Immigration Statistics: Information Gaps, Quality Issues Limit Utility of
Federal Data to Policymakers (Chapter Report, 07/31/98, GAO/GGD-98-164).

Pursuant to a congressional request, GAO identified: (1) policy-related
information needs for immigration statistics; (2) federal statistics
(and information gaps) on the full range of demographic concepts
relevant to immigration policy decisions, including what is known about
the quality of those statistics; and (3) strategies for improving
statistics.

GAO noted that: (1) Congress periodically makes decisions about numerous
immigration policies; (2) thus, informed decisionmaking by congressional
committees and members of Congress as well as interested members of the
general public, requires information on immigration flow, by legal
status; (3) Congress also decides on the eligibility of the foreign-born
for government benefits and services--with different benefits typically
allowed or restricted for different categories of the foreign-born
population; (4) GAO identified 33 discrete categories of demographic
information that could be relevant to congressional decisionmaking; (5)
information on immigration flow is reported in annual Immigration and
Naturalization Service (INS) Statistical Yearbooks; (6) statistics on
demographic categories other than flow are reported in a more scattered
fashion; indeed, a variety of INS and Bureau of the Census publications,
including the INS Web page, must be accessed in order to retrieve basic
information; (7) INS records that are maintained for administrative
purposes are the basis for most federal statistics on flow; (8) these
statistics describe the number of new legal permanent residents, new
refugees and asylees, and new naturalized citizens; (9) as reported in
the INS Yearbook, however, these statistics are limited by conceptual
problems and confused reporting, undercounts, and information gaps; (10)
the number of new asylees--persons granted asylum--and the number of
persons granted citizenship are undercounted in the Yearbook tallies
because the data omit certain groups of persons; (11) statistics for
other demographic categories are not available; (12) while Census
provides some information on the size of the resident foreign-born
population, annual net change in size, and emigration, Census has not
quantitatively evaluated these data with respect to coverage, accuracy
of reported place of birth, or nonresponse rates; (13) there are no
separate Census data on legal status because none of the surveys ask
questions about legal status; (14) INS has made efforts to fill
information gaps for some legal statuses by using the limited data that
are available and creating assumption-based models; (15) GAO attempted
to identify existing strategies or develop new ones to improve
immigration statistics; (16) GAO devised a new method for collecting
survey data on the legal status of foreign-born respondents; and (17)
GAO also identified strategies for evaluating survey data on the
foreign-born.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-98-164
     TITLE:  Immigration Statistics: Information Gaps, Quality Issues 
             Limit Utility of Federal Data to Policymakers
      DATE:  07/31/98
   SUBJECT:  Population statistics
             Immigration information systems
             Data collection
             Data integrity
             Aliens
             Immigrants
             Census
             Immigration or emigration
IDENTIFIER:  INS Alien Address Report Program
             Census Bureau Current Population Survey
             Census Bureau Post Enumeration Survey
             
******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.  Major          **
** divisions and subdivisions of the text, such as Chapters,    **
** Sections, and Appendixes, are identified by double and       **
** single lines.  The numbers on the right end of these lines   **
** indicate the position of each of the subsections in the      **
** document outline.  These numbers do NOT correspond with the  **
** page numbers of the printed product.                         **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************


Cover
================================================================ COVER


Report to Congressional Requesters

July 1998

IMMIGRATION STATISTICS -
INFORMATION GAPS, QUALITY ISSUES
LIMIT UTILITY OF FEDERAL DATA TO
POLICYMAKERS

GAO/GGD-98-164

Quality of Immigration Statistics

(966700)


Abbreviations
=============================================================== ABBREV

  CLAIMS - Computer Linked Application Information Management System
  CPS - Current Population Survey
  DOL - Department of Labor
  EOIR - Executive Office of Immigration Review
  EWI - Entry without inspection
  GPO - Government Printing Office
  INA - Immigration and Nationality Act
  INS - Immigration and Naturalization Service
  IRCA - Immigration Reform and Control Act
  LPR - Legal permanent resident
  NAWS - National Agricultural Workers Survey
  NCES - National Center for Education Statistics
  PES - Post Enumeration Survey
  RAPS - Refugee, Asylum, and Parolee System

Letter
=============================================================== LETTER


B-277490

July 31, 1998

The Honorable Stephen Horn, Chairman
The Honorable Dennis J.  Kucinich
Ranking Minority Member
Subcommittee on Government
 Management, Information, and Technology
Committee on Government Reform and Oversight
House of Representatives

The Honorable Carolyn B.  Maloney
House of Representatives

Policy debates on immigration have been important in recent years and
will probably continue to be high on the national agenda for some
time to come.  Relevant demographic, statistical information can
inform policy decisions, but concerns continue to be expressed about
an apparent lack of such information in the area of immigration. 

