Using an impressively detailed report from the CIS, I came up with a merit index for immigrants to the US by country of origin based on six factors: The percentage using one or more welfare programs, % self-employed, % lacking health insurance, % having graduated from high school or an educational equivalent, % having attained a Bachelor's and beyond, and the percentage living in poverty. Merit scores by country of origin and the estimated average IQ of the country correlated at a statistically significant .57, which jumped to .65 with India removed from the analysis.
The Department of Homeland Security's Yearbook of immigration statistics provides data on legal immigrants by country of origin as well as by the method used to attain legal residency (through family reunification, employment, or as refugees or asylees). I used the Yearbook's compilation for '07 to determine the percentage of immigrants from each country that are granted legal residency based on employment-based preferences and employer sponsorship and then related those percentages to the merit scores for the 25 countries detailed in the aforementioned CIS report*.
The percentage of minted residents from a country who gained their status through employment and the merit index of that country correlate at .66 (p<.001). Put simply, this suggests that immigrants who come for employment-related reasons do much better on a host of social indicators than do immigrants who are brought in by family members already in the US or who are let in as refugees and asylees.
That is not an endorsement of current work visa programs. The H-2A and H-2B programs allow for cheap, unskilled labor to be brought in, disincentivizing the mechanization of menial processes that drive secular increases in the standard of living over time. Tech companies pushing for the annual number of H-1B visas to be raised are effectively working to lower the cost of their labor, in turn making tech careers less lucrative to US natives who choose something less intellectually demanding but often as or more financially rewarding, like finance or accounting.
The point is not to argue the merits for more or less employment-based immigration, it's to show that family reunification and refugee granting are dumb ways to go about forging immigration policy** (and keep in mind we're only dealing with legal immigrants here--tolerating illegal immigration is even dumber). They lack merit testing and not surprisingly places that send large numbers of migrants who are granted residency through family reunification or as refugees are the places where the least successful migrants come from. Whatever the net effect of employment-related immigration is on the quality of life for US natives, the effects of family reunification and refugee-related immigration is worse.
An objection might be that refugees tend to come from impoverished, dysfunctional places and so by grouping them with those who are brought in via family reunifications, I'm making the latter look bad. Actually, the opposite is the case to the extent that it is occuring. There is no relationship between the refugee percentage and merit. Refugees constitute a significant percentage (20% or more, with 6% or less as refugees from the other 19) of immigrants from only six of the countries***, three of whom have done well in the US (Chinese, Iranians, and those from countries formerly part of the Soviet Union) and three of whom have struggled (Colombians, Cubans, and Haitians).
There is an inverse correlation of .43 (p<.04) between the percentage by country brought in via family-sponsorship and merit, however.
The strength of the relationship between the percentage of a country's migrants granted residency based on employment and that country's merit score exists even though the two are measuring different points in time. Most of the new legal residents have just recently arrived, while the merit index is based on the performance of those already here. This suggests a long-term pattern of those being brought into the US for family reunification or as refugees being more likely to fall to the bottom of society than are those who come for reasons directly related to employment.
* For a few countries, data was missing in multiple columns for 'disclosure' reasons. When one of this missing points was in the Diversity Visa Lottery field, I assumed it to be zero. When the two missing points were from other fields, I took the total, subtracted all other known fields, and split the remaining discrepancy equally between the two missing fields. This was necessary for four of the countries but its effects are negligible. It represented less than 1% of the total migration in all four cases.
** The Diversity Visa Lottery system is probably the dumbest approach of all, but "diversity immigrants" (really, that's what they're called) do not constitute a significant portion of immigrants from the 25 countries being considered, as these countries are major senders of people to the US and consequently their populations are not eligible (or their eligibility is severely limited) to participate in the DVL. Diversity immigrants make up less than 1.5% of the total new legal residencies granted for people from the 25.
*** Contemporary nations that were part of the Soviet Union are combined to mesh with the CIS study, which looked at immigrants from the former-USSR as a single 'country'.