Much of the analysis of U.S. immigration policy relies on data from the Bureau of the Census, gathered either during the decennial enumeration or the monthly Current Population Survey. However, these sources record only the place of birth of respondents, not their legal status; thus there is no reliable way to disaggregate legal from illegal immigrants or immigrants from nonimmigrants. In addition, data from the Immigration and Naturalization Service present snapshots of the overall immigration flow, but contain no information on changes over time, such as whether a legal immigrant originally entered the United States illegally, or entered legally but as a nonimmigrant.

In an attempt to remedy this situation, a group of prominent researchers, with funding from the National Institutes of Health, the INS, and the National Science Foundation, have developed plans for a New Immigrant Survey (NIS), which would track a large sample of new green-card recipients over a period of time. The first step in this undertaking was a pilot (to assess the cost and feasibility of the larger project) which surveyed a sample of immigrants who received green cards in the summer of 1996. The preliminary results, released last year, allow us for the first time to examine the status of legal immigrants alone. Although the findings contain no policy-related bombshells, they lend support to conclusions reached from other data.

Some findings from the survey's data on legal immigrants 25 and older:

• The hourglass distribution of immigrant skills applies even when looking exclusively at legal immigrants. The proportion of legal immigrants in the 1996 survey with postgraduate education (21.1 percent) was three times higher than native-born Americans counted in the 1990 Census, but the proportion of legal immigrants with less than nine years of schooling (19.5 percent) was also higher — more than twice as high as for the native born. Also, the rate of school enrollment among people aged 18 to 24 years was about 10 percent lower among legal immigrants.
   
• Since the survey examined only legal immigrants, their educational level was higher than the 1990 Census figures for the foreign born as a whole. This confirms the assumption that illegal aliens are the least-skilled component of the immigrant flow. This difference is particularly notable at the extremes of the educational spectrum: While the census found that 12.1 percent of all the foreign born had less than five years of schooling, the survey found only 6.8 percent; and the census found 5.2 percent of all immigrants with 19 or more years of schooling, while the legal immigrant survey found 9 percent. Although this also was apparent among Mexicans, the difference in education between legal immigrants and all immigrants was not large: the 1990 census found that 56.7 percent of recent immigrants born in Mexico had less than nine years of schooling, while the survey found that 50.7 percent of new legal immigrants from Mexico had that level of education.
   
• The visa category of admission made a big difference in the level of schooling. Naturally, those admitted under the employment preferences (including spouses) had the highest mean years of schooling (16.1), followed by diversity immigrants (winners of the visa lottery), at 14.5 years. Interestingly, there was little difference among immigrants admitted as refugees/asylees, siblings, and spouses, with years of schooling ranging from 13.1 to 13.6 years. The least educated, probably because of age, were those admitted as parents of adult citizens, with an average of 7.5 years of schooling.
   
• Almost 20 percent of legal immigrants in the survey had snuck into the United States illegally on their first or last trips here. Because the survey did not ask about intervening trips, and did not ask whether any of the newly minted legal immigrants had overstayed temporary visas, the proportion of legal immigrants who had been illegal aliens is certainly higher. This confirms earlier estimates that at least 25 percent of each year's legal immigrants are illegal aliens using the system to launder their status. Marriage to a U.S. citizen is the chief means for former illegal aliens to acquire green cards; almost 40 percent of the new legal immigrants who had snuck in illegally (on their first or last times across the border) acquired green cards through marriage to a U.S. citizen, as opposed to only 26 percent of those surveyed who had not done so. Mexicans accounted for almost half of the former illegals (47.4 percent), but only 7.4 percent of those who had not snuck in illegally.
   
• The survey also highlights the adulterated nature of employment immigration. For instance, immigrants (younger than 40) admitted under the employment-based categories had by far the longest prior residence in U.S. (4.5 years), which makes sense since an undergraduate degree usually takes four years to complete and H-1B "temporary" visas run out after 6 years. In other words, the survey suggests what is already widely suspected: "temporary" visas for work or study are, for many, merely the first step toward permanent residence. The survey also found that 22 percent of the sponsors of employment immigrants are individuals, over two-thirds of them foreign-born. Almost four percent of individual sponsors of employment-based immigrants are their relatives, and almost three percent of the U.S. citizens sponsors of spouses are also their employers.

These and other preliminary findings from the survey underline the potential value of a full-fledged New Immigrant Survey, planned to start some time in the next few years, if funding is secured. This information should help highlight some of the flaws in our legal immigration system and prepare the ground for meaningful reform.