INTEGRATION OF MULTIPLE DATA SOURCES
IN IMMIGRANT STUDIES
Teresa A. Sullivan
and
Marta Tienda
CDE Working Paper 84-18INTEGRATION OF MULTIPLE DATA SOURCES
IN IMMIGRANT STUDIES
Teresa A. Sullivan
University of Texas at Austin
Marta Tienda
University of Wisconsin-Madison
This paper was prepared for presentation at the 1984 Annual Meetings of the
Population Association of America, Minneapolis, Minnesota. Institutional support
was provided by the College of Agricultural and Life Sciences of the University
of Wisconsin and the Department of Sociology, University of Texas at Austin.
April 1984
(To appear in Review of Public Data Use)Abstract
‘This paper describes and classifies the types of data now available in the
“system” of U.S. immigration statistics and illustrates through a selective
teview of recent studies how these sources can be combined into discrete
investigations of immigration, First we outline an heuristic framework for
classifying and discussing 1mmigration statistics. This framework
distinguishes between the production and use of official and unofficial
statistics. The remaining discussion addresses the diversity of studies
employing unofficial data sources, and selectively highlights points of
convergence in substantive findings, policy inferences and the strengths and
limitations of varying research approaches.INTEGRATION OF MULTIPLE DATA SOURCES IN IMMIGRANT STUDIES
The rapidly growing literature about the dimensions, causes and
consequences of immigration to the United States is impressive, not only for
its volume, but also for its diversity in substantive concerns, theoretical
orientations, research approaches and types of evidence presented. The latter
two aspects of this diversity--data sources and methodological approaches-~are
the central concerns of this paper. Rather than an attempt at yet another
comprehensive literature review that sorts the vast number of immigration
literature by data sources and methods, this paper has two objectives: first,
to describe and classify the types of data now available in the “system” of
U.S. immigration statistics; and second, to illustrate how these sources can
be combined into discrete investigations of immigration. This paper, then,
might be called a metastudy of immigration.
Readers should be forewarned that the idea of building a master file which
links a variety of data sets through matching of selected identifiers and/or
characteristics of respondents or places is not contemplated. Rather, our
goal is to identify data gaps in existing statistical and non-statistical data
sources so that future data collection efforts might benefit, and so that
future analysts might pursue multi
data and multi-method strategies to
validate specific substantive conclusions.The "System" of Immigration Statistics
The word statistics shares an etymological origin with the word
statistics were originally the data produced by and for the state. When the
Nineteenth Century preoccupation with national boundaries led to the
inspection of persons crossing those borders, the data we now call immigration
statistics came into existence. By today, a variety of data have been
analyzed to draw conclusions about immigration; the sources are by no means
limited to data authorized or collected by the state. The very proliferation
of data sources, collected under differing auspices, complicates the analysis
of immigration statistics. These varying data sources and the mechanisms for
collecting the data may be described, for our purposes, as the immigration
statistics “system.” As we will show, however, in many respects this “system”
is quite unsystematic.
A system of statistics may be analyzed in terms of their characteristic
production and subsequent use. Within the U.S. system, it is also helpful to
distinguish the “official” statistics from the “unofficial” statistics. These
two dimensions~-function and auspices--may be combined to produce Figure 1,
which is offered as an heuristic device for classifying and discussing
immigration statistics.
(Figure 1 About Here)
Data production may be conceptually separated from data analysis, although
the design of the data constrains the possible analyses. Data production
refers to the collection, compilation, coding and storage of basic data
concerning immigration, Data production is intentional, but the intentions of
the producer and those of the analyst do not necessarily coincide. Some data
collected by INS, especially those collected in response to Congressionalmandate, are produced specifically for analysis. Other data, often called
by-product data, have analytic uses beyond their intended purposes. For
example, INS data on budgeting and personnel are produced to aid the
administration of the agencys nevertheless, in the hands of the analyst they
may provide important information about enforcement or shifting ports of entry
among immigrants. The absence of some useful data (such as preference
category upon entry), the loss of data, or data incompatibility are among
current data production problems in immigration studies.