We are pleased to send you this report, which responds to your
request that we study policy-related needs for demographic,
statistical information; report on federal statistics that address
those needs, including what is known about the quality of those
statistics; identify data gaps; and suggest strategies for improving
statistics. 

As agreed with your offices, unless you publicly announce the
contents of this report earlier, we plan no further distribution of
it until 30 days from the date of this letter.  At that time, we will
send copies to the Chairmen and Ranking Minority Members of other
committees with jurisdiction over immigration issues; the
Commissioner, INS; and Director, Bureau of the Census.  We will also
make copies available to others upon request. 

If you or your staff have any questions about this report, please
call me or Judith A.  Droitcour, Assistant Director, who served as
project director for this work, at (202) 512-7997.  Other major
contributors are listed in appendix V. 

Susan S.  Westin
Associate Director
Advanced Studies and Evaluation
 Methodology Group


EXECUTIVE SUMMARY
============================================================ Chapter 0


   PURPOSE
---------------------------------------------------------- Chapter 0:1

Immigration issues are high on the national policy agenda.  But there
have been repeated indications that statistics that could inform key
debates are lacking, misleading, or otherwise inadequate.  In
response to a request from the House Government Reform and Oversight
Subcommittee on Government Management, Information, and Technology,
this report (1) identifies policy-related information needs for
immigration statistics; (2) identifies federal statistics (and
information gaps) on the full range of demographic concepts relevant
to immigration policy decisions and determines what is known about
the quality of these statistics; and (3) identifies strategies for
improving immigration statistics. 


   BACKGROUND
---------------------------------------------------------- Chapter 0:2

Published immigration statistics include information on several
demographic concepts, such as immigration flow (i.e., the number of
persons who come to reside in the United States each year or the
number who transition from one legal status to another).  Other
immigration statistics indicate the size of the resident foreign-born
population, net change in size, and emigration (i.e., the estimated
number of foreign-born residents who leave the United States to live
in another country).  Statistics on the legal status of foreign-born
residents have also been published. 

Most federal statistical information on the foreign-born is provided
by two agencies:  the Immigration and Naturalization Service (INS)
and the Bureau of the Census.  INS provides information relevant to
flow.  INS and the Census Bureau each provide some information on the
size of the resident foreign-born population and on annual change in
size.  The Census Bureau provides an estimate of emigration. 


   RESULTS IN BRIEF
---------------------------------------------------------- Chapter 0:3

Congress periodically makes decisions about numerous immigration
policies, such as whether to decrease numerical limits for legal
immigrants or increase them for temporary workers.  Thus, informed
decision-making, by congressional committees and Members of Congress
as well as interested members of the general public, requires
information on immigration flow, by legal status.  Congress also
decides on the eligibility of the foreign-born for government
benefits and services--with different benefits typically allowed or
restricted for different categories of the foreign-born population,
such as legal permanent residents (LPRs) or illegal immigrants.  GAO
identified 33 discrete categories of demographic information that
could be relevant to congressional decision-making. 

Information on immigration flow is reported in annual INS Statistical
Yearbooks.  Statistics on demographic categories other than flow are
reported in a more scattered fashion; indeed, a variety of INS and
Bureau of the Census publications, including the INS Web page, must
be accessed in order to retrieve basic information. 

INS records that are maintained for administrative purposes are the
basis for most federal statistics on flow.  These statistics describe
the number of new legal permanent residents (persons with new
green-card status), new refugees and asylees, and new naturalized
citizens.  As reported in the INS Yearbook, however, these statistics
are limited by conceptual problems and confused reporting,
undercounts, and information gaps.  For example, data on annual
trends in the number of new green cards authorized, although
highlighted in the INS Yearbook, are significantly affected by INS'
administrative problems in processing green-card applicants and thus
may not accurately reflect flow over time.  The number of new
asylees-- persons granted asylum--and the number of persons granted
citizenship are undercounted in the Yearbook tallies because the data
omit certain groups of persons, such as those granted asylum on
appeal or minor children of new citizens.  Statistics for other
demographic categories, such as the total number of foreign-born
persons who take up residence in the United States each year, are not
available. 

