests purported to measure individual differences in human intelligence. The score is derived from tests which themselves are not said to measure intelligence, but when calculated in aggregate they do. If instead this means different entire IQ tests instead of subtests, it should be changed to "score derived from a test . . ."
• The use of the word "purported" implies that IQ tests do not really measure intelligence: [1]
• The words "individual differences in human" are irrelevant for a short lead, and also not the most accurate. IQ tests don't usually measure differences in intelligence; they usually measure a single individual's intelligence. "Human" and "individual" are also redundant, as only humans can take IQ tests (lol) and more than one person can't take the same test.

My proposed changes: "Standardized score from a test designed to assess intelligence."

• An IQ test score is by definition both a standardized score and standardized in the sense that it is a normed test.

BooleanQuackery (talk) 20:10, 31 July 2022 (UTC)

Some comments on the wording:

• There is considerable controversy about the definition of intelligence and whether or not it's meaningful to assess it with a single number. In his book The Mismeasure of Man Stephen Jay Gould criticized the tendency to reify intelligence. We should not assert in wikivoice that something is a measure of intelligence. By using the word purported, we avoid taking a position on whether or not IQ measures intelligence.
• The commonly understood use of the word standardized is in the term standardized tests, meaning tests - such as IQ tests, the SAT in the U.S., and the tough college-admissions exams that are given in several Asian countries - that are given throughout a region or country and do not vary from school to school or city to city. In this context it is confusing to call the score standardized. What's meant is that the score is normalized so that it conforms to the normal distribution. This technical detail doesn't belong in the short description, especially since the average reader won't understand standardized to mean that the graph of the distribution of scores is made to look like a bell curve.
• I agree that "purported to measure differences in human intelligence" should be changed to "purported to measure human intelligence". Logically, the scores are purported to measure the intelligence of the people tested; the differences between scores are purported to measure differences in human intelligence. NightHeron (talk) 22:39, 31 July 2022 (UTC)

• The word "designed" also avoids stating that IQ is a measure of intelligence, though it is consensus among scientists that it is.
• IQ tests are both standardized in the layman sense (given throughout a region or country and do not vary from school to school or city to city) and standardized in the "normalized" sense. BooleanQuackery (talk) 00:26, 1 August 2022 (UTC)

• I won't debate the question of consensus here, except to note that a consensus among a certain subculture of psychometricians is not the same as a consensus of scientists.
• What I objected to was the term "standardized score" that you wrote, because that makes no sense to a nontechnical reader. NightHeron (talk) 00:56, 1 August 2022 (UTC)

The "no evidence" statement

This is a continuation of the discussion at Talk:Heritability_of_IQ#Comments_on_sourcing_and_consensus, regarding the statement in this article, "there is no evidence for a genetic component". In the Heritability of IQ article, the same statement was recently changed [2] to something that I think more accurately reflects its sources, but in this article (Intelligence quotient) it has not been changed. I suggest it should be changed the same way in both articles. In the other discussion, Firefangledfeathers suggested opening a new discussion about the same statement in this article, and I'd like to know his or her opinion about potentially making the same change here. --AndewNguyen (talk) 18:45, 5 August 2022 (UTC)

A key piece of information that is missing from the above post is that this statement is in reference to racial differences in average IQ test performance. As far as I'm aware, this was first discussed a year and a half ago on the Race and intelligence talk page (specifically here and here) and I do not see any reason to revisit the consensus that was established on the matter at that time: the sources do support the statement that no evidence exists linking racial differences in average IQ test performance to underlying genetic causes. For those who are not yet aware, there is a very firm consensus on the topic which was solidified by a nearly unanimous RfC last year. Happy to improve the language of the article for valid reasons, but I do not believe that AndewNguyen has yet presented one. Generalrelative (talk) 20:43, 5 August 2022 (UTC)

And now Generalrelative has raised the issue at the Fringe theories noticeboard. [3] --AndewNguyen (talk) 21:52, 5 August 2022 (UTC)

Please change the sentence or remove it. It isn't consistent with the sources it cites. Nor is it consistent with sources such as [4] from the American Journal of Psychology, about the various lines of indirect evidence that the average differences include a genetic component. Mr Butterbur (talk) 16:26, 7 August 2022 (UTC)

Russell Warne is far outside the mainstream. In particular, he contradicts the conclusions of the panel of experts convened by the American Psychological Association, whose report is cited in the article. NightHeron (talk) 17:30, 7 August 2022 (UTC)

There's not an actual contradiction between the two papers. The APA report says that genetic interpretations of group differences have no "direct evidence", like all of the other sources cited for the "no evidence" statement. (In this article the statement is sourced only to Hunt, but the APA report is cited for it in the Race and intelligence article.) Here are the relevant portions of the APA report:

It is sometimes suggested that the Black/ White differential in psychometric intelligence is partly due to genetic differences (Jensen, 1972). There is not much direct evidence on this point, but what little there is fails to support the genetic hypothesis.

A later part of the report says that genetic interpretations have no "direct empirical support":

The differential between the mean intelligence test scores of Blacks and Whites (about one standard deviation, although it may be diminishing) does not result from any obvious biases in test construction and administration, nor does it simply reflect differences in socio-economic status. Explanations based on factors of caste and culture may be appropriate, but so far have little direct empirical support. There is certainly no such support for a genetic interpretation. At present, no one knows what causes this differential.

Definitely the emphasis is different, because the APA report emphasizes the importance of environment for group differences and does not say much about the possible role of genetics. But the report doesn't comment about the existence or non-existence of indirect evidence for genetic factors, which is the subject of Warne's paper, except for by including that subtle qualifier of "no direct evidence" when the report discusses this matter. --AndewNguyen (talk) 18:26, 7 August 2022 (UTC)