Data analysis refers to tabulating, comparing, modelling, or in other ways
manipulating data to draw substantive conclusions about immigration. Primary
analysis carries out the intentions for which the data were produced.
Secondary analysis uses data ostensibly collected for another purpose.
Administrative by-product data are nearly always subjected to secondary
analysis, but in addition secondary analysis of purposive data collection is
common (for example, secondary analyses of the U.S. Census or the Current
Population Survey).
Immigration data may be further classified by the auspices under which
they were collected. Official data are collected by a government agency.
“Unofficial” data includes those produced by not-for-profit or private sector
firms. There are two intermediate cases as well. The first is the use of
government funds to finance the collection of data by non-government
investigators. The General Social Survey, which has sometimes been used to
compare immigrants with the native-born, is one such data set. However,
because government grant programs usually disclaim responsibility for these
data sets, the data are perhaps best considered to be “unofficial.” A second
intermediate case is that of data collected by a foreign government or by a
quasi-governmental agency, such as the United Nations High Commission forRefugees. These data, while official in some contexts, are unofficial as used
within the United States. The two cross-cutting dimensions of
production/analysis and offictal/unofficial auspices provide four analytic
categories for understanding the studies of immigration that have been done
and the gaps remaining.
Cell A, production of official statistics, is notable for its
decentralization. Within the federal government a number of agencies produce
data on immigrants or refugees, either intentionally or as a by-product of
their other services. The information system is a federal one, with the
states providing at least some information on vital events occurring to
immigrants, refugee program use, bilingual education, and other topics that
vary by state.
The decentralized nature of this production system leads to many of the
frustrations of analysts. Because of the decentralized nature of the system,
data are not always comparable from one agency to another, nor even from one
time period to another. Census data permit one to identify the foreign-born,
but INS data distinguish among categorizing aliens. Some critical data, such
as the decennial census, are collected infrequently, while other data are
available on an annual, quarterly, or monthly basis. Further, because of the
relative rareness of immigrants and their great heterogeneity, many official
data sets that are representative of the population yield small case bases of
immigrants. These frustrations have encouraged the development and use of
unofficial statistics.
Cell B of the figure, the production of unoffical statistics, fits loosely
if at all into the official ”
ystem.” Included in this category are intensive
ethnographic studies of immigrants and immigrant sending and receiving
communities; local surveys and compilations of local data; reconstructions ofhistorical series from extant but unsystematized data; and many other
possibilities. Added to this unofficial aspect of the “system” are the data
of foreign consulates and embassies, foreign government agencies, and
international agencies, all of which are “official” in another context. Even
with all the supplementary unoffical data that are produced, however, existing
data sets may remain inadequate in terms of the coverage, validity, and
reliability of the data they contain, or they may be inadequate because they
are impossible to compare or to splice with other extant official or
unofficial data. To some extent, analysts have tried to solve these problems
with more creative uses of the existing data. Our discussion focuses on the
diversity of studies employing unofficial data sources, and selectively
highlights points of convergence in substantive findings, policy inferences
and the strengths and limitations of varying research approaches.
Cell C of the figure is the analysis of data from unoffictal sources.
This may be primary analysis, but because of the probibitive expense of
collecting unofficial data, secondary analysis is more common. The agenda for
unofficial analysis is set by current intellectual issues, policy debates,
theoretical developments in one or more of the social sciences, or simply the
curiosity of the investigator. Such studies, usually published in the
scholarly journals, monographic literature, ot popular press, may not be
cumulative because their audiences are specialized along political, class, or
disciplinary lines. To the layperson or the policy-naker, the resulting
debates and controversies, especially those centering on issues of data
adequacy, may seem partisan or merely arcane. Just as diverse motivations
underlie the collection of data by unofficial sources, so too the analysis of
data by unofficial analysts responds to many goals. A lack of consensus among,
the analysts may be a sign of healthy diversity ané not merely offactionalism. On the other hand, an effort such as this one to integrate
findings from different data sources is limited to the major substantive foci
of recent research.
The fourth cell of Figure 1, Cell D, is the analysis of official data.