While the Bureau of the Census provides some information on the size
of the resident foreign-born population, annual net change in size,
and emigration, Census has not quantitatively evaluated these data
with respect to coverage, accuracy of reported place of birth, or
nonresponse rates.  Moreover, there are no separate Census data on
legal status because none of the surveys ask questions about legal
status.  Such questions are very sensitive, and negative reactions to
them could affect the accuracy of responses to other questions on the
survey. 

INS has made efforts to fill information gaps for some legal statuses
by using the limited data that are available and creating
assumption-based models.  The resulting estimates are necessarily
uncertain because assumptions and judgments are substituted for data. 

Because of the above problems, GAO attempted to identify existing
strategies or develop new ones to improve immigration statistics.  In
particular, GAO devised a new method for collecting survey data on
the legal status of foreign-born respondents.  The "three-card
method" asks questions that are less sensitive than a direct question
about legal status, and it ensures privacy of response.  Yet this
method allows statistically unbiased survey estimates for all major
legal statuses.  GAO conducted a preliminary qualitative test of the
new method and concluded that this method appears to show promise and
to deserve further testing and development.  GAO also identified
strategies for evaluating survey data on the foreign-born. 


   PRINCIPAL FINDINGS
---------------------------------------------------------- Chapter 0:4


      POLICY-RELATED INFORMATION
      NEEDS INCLUDE DATA ON FLOW
      AND OTHER MAJOR DEMOGRAPHIC
      CONCEPTS
-------------------------------------------------------- Chapter 0:4.1

Informed decision-making on immigration issues requires data on

  -- immigration flow, including separate information on flow into
     the United States and transitions to new legal statuses,

  -- size of the foreign-born population,

  -- net change in size, and

  -- emigration. 

Because almost all immigration laws involve specific legal statuses,
it is crucial that information on each major demographic concept be
reported separately for each of the major legal statuses (legal
permanent residents, refugees and asylees, persons here with a
temporary visa, illegal immigrants, and naturalized citizens).  All
together, there are 33 relevant demographic categories.  GAO
summarized these categories and their interrelationships in an
information typology that helps clarify (1) the differences between
demographic concepts and (2) which statistics can--and which
cannot--be validly compared. 

Most of the information specified in GAO's typology is required,
either generally or specifically, by existing laws or has been
requested by congressional committees.  Virtually all of it is
directly or indirectly relevant to recurrent congressional
activities, such as revising numerical limits for visas issued to
immigrants or temporary workers. 


      INS STATISTICS ON
      IMMIGRATION FLOW ARE LIMITED
-------------------------------------------------------- Chapter 0:4.2

The INS Yearbooks present a variety of data on immigration flow,
including counts of new LPRs, new refugees and asylees, and new U.S. 
citizens--all based on administrative records.  But as presented,
these data do not convey a clear and valid picture of immigration
flow that is relevant to policy information needs.  Rather, there are
instances of conceptual problems and confused reporting, as well as
undercounts and information gaps. 

Conceptual problems and confused reporting are evident in the
introductions to the INS 1995 and 1996 Yearbooks, which alternatively
highlight a "decrease" (1995) and a "rise" (1996) in immigration to
the United States.  These annual trends mix data on two types of
flow:  (1) flow of new LPRs into the United States and (2)
transitions by persons already living here into LPR status.  These
trends in "combined-flow" can reflect

  -- a change in the number of persons coming into the United States,

  -- an administrative logjam (slowdown) or speed-up in INS' issuing
     of green cards to foreign-born persons already living here, or

  -- some combination of the two. 

Mixing two different forms of flow is not a trivial problem because,
according to the Yearbook, the majority of new green-card holders
were already residing here--often either illegally or on temporary
visas--and because, in recent years, INS has experienced problems in
processing green cards.  (The agency is struggling to catch up with
backed-up applications, which numbered approximately 775,000 as of
April 1998.) Thus, the trends that INS highlights are very difficult
to interpret, but readers of the Yearbook introduction are not
alerted to this--unless they turn to the body of the report. 

