The federal New Immigrant Survey, which surveyed 9,400 new legal immigrants in 2003 and intends to track them for decades, offers a fascinating source of data, much of it free to the public. The databases can be downloaded formatted for users of each of the three high-end statistical packages: SAS, SPSS, and Stata. So, if you are a data analysis heavyweight, here's a relatively unplumbed but important data source.

Jason Richwine calculated the Wechsler Digit Span IQ subtest scores for different immigrant groups from it, but nobody has yet published the IQ scores from the four Woodcock-Johnson III IQ subtests given to 3 to 12 year-old children of immigrants with legal permanent residency in the NIS.

You'd have to be careful in evaluating the scores, since on most of the subtests, between 25-50% of the kids got scores of zero. Presumably, they were utterly defeated by language problems due to the Woodcock-Johnson IQ test not having been translated into their exotic language. They should be left out. (In contrast, the very simple Wechsler Digit Span subtest was translated into a huge number of languages. One immigrant's child took it in Amharic.)

However, 704 of the Woodcock-Johnson tests were given in Spanish, so those scores are probably more or less fair. And you'd also need to work on how to convert Woodcock-Johnson scores to IQ scores.

It would be an ambitious project, but a significant one.