Primary analyses may be guided by official or quasi-official agendas for the
types of statistical tests desired and the frequency with which they are to be
done. The U.S. Gensus Bureau is an example of @ statistical agency that not
only produces data, but also publishes analyses of these data, many at regular
intervals, others as occasional papers. Comparisons of official data sources
may occasionally be undertaken to provide statistical benchmarks. But
analyses of immigration data under official auspices have been relatively
rare. The intellectual division of labor has allocated this task to
individual researchers in the private sector.
Although the greatest volume of recent social science research about
immigration has been based on census-type surveys (especially the 1960 and
1970 Public Use Samples and the 1976 Survey of Income and Education), reliance
on such data greatly restricts the range of substantive issues that can be
addressed and the kinds of analytical approaches that can be pursued. Given
the restricted range of social and economic variables available on census
files, and the virtual absence of cultural indicators, it is not surprising
that most immigration studies that rely on census data focus on the
socioeconomic characteristics of the foreign-born population. The latter
usually differentiated by national origin and occasionally by year of arrival
cohorts. Census micro-data encourage the use of individuals or households as
units of analysis, but the use of places or some other aggregate unit can
portray macro dimensions of sociél phenomena. Multi-level analyses combining
person and place variables are less common, but not entirely absent from the
literature.Because of the extensive reliance of previous researchers on official
data, a discussion of how these data have been analyzed in the past and/or how
to link census data with various administrative and non-administrative data
would entail an entire paper. In the interest of adhering to the focus on
unofficial data sources, we limit our discussion to unofficial survey data and
to ethnographic data.
A Closer Look at Unofficial Data Sources
An important distinction must be made between statistical data (e.g.
social surveys) and non-statistical data sources (e.g. ethnographic and
archival or historical data). This section outlines the relative strengths of
each type of data. The potential complementarity among various types of
unofficial data sources requires some discussion of the analytical approaches
that have been employed to address specific substantive questions.
Understanding this complementarity is one key in integrating multiple data
sources.
Studies based on both ethnographic and survey data have generated useful
Ansights about immigration as a social phenomenon, although the relative
strengths, limitations, and types of inferences based on both kinds of data
clearly differ. Both survey and ethnographic data sources can be tailored to
specific substantive questions about immigration as a social process whose
causes and consequences extend from individuals to sending and receiving
communities, but the scope, depth and generalizability of the information
differs considerably. In the main, this results from the qualitative nature
of ethnographic data compared to the quantitative nature of survey data as
well as the manner of soliciting and “coding” information. Ethnographic databenefit from greater respondent flexibility, whereas survey data often
compromise the richness of open-ended responses to facilitate coding.
Ethnographic Data
Ethnographic data are, by definition, qualitative and richly textured
information, and their main strengths reside in their depth and comprehensive
coverage of an immigrant conmunity or a specific aspect of the inmigration
process. Ethnographic data tell us a great deal more about social processes
and interaction patterns that structure immigration flows than conventional
survey data, but the comprehensiveness of information produced depends on the
amount of time spent in the field and the number of field sites. Ethnographic
data are generally not portable among different research enalysts, therefore
they are not easily subjected to secondary analyses or verification except
through restudy by different investigators. However, the practice of
ethnographers to solicit information from multipie sources--participant
observation, key informants, and respondent interviews--serves as an internal
data consistency check and as a form of response validation.
In sun, the major drawbacks of ethnographic information are, the
nonportability of information among researchers, the limited generalizability
of the results beyond the population or locality of study, and the difficulty
of subjecting the data to rigorous hypothesis testing using aultivariate
statistical techniques. The major advantages afforded by ethnographic data
are the depth-perception they provide about the dimensions of human
interaction that structure geographic movements and ultimately the social and
economic characteristics of immigrants which are efficiently portrayed, but
often misinterpreted by survey data analysts.Survey Data
The strengths of unofficial survey data complement the weaknesses of
ethnographic data and vice versa. Primary strengths of unofficial survey data
reside in their pliability for defining the number and scope of topics covered
by an instrument, the definition of the universe, their degree of statistical
representativeness, and their anenability to rigorous empirical analysis.