Another instance of a conceptual problem exacerbated by confused
reporting occurs when the INS Yearbook presents information relevant
to flow for legal immigration (i.e., the number of new green cards
authorized in the past year), but not for illegal immigration. 
Instead, the Yearbook presents information on net change in the size
of the illegal population.  For a valid comparison of data on legal
flow to data on illegals, comparable categories must be used; that
is, legal flow should be compared to illegal flow--not to net change
in the size of the illegal population.  Because the INS Yearbook does
not point out the difference between flow and net change, there is a
potential for a reader to make an invalid comparison.  Net change in
the illegal population is likely to be considerably smaller than
illegal flow because the net change statistic subtracts out deaths,
emigration, and legalizations for all illegal immigrants, regardless
of how many years they have resided here.  A likely consequence is
that--for readers of the INS Yearbook--the flow of illegal immigrants
is understated relative to the flow of legal immigrants. 

The 1996 INS Yearbook also presents figures that undercount two other
categories of flow: 

  -- Persons granted asylum are undercounted by about a third or
     more, based on information that GAO obtained from computer
     systems maintained by INS, the Department of State, and the
     Department of Justice's Executive Office of Immigration Review
     (EOIR).  Groups omitted from the Yearbook tally include (1)
     persons who were granted asylum on appeal and (2) trailing
     relatives of persons granted asylum.  (These trailing relatives
     who "followed to join" a principal asylee were not included in
     that asylee's original application--often because they were not
     in the United States at the time.)

  -- The number of new naturalizations is undercounted by an unknown
     amount.  INS administrative records on naturalizations do not
     include most minor children who naturalize along with their
     parents. 

There are also information gaps for certain categories of flow.  For
example, no statistic is reported for the number of new residents who
entered the United States with temporary (e.g., student, worker)
visas.  This number cannot be tallied accurately because INS
record-keeping systems are not designed to identify reentries.  Thus,
a foreign student residing here for 4 years who visits his home
country twice a year would be tallied as eight separate 6-month
visits by up to eight persons, rather than as a single individual's
4-year stay. 


      FEDERAL STATISTICS ON OTHER
      DEMOGRAPHIC CONCEPTS ARE
      MARKED BY GAPS AND UNCERTAIN
      ESTIMATES
-------------------------------------------------------- Chapter 0:4.3

Information on the size of the foreign-born population, change in
size, and emigration is limited by important data gaps.  The data
that do exist have not been evaluated, despite important questions
about their quality. 

The Census Bureau provides census and survey data on size and net
change in size for the total foreign-born population and for
naturalized citizens.  The Bureau also estimates emigration by
comparing 1980 and 1990 census data.  Analysts have raised a variety
of questions about possible problems with data on the foreign-born,
such as undercoverage (because, e.g., illegal immigrants may
deliberately avoid enumeration).  Yet Census has not conducted a
quantitative evaluation of its census or survey data on the
foreign-born. 

The Census Bureau's estimate of emigration represents one instance
where census undercoverage of the foreign-born clearly limits the
utility of the data.  The uncertainty of the emigration estimate is
important because information on emigration is needed to balance
information on flow--and also because estimates of emigration figure
importantly in INS' indirect estimates of population size for
illegals and legal permanent residents.  There are no direct data on
emigration.  The Census Bureau attempts to fill this gap through a
method of examining change in the size of various arrival groups
across two points in time.  For example, the 1970-79 arrival group
(i.e., foreign-born residents who arrived here between 1970 and 1979)
was counted in the 1980 and again in the 1990 decennial census. 
Emigration for that group is estimated according to the amount it
dwindled between 1980 and 1990.  But the data for certain pre-1980
arrival groups (e.g., Mexican and Salvadoran arrival groups) showed
growth between 1980 and 1990 rather than dwindling--a logical
impossibility since all had arrived before 1980.  This pattern of
apparent growth, which could be explained by differential census
coverage (i.e., lower coverage of Mexican and Salvadoran residents in
1980 than in 1990), raises questions about the census counts that
underlie the emigration calculation. 

Data gaps occur for the number of residents in specific legal
statuses--from LPRs to illegal immigrants--and for net change in the
sizes of these resident population groups.  Direct questions on legal
status are very sensitive and, according to the Census Bureau, have
not been asked in the census or in any survey that it has conducted. 
The gaps for the population sizes of various legal statuses cannot be
easily filled by using INS records, as these are limited to flow.  To
fill data gaps for the population sizes of certain legal statuses,
INS has developed indirect estimates of the number of illegal
immigrants and legal permanent residents.  These estimates address
key gaps and thus would seem to represent a step forward in providing
information, but multiple sources of uncertainty remain--owing to the
lack of direct empirical data and the need for major assumptions
that, in some cases, are not fully supported. 


