1a. I think you've misunderstood. A low IQ alone is not intellectual disability, so the "exact same level" remark after e.g. isn't right.
1b. What are the equivalent measures you're talking about? Do you mean to say similar measured IQs in different places? To be abundantly clear regarding the interpretation of different IQs in different groups, though an IQ of 60 may signal intellectual disability reliably in a population with a mean of 100 but not in a population with a mean of 65, that doesn't imply or even suggest that IQ tests measure different things in those populations. We're talking about the far left tail of one population having a different meaning than the mean of another population, but around the mean, the thing being measured is still the same or measurement invariance would be impossible.
2a. Yes, imputations are always to be treated cautiously. It's lucky that they hold up so well.
2b. Well, given the imputations are reasonably accurate and given they should be reasonably accurate theoretically, I would say they're not *that* uncertain. They're usually +/- about 2 points from the later sampled value.
3. My spatial imputation of racial gaps was not intended to mirror Lynn's methods, just to show that spatial imputation is possible. The motivation for using water borders was just that they seemed right, since people often consider those to be real borders, and people cross those borders, so they ought to help with spatial imputation. Without them, the results are unchanged though.
4. A huge amount of uncertainty about doing them at all. I don't like Flynn effect corrections, but because of the explicit nature of the corrections for trends, the corrections applied still end up making it non-viable to say "What about the Flynn effect?" and that's what I was getting at.
1a. Pretty much. This is why IQ tests are not the way we define mental retardation. If there are no adaptive behavior deficits but a person is still just not that smart, then, well, they're not retarded, they're just not very smart.
1b. No, it's just saying that comparing the far left tail of one distribution to the central tendency of another doesn't really tell us that the meanings of their respective central tendencies are different. We know that it's possible to see strict invariance in comparisons between groups differentiated by 30+ IQ points and for all of this commentary about intellectual disability to still hold. Maybe if we oversampled people in the far left tails of a higher-performing group then that wouldn't hold. Might simulate this.
Thanks for the extensive, in-depth analysis of this highly maligned research.
1a. I think you've misunderstood. A low IQ alone is not intellectual disability, so the "exact same level" remark after e.g. isn't right.
1b. What are the equivalent measures you're talking about? Do you mean to say similar measured IQs in different places? To be abundantly clear regarding the interpretation of different IQs in different groups, though an IQ of 60 may signal intellectual disability reliably in a population with a mean of 100 but not in a population with a mean of 65, that doesn't imply or even suggest that IQ tests measure different things in those populations. We're talking about the far left tail of one population having a different meaning than the mean of another population, but around the mean, the thing being measured is still the same or measurement invariance would be impossible.
1c. We do have adaptability measures, they're just not great. Usually you use clinical judgment rather than questionnaires and rater inventories to describe adaptive behavior. These measures usually correlate with IQ tests at around r = 0.4-0.6 (e.g., https://www.mdpi.com/2076-328X/13/3/252, https://www.sciencedirect.com/science/article/pii/S2405844024070798).
2a. Yes, imputations are always to be treated cautiously. It's lucky that they hold up so well.
2b. Well, given the imputations are reasonably accurate and given they should be reasonably accurate theoretically, I would say they're not *that* uncertain. They're usually +/- about 2 points from the later sampled value.
3. My spatial imputation of racial gaps was not intended to mirror Lynn's methods, just to show that spatial imputation is possible. The motivation for using water borders was just that they seemed right, since people often consider those to be real borders, and people cross those borders, so they ought to help with spatial imputation. Without them, the results are unchanged though.
4. A huge amount of uncertainty about doing them at all. I don't like Flynn effect corrections, but because of the explicit nature of the corrections for trends, the corrections applied still end up making it non-viable to say "What about the Flynn effect?" and that's what I was getting at.
1a. Pretty much. This is why IQ tests are not the way we define mental retardation. If there are no adaptive behavior deficits but a person is still just not that smart, then, well, they're not retarded, they're just not very smart.
1b. No, it's just saying that comparing the far left tail of one distribution to the central tendency of another doesn't really tell us that the meanings of their respective central tendencies are different. We know that it's possible to see strict invariance in comparisons between groups differentiated by 30+ IQ points and for all of this commentary about intellectual disability to still hold. Maybe if we oversampled people in the far left tails of a higher-performing group then that wouldn't hold. Might simulate this.