None of these features are absolutes, for there are implied trade-offs in
scope, representativeness, breadth, population coverage and quantifiability
that are determined by cost and time considerations. Moreover, many aspects
of immigration are difficult to capture using conventional random sampling
survey techniques. The obvious example is the phenomenon of illegal
immigration, which defies the use of random sampling techniques. Non-random
sampling methods, including incidental or snowball samples, have been employed
with varying degrees of success, but the process of statistical inference is
seriously impaired. Other concerns not easily pursued with survey data (or
ethnographic data) are the macro-structural properties of the immigration
process, including the changing nature, direction, and composition of the
flows, but the official data sources are especially well suited to address
such matters.
Despite the many virtues of survey data for addressing questions about
immigration, these unofficial data sets present several drawbacks. Most
surveys are one-time a static snapshot of social and economic outcomes that
provide relatively limited information about how various outcomes occurred.
Causal relationships are imposed by the analyst through statistical inference
and substantive interpretation. Thus, mst cross-sectional surveys of
immigrants are limited in their ability to eddress questions about process, or
to establish causal relationships clearly.10
Notable exceptions are those few surveys that contain retrospective
modules about the timing of various events over individual life cycles,
inclu¢ing migration, employment, and fertility histories. The dating of
changes in social and demographic statuses permits greater precision in
estimating causal models, but inevitably the date of the interview introduces
censoring and truncation biases. Gurak's (1980) recent survey of Dominican
and Colombian immigrants in New York City, and Massey's (1981) study of four
Mexican “sending communities” are excellent recent examples of the uses of
retrospective modules in cross-sectional surveys of immigrants. Because both
surveys have been undertaken within the last two years, no specific findings
illustrating the use of retrospective life history modules for the study of
immigration as a social process have been released. ‘Two recent examples are
papers by Mullan (1984) and Kritz and Gurak (1984) which are based on these
surveys.
An alternative to the use of retrospective modules in sample surveys for
incorporating a temporal dimension in cross-sectional surveys involves
re-interviewing a sample of respondents at several points in time. This
approach 1s preferable to that of repeated cross-sections for making direct
inferences about causal sequences leading to specific outcomes. However, cost
factors have greatly inhibited individual researchers from undertaking truely
longitudinal surveys of immigrants. Moreover, attempting a longitudinal
survey of immigrants, a population which by definition has a high propensity
to move, increases the extent of sample attrition and this, in turn, can
impair the representativeness of the sample, assuming it was random in the
firet place. Complexity and cost factors aside, it is noteworthy that
currently there does not exist a nationally representative longitudinal study
of recent immigrants. In most longitudinal studies of the general populationuw
the number of immigrants included in the sample is too small to permit
separate analysis, especially if separate analyses of nationality groups are
contemplated.
One of the most well-known recent longitudinal surveys of immigrants was
conducted by Alejandro Portes (1972) between 1973 and 1979. His universe was
defined by an incidental sample of legal Cuban and Mexican immigrants who
entered the United States in 1973 at specified ports, and who were
re-interviewed in 1976 and 1979, Although this universe is less than optimum
from the standpoint of statistical randomness and representativeness of the
total population of Cuban and Mexican immigrants, the longitudinal dimension
of Portes's survey has allowed novel analytical approaches and insights about
immigrant incorporation experiences. Moreover, early panels informed
subsequent ones, so that by the third panel Portes solicited more information
about the structural contexts in which specific individual outcomes occurred.
As a result, the last wave of data permitted more extensive multi-level data
analysis than the earlier waves.
