Previous estimates of the foreign-born population by migrant status used a variety of often unrelated data sets. Using different data sets to estimate types of international migrants is problematic given the residual methodology used previously and in this analysis. To minimize inconsistencies, we used an integrated approach to calculate the migrant status of the foreign-born. Additionally, we generated standardized files for the 1990 Census and Census 2000 data which were used for the calculations of the number of each type of international migrant. We also used a standard method to impute values for missing variables and characteristics in these files.
Data Sets Used for Calculations
For temporary migrants, data from the Census 2000 Supplementary Survey were used as a proxy for Census 2000 data that were not yet available. A review and evaluation of these data suggest they are a reasonable approximation for yet unavailable detailed Census 2000 sample data (Malone, 2001; Deardorff and Malone, 2001).
For 1990, we used the census sample edited detail file modified to remove the category of "some other race." Missing data for country of birth were imputed using responses to the country of birth question, independently for each state. For 2000, we used preliminary census sample data, based on intermediate weighting schemes and editing procedures, and modified to match the 1990 racial categories (Malone, 2001). The preliminary Census 2000 sample data were available only for certain variables, including age, sex, race, Hispanic origin, country of birth, citizenship, and year of entry into the United States.
Using these data sets, we estimated the foreign-born in 2000 by migrant status (legal immigrants, temporary migrants, and a residual component consisting of quasi-legal and unauthorized migrants) by DA race (Black, NonBlack), sex, and A.C.E. age groups (ages 0-17, 18-29, 30-49, and 50 and older). In addition, we estimated the number of foreign-born by migrant status, sex, A.C.E. age groups, and mutually exclusive race/ethnic categories (non-Hispanic White, non-Hispanic Black, non-Hispanic Asian and Pacific Islander, non-Hispanic American Indian and Alaska Native, and Hispanic).
Review of Previous Methodology Used to Calculate the Foreign-Born Population by Migrant Status
Historically, the Census Bureau employed demographic analysis to evaluate the accuracy of census results. In the course of these evaluations, the Census Bureau made assumptions regarding the level of legal migrants and the residual foreign born. Based on previous research about census coverage of these populations, the Census Bureau traditionally assumed a higher coverage rate for legal immigrants than for the residual foreign born (Costanzo et al., 2001). After the 1990 Census, the Census Bureau expanded estimates of international migrants to include temporary migrants to the United States, as previous estimates of temporary migrants were limited to the number of foreign students in the country. A primary reason for estimating temporary migrants was to account for this group independently of the unauthorized population in the decennial census. Other reasons were to develop better demographic characteristics of the foreign-born population (specifically, temporary migrants do not age during the decade because of legal requirements restricting length of stay in the United States), and to evaluate the upcoming results of Census 2000.
A major component of the DAPE project was to validate estimates of the number of international migrants (legal immigrants, temporary migrants, and unauthorized migrants) in 1990. After our validation work, we used the same methodologies to develop estimates of the number of international migrants for 2000 using available data. Independent teams were formed to evaluate work on each of these components of international migration. For detailed descriptions of how the teams revised and improved previous estimates, see Costanzo et al., 2001; Mulder et al., 2001; Cassidy and Pearson, 2001; and Perry et al., 2001.
Evaluation of the Methodology Used to Calculate the Foreign-Born Population by Migrant Status
Although researchers have routinely adjusted census level estimates of unauthorized migrants to account for those missed in the census, they usually do not adjust explicitly for similar undercounts to the legal immigrant and temporary migrant populations (Passel, 2001; Bean et al., 2001).
To assess the robustness of these levels to varying assumptions about the undercount of legal immigrants and temporary migrants, we developed several scenarios. As discussed later, the application of alternative assumptions results in different implied total foreign-born populations by migrant status. Nevertheless, the totals are not different enough to greatly affect the total DA estimates. Thus, while the results based on the 15-percent assumptions discussed above could vary, the variations would not be substantively different.
This evaluation of the methodology used to calculate the components of international migration addressed several questions:
- Was the assumption of complete coverage of legal immigrants and temporary migrants in the census reasonable?
- Was the assumption of 15-percent undercount for all residual foreign-born reasonable?
- Was the resulting estimate of the residual foreign-born a reasonable approximation of unauthorized migrants?
Evaluation Question 1
When assigning the foreign-born counted in the census to migration statuses, previous researchers at the Census Bureau assumed complete (100 percent) coverage of legal immigrants and temporary migrants in the decennial census. Because the residual foreign born were calculated in the residual category (foreign-born population minus the sum of legal immigrants and temporary migrants), the number of foreign-born counted in the census who were categorized as the residual foreign born would be even higher if the assumption of complete coverage of legal immigrants and temporary migrants was dropped.
Researchers studying the foreign-born, both inside and outside the Census Bureau, agreed that an assumption of complete coverage for legal immigrants and temporary migrants was unreasonable (Deardorff and Cresce, 2001). A change to this assumption of full coverage in the census would mean fewer foreign-born being categorized as legal immigrants and temporary migrants, and more foreign-born being categorized as residual foreign born during census level calculations.
Evaluation Question 2
Due to time constraints of the DAPE project, we assumed an average 15-percent undercount rate for the residual foreign-born, before meeting with external experts on international migration, even though we expected rates to differ for all groups (legal immigrants and temporary migrants, as well as the residual foreign born) and to vary by demographic characteristics (age, sex, race, and Hispanic origin) and country of birth. Although no consensus emerged on the appropriate levels of undercount to assume, experts we consulted agreed that the previously assumed average undercount of 15 percent was probably too high, especially given the undercount rates of other hard-to-count groups from Census 2000 (e.g., the undercount rate for Hispanic renters was less than 5 percent), (see Hogan and Whitford, 2001). Additionally, a 15-percent undercount represented the midpoint of previously used rates, but evaluation results suggest census coverage improved from the 1990 Census to Census 2000.
Evaluation Question 3
Most importantly, researchers were concerned about the possible implications of not correcting the assumptions discussed above. Although an explanation that some legal immigrants and temporary migrants were categorized as residual foreign-born was helpful, the media and policy makers could mistakenly interpret our results for the residual foreign-born as a "best" guess of the size of the unauthorized migrant population. Furthermore, because we had not included "quasi-legal" immigrants (e.g., refugees who had not adjusted to legal permanent resident status because of processing backlogs at INS) in the legal immigrant category, additional foreign-born were included in this residual category. For a more detailed discussion of these populations, see Costanzo et al., 2001.
Based on these discussions, we decided to produce alternative undercount assumptions for the foreign-born population and to evaluate the initial, detailed set of estimates against the alternatives. In addition, we are emphasizing that the residual group (as identified by our initial equation) is not an accurate portrayal of the unauthorized foreign-born. Finally, we identified additional information about the foreign-born population to separate the residual foreign-born category into two components: known components of the foreign-born (or those identified as quasi-legal) and the implied unauthorized population (Costanzo et al., 2001; Deardorff and Cresce, 2001).