I think another study that would be worthwhile doing was to look at kids that are specifically WAY ahead in milestones (not just slightly advanced but years advanced). So look at the kids reading at 2 or speaking 100 words at age 1. I'd be pretty surprised if they don't find a high correlation then!
I agree, that would be very interesting. It also might be doable, in a way: you'd have to publicise that [so and so] was interested in seeing children who [whatever]; if, on following those children, you found that they were far off the mean on some other criterion, like IQ, it would be reasonable to think there was something going on. You'd have the children seen by a psychologist at the time they were exhibiting the unusually early ability, so no recall problem, and you'd be able to eliminate problems like "does that really count as reading?" and "is that really a word?" by applying standard criteria.
Objections I can see would still have to be dealt with:
- possibly being singled out for this study might affect people's expectations of these children in a self-fulfilling way. If one could think of milestones that everybody agrees are not expected to be correlated with IQ, one thing you could do would be to recruit children early at those too, and use them as a "control group", without telling the parents which conditions constituted your "control group". You'd have to be careful about the ethics, but the problem that seems hardest to me is that I can't off-hand think of any milestone which has never been claimed by anyone to be correlated with giftedness! Age of first tooth eruption might be a candidate, since that's moved pretty firmly into the "old wives tale" category I think.
- in a study that relied on parents self-referring, or having access to a doctor who would refer them and then going along with it, there would be a danger that the parents who actually did refer their children for the study would not be representative of those who would be eligible. E.g. suppose 50% of all children, randomly distributed, meet criterion X, but that hearing about the study and being interested in taking part are highly correlated with education level and IQ. Then you'd expect the referred children to have higher than average IQ, even if that isn't true for the whole population of children meeting criterion X. This I think would be harder to deal with in experimental design, though I daresay a clever statistician could do something convincing to compensate.