Obviously, the strategy used to define a universe and devise a sampling
scheme may limit the usefulness of the survey for studying immigrants. (For
example, the General Social Surveys include only the English-speaking
population over age 18 in their universe), Leaving aside the special problem
of sampling the undocumented immigrant population, there are no obvious
guidelines about how to conduct a sample of immigrants to study the
determinants of migration and the consequences of immigration for the sending
and receiving communities. With few exceptions, most primary sample surveys
about immigrants have defined the universe on the basis of those who actually
move across international boundaries and settle in a specific locality
(locality defined universes) or who cross at a specific time (time-defined
universes), or both.12
Such strategies for delimiting a universe are appropriate for addressing
questions about the experiences of immigrants in a host society, but these
samples easily limit analysts' interpretations of the causes and consequences
of international migration in at least two ways that bear importantly on
policy issues. First, by excluding those who decide not to emigrate, studies
based on samples of individuals who have migrated across international
boundaries distort our understanding of the determinants of immigration and
lead to potentially erroneous conclusions about the nature of migrant
selectivity. Second, time and locality defined universes, especially the
latter, ignore the fact that an unknown number of immigrants may have returned
to their place of origin, or moved on to another destination. This latter
problem can be partly resolved by inquiring about past migration history,
intended moves, and the existence of friends and relatives in other
localities, but it introduces selection problems of unknown magnitude in the
statistical analysis of the survey date, and may ultimately distort
conclusions about what mix of individual, familial, and iccational features
combine to structure aggregate flows.
Depending on the craftiness of the analysts and the universe used for a
sampling frame, which can range from national to local, survey data can be
manipulated to create hierarchical files representing different structural
contexts (based on characteristics of the workplace, the neighborhood, the
labor market, etc.), but such multi-level research approaches, although
promising, have only recently begun to emerge. Nevertheless, the insights
generated by such approaches have implications for the undertaking of further
data collection.
To summarize, the main advantages of survey data reside in their
generalizability, the scope of topics covered, their amenability to rigorous13
hypothesis testing, and the high degree of portability among researchers. In
practice, however, the access to and distribution of specialized surveys about
immigrants is not extensive, as there is no central clearinghouse which
receives and distributes primary or secondary data sources containing
information about immigrants to interested researchers. The generalizability,
substantive content, and amenability of unofficial surveys to rigorous
secondary analyses of immigration issues varies greatly as a function of the
motivation for the original research. Although cross-section surveys can be
designed to handle temporality of events through retrospective techniques, or
through repeated cross-sectional approaches, each strategy poses different
problens for the analyst, of which cost, sample attrition, and recall problems
may be the most substantial from the standpoint of data production, and
truncation, censoring, and selection bias are the more important from the
standpoint of data analysis.
Substantive Complementarities Among Unofficial Data Sources
Although there are a number of possible combinations of data types, three
particular combinations have led to fruitful studies of U.S. immigration. We
have termed these three combinations the multi-level studies, the multi-method
studies, and the multi-dataset validations. The first category uses
combinations of data aggregated at different levels to establish a finding.
For example, micro-data or household data might be used to enhance or specify
conclusions based on aggregated data. The study of Borjas(1983) , discussed
below, is one example. Such a study may be conducted entirely within one cell
of Figure 1.14
The second combination, the multi-method study, combines fundamentally
different types of data to bolster a finding or to provide quasi-replication.
An example would be the use of ethnographic data to confirm findings of a
sample survey. Such studies frequently require the combination of cells in
Figure 1, particularly the combination of cells A and B or B and D.
The third category, the multiple dataset category, involves the
cross-validation of a finding using a different dataset that measures the same
population or variable of interest. For example, U.S. estimates of
immigration from a country can be compared with that country's estimated
emigration to the United States. Such studies typically require combinations
of cells A and B of Figure 1, although other combinations might also be
pursued.
Given the diversity of substantive concerns in the empirical literature,
one way to manage such an illustration is to select a few research and policy
issues, such as labor displacement, labor recruitment, and other reception
factors encountered by recent immigrants. Both are central concerns for
understanding the societal consequences of contemporary immigration patterns
and the integration experiences of recent immigrants, and they represent
research domains where the official data sources, including nationally
representative census survey data, have been of limited value. The following
section illustrates the complementarity of ethnographic and survey information
for validating conclusions about specific aspects of these substantive areas.
Labor Market Displacement
Much of the current debate about the need for reform in the immigration
legislation centers on the alleged negative labor market impacts of a large
influx of foreign workers, particularly during times of economic recession.15
The major research issue at the crux of the current policy debate is whether
immigrants displace native born workers in the labor market, and, by
extension, whether the unemployment rates observed among domestic workers
would be lower in the absence of immigration. Obviously, this assertion
assumes perfect, or near perfect substitutability among native and immigrant
workers, a condition that cannot be established by intuition or assertion.