      STRATEGIES FOR IMPROVEMENT
-------------------------------------------------------- Chapter 0:4.4

INS is working toward correcting some of the problems in flow
statistics discussed above.  For example, a system that tracks
foreign students over time (and avoids duplicate counts when
reentries occur) is being pilot-tested at INS.  This system is based
on assigning each foreign student a unique high-tech identification
card, which he or she must use for all entries to and exits from the
United States.  INS is also working on an indirect estimate of
illegal flow.  (INS has worked with Census to obtain some of the
necessary data for this estimate.) But INS' current efforts do not
address all the problems identified above; for example, the Yearbook
tally of persons granted asylum does not include trailing relatives
of principal asylees. 

Various strategies for evaluating the quality of census and survey
data on the foreign-born would involve joint work on the part of INS
and the Bureau of the Census.  For example, coverage problems might
be assessed by matching INS administrative data and census records. 

In addition, testing and developing the "three-card method"--which
GAO devised for asking less sensitive questions about legal status in
surveys--might help fill important data gaps in the future.  That is,
if successfully developed and implemented, the three-card method
would provide survey data on the number of legal permanent residents
residing in the United States, the number of refugees and asylees
(without green cards) residing here, and so on; an indirect estimate
of the number of persons residing here illegally could also be
obtained.  By using the method in a series of surveys, net change
could also be tracked.  GAO's preliminary test indicated that the
less sensitive questions on legal status were acceptable to the
interviewers and foreign-born Hispanic farmworkers who participated. 


   RECOMMENDATIONS
---------------------------------------------------------- Chapter 0:5

GAO recommends that

  -- to help correct undercounts, reduce conceptual problems, and
     where possible, fill gaps for information on immigration flow,
     the Commissioner of INS should (1) evaluate and, where feasible,
     work toward improving data on flow and (2) utilize a set of
     demographic categories (or an information typology) that clearly
     distinguishes different concepts and helps determine which
     statistics can fairly be compared to others;

  -- to eliminate confused reporting of data and estimates concerning
     immigration flow, the Commissioner of INS should more clearly
     report information about trends in legal immigration flow and
     about the difference between the concepts of flow and net change
     in the INS Yearbook--or develop a new reporting format that
     communicates more effectively to policymakers and interested
     members of the general public; and

  -- to reduce the uncertainty associated with statistical estimates
     of demographic concepts other than immigration flow, fill
     information gaps for specific legal statuses, and address
     fragmented reporting, the Commissioner of INS and the Director
     of the Bureau of the Census should together (1) devise a plan of
     joint research for evaluating the quality of census and survey
     data on the foreign-born; (2) further develop, test, and
     evaluate the three-card method that GAO devised for surveying
     the foreign-born about their legal status; and (3) either
     publish a joint report or coordinate reports that present
     information on population size, net change, and emigration. 


   AGENCY COMMENTS
---------------------------------------------------------- Chapter 0:6

GAO met with INS officials who had reviewed a draft of this report,
including the Director of the Statistics Branch.  INS indicated that
it is currently working to improve the clarity of statistical
reporting in the Yearbook and that it finds GAO's typology very
useful. 

With respect to GAO's recommendation concerning the three-card
method, INS made two comments: 

  -- First, INS suggested that because it is not an expert in survey
     methodology, its appropriate role would be limited to providing
     support and consultation to Census in that agency's efforts to
     develop, evaluate, and test the new method.  GAO believes that,
     as stated above, the recommendation for joint INS-Census work
     allows latitude for INS and Census to determine their
     appropriate roles. 

  -- Second, INS indicated that it would need an independent
     evaluation of the three-card method before committing funds to
     the method's development.  GAO agrees that INS' obtaining an
     independent evaluation of the method before proceeding with
     further development would be prudent. 

The Bureau of the Census provided written comments (see app.  IV),
which raised no objections to GAO's findings on data gaps and the
quality of federal statistics on immigration. 

The Census Bureau stated a concern about its involvement in a survey
designed to obtain information on the legal status of the
foreign-born.  GAO's recommendation is only that the Census Bureau be
involved in the development, testing, and evaluation of the new
method, not necessarily in any resulting survey.  GAO believes that
Census would bring essential expertise to designing and overseeing
this work.  Testing--even large-scale testing--need not involve data
collection by the Census Bureau. 

