This is a topic that I've spend some time thinking about. I believe that "160" doesn't really mean what it seems to mean and the tests just aren't good in the tails.
First, there's the problem of the test itself. To get a math score with a GE, there are a few problems at each grade level -- not many and many of them are quite easy so that vague and passing familiarity would allow a child to answer the question despite not knowing that much. A score that returns "11th grade GE" does not mean that a child knows the intricacies of pre-calc and trig. So I think the tests are pretty weak.
Next, there's the issue of norming. I looked harder at IQ tests than at achievement when researching this, but a norming sample of 2000+ children may not include any children scoring very high. If there's no norming sample in the tails, it's much harder to get an accurate measurement of what number right would be something = to the rarity of 4 sd above the mean. For IQ, none of the original WISC-IV sample scored above 150 for full scale. I understood this to mean that there was a lot of extrapolation and guessing in determining the raw score differences that would give 150, 155, or 160 for full IQ. If your sample is miniscule, the error bars are huge. If the error bars are huge, a random guess or miss makes way too much impact on the score.
I agree that GE helps with the "out there" factor for a 160 at a young age. At late elem, there's probably not enough ceiling to tell, but a 6 yo who scores 160 with a GE of >12.9 has answered more questions than a 6 yo with 160 and a GE of 6.1.
Finally, I personally have seen way too many crazy out there scores to believe that 160 really indicates the rarity of 4 sd above the mean. I've seen multiple scores with all subtests at 160 indicating even more rarity and although the children are definitely smart, they are not that unusual.
I'm not sure what good such tests are except for another clue in putting together an understanding of a child. If a kid hates school at 6 and complains of boredom and the WIAT shows >12.9 in multiple subtests, the kid is probably bored. If a kid has 160 achievement test scores overall and an IQ of 105, I'd wonder if there was some missing information to explain the disconnect. Similarly, if an IQ score is 160 and achievement averages 110, that seems to be a mismatch worth investigating. I'm not convinced that an achievement score of 150 is really worse than 160 or that there is any useful information about what a child actually knows from the WIAT or WJ achievement.