The counter-argument to the displacement thesis is that immigrants expand the
Jabor market; a weaker form of this argument is that immigrants take those
Jobs and accept wage rates domestic workers would not consider. Either way,
the net aggregate impact of immigration is benign.
While 4t is very likely that immigration policy decisions will continue to
be made without much regard for recent research findings, it 1s important to
note that our limited understanding of the articulation of native and
foreign-born workers in the U.S. labor markets begins with the lack of
appropriate data to study the problem of displacement and complementarity of
dmmigrant and domestic labor. Nevertheless, the centrality of this issue for
the policy debate has encouraged researchers to address the question using
census survey data in innovative ways and conducting intensive case studies of
immigrant firms. At a very general level, the basic findings of these studies
tend to corroborate each other in showing that the extent of displacement
between domestic and immigrant labor is very small, if it exists at all,
partly because these groups of workers do not find employment in the same
kinds of enterprises, and partly because of differing skill levels between
native- and foreign-born workers. However this generalization requires much
more specificity for us to suggest that this is an example of
quasi-replication involving multiple data sets and methodologies. Further
elaboration clarifies why this is the case.16
Borjas’ (1983; 1984) econometric analyses of several nationally
representative census surveys show no evidence that immigrants displace native
workers. fis various analyses define labor markets using SMSA boundaries and,
when sample sizes permit, these are replicated within broad industry
categories (e.g-, manufacturing, construction, services, etc.) to bolster his
basic finding of the extent of complementarity among native and immigrant
workers, Two features of his data sources and methodology are noteworthy for
discussion. One is his use of a multi-level methodology using individuals as
units of analysis. Borjas specifies wage functions as determined by
individual, group, and industry characteristics. This methodology required
merging of information from secondary sources and also appending aggregated
census data on individual records. Second, the level of aggregation he used
to establish the complementarity of domestic and foreign workers was rather
coarse.
Even 1f there is general agreement that the labor market impacts of
immigrants are specific to labor markets where large volumes of foreign
workers settle, the ethnographic evidence shows that even within detailed
industries—-the garment industry for example—-there is substantial segregation
between native and foreign workers across specific firms. The pervasiveness
and extent of inter-firm segregation between native and foreign-born workers
4s consistent with results showing limited, if any, labor market
displacement. However, the industry and occupational segregation of workers
does not preclude the existence of competition and substitution. For this
reason, more ethnographic studies, and firm-level studies in particular, are
needed to pursue the question of complementarity and substitution between
native and foreign workers. Moreover, some intermediate level studies which
can illustrate the social processes that result in inter-firm segregation ofwv
domestic and immigrant workers, and others which simulate the wage structures
that could prevail in the absence of low-skill foreign-born workers would help
to reconcile better the generally consistent, but specifically inconclusive
inferences about the nature and extent of articulation of the domestic and
foreign born workforces.
Labor Recruitment
Roger Waldinger's (1984) ethnographic study of the work patterns of
Colombian and Dominican immigrant women residing in New York City revealed
that labor recruitment in predominantly immigrant firms occurs within family,
friend, and kinship networks; this information is not new. The insight his
study affords is that an apparently stable ethnic labor force in small garment
firms can mask extensive turnover in the job incumbents. This results as new
friends and relatives assume the jobs vacated by returning migrants.
To elaborate a bit further, consider the well documented fact that recent
immigrants with limited skills and poor English proficiency usually find
employment in the secondary sector, often in industries employing substantial
numbers of other like minorities. What might one infer about labor market
Ancorporation experiences of immigrants from such facts? Survey information
about the ethnic composition of workplaces, as Portes obtained in his 6-year
longitudinal survey of Mexican and Cuban immigrants, help document the extent
of ghettoization of foreign born workers of a given national origin, but such
data seldon aid in establishing ties or antagonisms among the workers
themselves, nor do they comment on the operation of recruitment channels for
the workers involved. It is striking to note that despite the belief that
common ancestry (i.e., Mexican origin) provides a ready base for elaboration
of ethnic and class solidarity, Browning and Rodriquez's (1984) ethnographic18
study of indocumentados in Austin has demonstrated why this image of the
Chicano-Mexicano interface is incorrect. He has elaborated in some detail the
existence and nature of antagonism between Chicano and Mexicano residents in
the workplace as well as in the social and cultural spheres of life.