The Department of State and the Department of Labor provided
technical comments, which were incorporated as appropriate.  The
Department of Health and Human Services had no comments on the
report. 


INTRODUCTION
============================================================ Chapter 1


   BACKGROUND
---------------------------------------------------------- Chapter 1:1

"Reliable data is a necessary ingredient for credible policy and its
implementation." So stated the U.S.  Commission on Immigration Reform
in 1994 (p.  xxxi).  But the Commission found that throughout its own
inquiry, inadequate data made it difficult to assess the impact of
immigration policy and of immigration itself on American society. 

Dissatisfaction with information on immigration has also surfaced on
Capitol Hill.  In spring 1996, the Immigration and Naturalization
Service (INS) issued a press release with the headline:  "U.S.  Legal
Immigration Down 10.4 Percent in 1995." The headline is based on a
1994-95 reduction in the number of green cards authorized, and the
reduction appears to have been caused mainly by a logjam in INS'
processing of green-card applications.\1 Subsequently, INS and most
of the experts testifying at a congressional hearing reported that,
in general terms, legal immigration was increasing.\2 At that
hearing, it also became apparent that some policymakers and reporters
had been confused about whether immigration was increasing or
decreasing. 

Recent dissatisfaction with immigration statistics comes roughly a
decade after a thorough review of immigration statistics (Levine et
al., 1985) summed up the situation as a history of neglect and
despite some efforts to provide better information.\3 It has been a
decade during which levels of legal immigration increased while
patterns of immigration shifted further away from the European
dominance of the early and mid-20th century to heavier flows of Asian
and Latin American immigrants.  At the same time, illegal immigration
emerged as a major concern, and the INS budget increased
dramatically, mainly because of more intensive efforts to curtail
illegal immigration.  (During the 1990s, INS' budget quadrupled; it
is expected to reach $4 billion in fiscal year 1999--up from less
than $1 billion in 1992.)

In recent years, the public--particularly California voters--entered
the debate about the value and cost of immigration.\4 And a series of
major bills--affecting legal immigration, the transition from illegal
to legal status, and the public benefits for which immigrants in
various statuses are eligible--were introduced and debated in
Congress. 

Debates concerning immigration continue.  For example, this year
Californians passed a hotly contested proposition to end bilingual
education.  Also, the computer industry asked Congress to increase
the number of temporary visas for high-tech workers.  Thus, there is
an increased need for valid, reliable, and clear policy-relevant
information.  Congressman Ed Bryant of Tennessee called for INS to
answer direct questions, such as:  "How many people--in total,
including every category
.  .  .  enter the United States each year?"\5 The Commission on
Immigration Reform (1994) also issued a call for new methods to be
developed to meet some of the difficult challenges in immigration
statistics, such as estimating the flow of illegal immigrants into
the United States. 


--------------------
\1 Identity cards attesting the legal permanent resident (LPR) status
of an alien in the United States are termed green cards because of
their former color. 

\2 Hearing before the House Subcommittee on Immigration and Claims,
Committee on the Judiciary, May 16, 1996. 

\3 For a follow-up on implementation of recommendations from two
National Academy of Sciences reports on immigration statistics
(Levine et al., 1985; Edmonston, 1996), see GAO/GGD-98-119. 

\4 In California, where immigration increases have been concentrated,
Proposition 187 was passed in 1994 calling for public agencies, such
as schools, to report suspected illegal immigrants to INS--a measure
that was blocked in court. 

\5 Hearing before the House Judiciary Subcommittee on Immigration and
Claims, May 16, 1996. 


      DEFINITION OF FOREIGN-BORN
      U.S.  RESIDENTS
-------------------------------------------------------- Chapter 1:1.1

The resident foreign-born population is defined here as all persons
who were born abroad (to parents who were not U.S.  citizens) and who
now either (1) are in a permanent legal status (naturalized citizen,
legal permanent resident, refugee, person granted asylum) or (2) if
in a temporary legal status or here illegally, remain in this country
for over a year.  The requirement for remaining more than a year is
based on the U.N.  definition. 


      DEMOGRAPHIC CONCEPTS
-------------------------------------------------------- Chapter 1:1.2

Four basic demographic concepts (categories of statistics) are
crucial to understanding information on the foreign-born population: 

  -- The inflow--or in this report, flow--refers to the movement of
     foreign-born persons into the United States and, as explained
     below, their transition into specific legal statuses. 