Additional Reception Factors
The questions of which and how reception factors structure the social and
economic integration of recent immigrants might further help clarify processes
of industry and occupational segregation, as well as help delineate the nature
of tensions between domestic and foreign-born workers of common or different
national origins. Besides the obvious need for ethnographic studies of social.
networks among immigrants and especially the existence of recruitment
strategies based on kinship ties, an intriguing line of research that could
shed new light on the old questions of the manifestations and consequences of
ethnic antagonism might derive from the policy sensitive and data illusive
issue of illegal migration. For example, ethnographic work by Browning and
Rodriquez (1984) has begun to illustrate the extent of antagonism between
Mexican immigrants and Chicanos in the Austin labor market, but it is unclear
whether similar patterns are found among the border towns or in agricultural
areas.
Another important aspect of reception factors concerns the under-studied
phenomenon of immigrant enclaves. It 1s unclear whether the formation and
maintenance of economic enclaves by Cubans in Miami and by Koreans in Los
Angeles actually serve to shield new immigrants from the corrosive forces of a
competitive labor market as Portes and his associates (Portes and Bach, 1980;
Wilson and Portes, 1980), have suggested, or whether the existence of economic
enclaves, while affording advantages for new entrants in the short run,19
ultimately serve to verify divisions between native and foreign-born workers
and thereby mask real labor market displacement. Neither the ethnographic nor
survey evidence is clear on this question, for there 1s a great deal of
dispersion anong studies and sources in defining enclaves, and other forms of
positive or negative reception factors, including the recruitment networks
that make ethnic enclaves viable economic enterprises. This, then, suggests a
major research and data gap which is critical to advance our understanding of
how reception factors influence labor allocation patterns which, in turn, may
ameliorate the extent of labor market displacement resulting from the
continued influx of innigrant workers.
Despite the appeal of undertaking multiple-source immigration studies for
validating or clarifying ambiguous or unexpected findings, researchers
encounter numerous obstacles in practice. Many of the obstacles to
multiple-source studies are traceable directly to the tension pointe implicit
within Figure 1. A brief catalogue of obstacles would include definitions of
the universe, legal definitions of the terms used, conceptual variations,
differential quality of coverage, and incomparabilities in time. Thus, some
rather obvious studies have been impossible to do. We reconsider this point
in our concluding section where we specify future data production and analysis
strategies.
Structural Complementarities and Tension Points
of the Innigration Statistics "System
The interplay of several levels of government and of both official and
unofficial parties generates within the current immigration statistics system
@ series of checks and balances that bears some resemblance to those
underlying the design of the U.S. government. ‘here are three bases for these
checks and balances: differential abilities, differences in subject matter,
and differences in analytic techniques.20
The official sources are bolstered by the coercive power and scope of the
atate, which can compel cooperation with the U.S. census and with the INS.
Federal funding maintains high quality procedures and the continuity needed to
produce time series. These advantages can be balanced against the relative
freedom of the unofficial researcher to use untried or experimental
procedures, and against the possibility that a non-governsental researcher can
obtain better rapport and cooperation from at least some populations.
The subject matter of the official data sources is limited by the
centrality of the variable to the government mission. Certain subjects are
not investigated by government agencies (e.g-, one’s religion) while other
subjects may not be reported honestly to the government (e.g., illegal
activities). Aside from regulations protecting human subjects, unofficial
investigators enjoy wide latitude in the sorts of information they seek. This
provides a means of balancing the official data with additional unofficial
data or replacement unofficial data.