  -- The size of the foreign-born population in the United States
     (sometimes referred to as "stock") is the total number of
     foreign-born persons residing here at any given time, including
     those who have naturalized. 

  -- Net change in the size of the foreign-born population over a
     specific period of time can be calculated using a demographic
     balancing equation, accounting for flow, deaths, and emigration
     (see Bogue et al., 1993; Pollard et al., 1981).  The
     foreign-born population increases when the flow of new residents
     into the country exceeds the number who emigrate or die. 

  -- Emigration refers to persons moving out of the United States to
     take up residence in a foreign country.  In this report, we are
     concerned with the emigration of only foreign-born persons--not
     the emigration of persons born in the United States. 


      LEGAL STATUSES
-------------------------------------------------------- Chapter 1:1.3

The foreign-born population may be subdivided into groups, such as
those defined by the various legal immigration statuses:  legal
permanent residents, refugees and asylees, those legally permitted to
reside here on a temporary basis, illegal immigrants, and naturalized
citizens. 

Virtually all laws and policies on immigration differentiate
foreign-born persons according to their legal status.\6 Thus, from a
policy perspective, legal status is critical information.  Consistent
with this view, in 1994, the U.S.  Commission on Immigration Reform,
commenting on the need to improve estimates of the costs and benefits
of legal and illegal immigration, stated that confusion results from
grouping together illegal immigrants, legal immigrants, and refugees. 


--------------------
\6 For example, the Immigration and Nationality Act (INA) of 1952, as
amended (8 U.S.C.  1101 et seq.) defines various classes of
foreign-born persons (legal statuses) and indicates rights and
limitations of privileges for persons in those classes. 


      TWO TYPES OF FLOW
-------------------------------------------------------- Chapter 1:1.4

In considering the process of immigration flow, we recognized that
transitions from one immigration status to another must be
considered.  Persons who enter the United States in one immigration
status often adjust or change to a different status, and statistics
have been reported on this process.\7 Such transitions are a form of
"flow" (i.e., flow to a particular status).  This means that flow can
indicate not only new entries into the United States, but also
transitions or adjustments to a different legal category or status. 
We believe that it is important to recognize--and to clearly
distinguish between--these two, very different types of flow. 


--------------------
\7 For example, statistics are currently reported on the aliens who
adjust to LPR (or green- card) status--and on LPRs who transition to
naturalized citizenship. 


      FEDERAL AGENCIES PROVIDING
      STATISTICS ON IMMIGRATION
-------------------------------------------------------- Chapter 1:1.5

Most of the available federal statistical information on the
foreign-born is provided by two agencies:  the Immigration and
Naturalization Service and the Bureau of the Census.  INS provides
information relevant to flow.  INS and Census each provide some
information on the size of the resident foreign-born population and
net change.  Census also provides an estimate of emigration (i.e.,
the estimated number of foreign-born residents who leave the United
States to live in another country).  Some additional information is
maintained by the Departments of State, Health and Human Services,
and Labor. 


   OBJECTIVES, SCOPE, AND
   METHODOLOGY
---------------------------------------------------------- Chapter 1:2


      OBJECTIVES
-------------------------------------------------------- Chapter 1:2.1

The objectives of this report are (1) to identify policy-related
information needs for immigration flow and other key demographic
concepts that are relevant to migration; (2) to identify federal
statistics on the flow of immigrants (and information gaps) and to
determine what is known about the quality of existing statistics on
flow; (3) to identify federal statistics relevant to other key
demographic categories and to determine what is known about their
quality; and (4) to identify strategies for improving immigration
statistics. 


      SCOPE
-------------------------------------------------------- Chapter 1:2.2

The scope of this report is limited to current federal statistics
that provide basic demographic, statistical information on the
resident foreign-born population.\8 Current federal statistics are
defined here as those published by a federal agency for fiscal or
calendar year 1996--the most recent year for which statistics were
generally available during the time we collected data from federal
agencies (from Nov.  1997 to Apr.  1998).\9

With respect to policy-related information needs, our focus is on
congressional information needs--that is, the statistical data that
can inform congressional debates on immigration issues. 