With regard to methods, official data are usually comprehensive in scope
and amenable to quantitative amlysis, Unofficial data, although often more
restricted in scope, may have a richness and extensiveness of detail that is
unavailable in official reports, as well as providing more qualitative and
attitudinal information,
The major disadvantage of this statistical “system” is that official and
unofficial research agendas may be sufficiently different that large gaps
remain in our understanding about immigration. This may occur because needed
data have not been produced, or because the data are not available or
accessible for analysis, Aside from these structural problems, there are also
imtellectual gaps that arise through the failure to exploit the extant data
adequately.21
One example will suffice to demonstrate the tension points in the current
system. Demographic accounting is not an explicit function of official
statistice, but the constitutional mandate to conduct the census and the
subsequent Congressional requirements for demographic data of various sorte
may be construed as an obligation to keep demographic accounts. In the basic
demographic equation, the population at time two is a function of the
population at time one plus natural increase plus net migration. This
equation requires estimates of births, deaths, immigration, and emigration.
Birth and death data are collected by the states; immigration data are
provided by INS, and emigration estimates are made by the U.S. Census Bureau.
Differences in definitions, timing, and coverage contribute to the errors in
computing these variables, but the need to coordinate all these different data
sources is in itself sufficient to introduce error. Many additional issues of
substance and method may immediately be raised, including accounting for
illegal immigration, return migration, the mortality of immigrants, and the
fertility of immigrants. Each of these issues, in turn, raises a host of
subissues. For example, to what extent do normal and reverse selectivity
among immigrants contribute to their differential mortality? How can we
adjust our mortality estimates to account for person-years of exposure to
health conditions in the United States?
Conclusion
The present “systen” of immigration statistics is complex and
decentrailzed. The nature of immigration is a complex process and the
decentralization of the official agencies and individual research agendas,
seem to be irreducible aspects of the system. Improvements in the system are
of course possible, but the structural issues seem to be more or less22
permanent. On the other hand, the variety in the data sources offer more
opportunities for analysts to compare and contrast data, verifying findings
with several data sets, or combine information innovatively. Thus, while the
integration of the system may be impossible, the integration of findings is
quite possible,Borjas, George
1983. “The Substitutability of Black, Hispanic and White Labor”, Economic
Enguiry, Vol 21 (January):93~106
1984 “The Impact of Immigrants of the Earnings of the Native-Born:
Evidence from the 1970 Census.” Paper presented at the meeting of
the National Council for Enployment Policy.
Browning, Harley and Nestor Rodriquez
1984 "Migration of Mexican Indocumentados as a Settlement Process.
Forthcoming in Borjas and Tienda (eds.), Hispanics in the U.S.
Economy. New York: Academic Press.
Gurak, Douglas
1980 Hispanic Immigrants in New Yor!
Research Proposal funded by NIMH
Work, Settlement and Adjustment.
Keite, Mary M. and Douglas Gurak
1984 Kinship Networks and the Settlement Process: Dominican aud
Colombian Immigrants in New York City. Paper presented at 1984
Meetings of PAA, Minneapolis.
Massey, Douglas
1981 U.S. Mexican Migration: The Settlement Process. Research Proposal
funded by NICHD,
Mullan, Brendon
1984 Occupational Mobiliby of Mexican Migrante to the United States.
Paper presented at the 1984 Meetings of PAA, Minneapolis.
Portes, Alejandro
1972 Assimilation of Latin American Minorities in the U.S. Research
proposal funded by the National Science Foundation
Portes, Alejandro and Robert L. Bach
1980 “Inmigrant Earnings: Cuban and Mexican Immigrants in the U.S.
International Migration Review, 14(3):315-337,
Waldinger, Roger
1984 “Immigration and Industrial Change in the New York Apparel.
Industry.” Forthcoming in Borjas and Tienda (eds.), Hispanics in
the U.S. Economy. New York: Academic Press.
Wilson, Kenneth and Alejandro Portes
1880 “Immigrant Enclaves: An Analysis of the Labor Market Experiences on
Cubans in Mian,” American Journal of Soctology, 86(2):295~319.Figure 1
HEURISTIC CLASSIFICATION OF IMMIGRATION DATA
Auspices of Data Data
Collection/Analysis Product ion Analysis
official A D
Unofficial