In assessing the quality of federal statistics on the foreign-born,
we limited our work to determining what is known about their quality,
including what can be determined from logical comparisons and
analysis.  In exploring new strategies, we initially limited our
scope to the development of ideas; in one instance, we were
subsequently able to conduct a preliminary, qualitative test of a new
method for estimating legal status in a census or survey. 


--------------------
\8 This report does not address needs for special information, e.g.,
information on criminal aliens (see GAO/T-GGD-97-154). 

\9 Published federal statistics refer to numbers included in an
agency publication or release (e.g., a press release), an agency
"working paper" that has been distributed externally, or a specific
listing on an agency Internet Web page.  (In some instances,
published federal statistics based on sample surveys were presented
without confidence intervals.  In those instances, we requested
confidence intervals from the agency and presented that information
together with the published federal statistic.)


      METHODOLOGY
-------------------------------------------------------- Chapter 1:2.3


         IDENTIFYING
         POLICY-RELATED
         INFORMATION NEEDS
------------------------------------------------------ Chapter 1:2.3.1

To identify policy-related information needs, we reviewed recent
congressional debates, bills, and laws concerning immigration to
determine the kinds of basic demographic, statistical information on
the foreign-born needed by congressional policymakers.  We also
reviewed basic texts on demography, which identify key concepts, and
consulted with immigration experts.  (App.  I lists the immigration
experts we consulted.)

In reviewing federal agency literature, material from relevant
hearings, and laws requiring information on the foreign-born, we
found that such information has not been gathered or reported
according to a common framework or typology.  As noted earlier, there
have been instances of confusion in interpreting these kinds of
information.  Therefore, to identify policy-related information needs
concerning the legally and illegally resident foreign-born
population, we proceeded through a two-step process, which included

  -- developing a basic typology or set of demographic categories
     (i.e., a systematic framework defining various types of
     policy-relevant demographic, statistical information on the
     foreign-born population and their interrelationships) and

  -- examining, in a general sense, whether these types of
     information were, in fact, needed by Congress and interested
     members of the general public. 

We developed our information typology (set of demographic categories)
in consultation with immigration experts.  We then examined the need
for these types of information by reviewing laws requiring
demographic, statistical information on the foreign-born, past
congressional requests for information, and recurrent congressional
activities.  We discussed our typology with staff at the Immigration
and Naturalization Service. 


         IDENTIFYING AND ASSESSING
         CURRENT FEDERAL
         STATISTICS
------------------------------------------------------ Chapter 1:2.3.2

Using our typology, we identified relevant federal statistics and
gaps.  Briefly, we reviewed literature published by federal agencies
and followed up with officials and staff at INS, the Bureau of the
Census, and other agencies. 

We then evaluated the quality of the relevant statistics, considering
technical adequacy and timeliness as well as the adequacy with which
the information was reported.  To guide our work, we developed
checklists for statistical quality, based on a review of literature
(including federal agency standards, published empirical assessments
of statistical quality, and evaluation and statistical texts) and
discussions with agency staff.  We used these checklists to ensure
comprehensiveness in interviewing federal agency staff, experts, and
users about statistical quality and in reviewing relevant literature. 
We then developed quality ratings to describe published statistics in
each demographic category as problem-free or as limited by conceptual
problems and confused reporting, as overcounts or undercounts, or as
uncertain or unevaluated statistics.  If no published federal
statistic could be identified for a demographic category, the
descriptive rating consists of the notation that a gap exists. 


         IDENTIFYING STRATEGIES
         FOR IMPROVEMENT
------------------------------------------------------ Chapter 1:2.3.3

To identify strategies to improve federal statistics on immigration,
we (1) logically analyzed the problems we had identified, (2) talked
with agency staff and experts about possible approaches, (3) reviewed
literature, and (4) developed our own new strategy for collecting
relevant data.  In particular, based on previous research on survey
methods for asking sensitive questions and on demographic methods for
estimating illegal immigrants, we devised a new method for
interviewing foreign-born respondents and collecting data on their
immigration status while protecting privacy--the three-card method. 
We pretested the three-card method in interviews with foreign-born
Hispanics in farmwork settings, at a legal clinic for immigration
problems, and at a city "drop-in" center.  (These interviews were
conducted by members of our staff who are fluent in Spanish.) We then
conducted a preliminary test of the acceptability of this method to
interviewers and respondents by contracting for 81 interviews with
foreign-born farmworkers as a supplement to the National Agricultural
Workers Survey (NAWS), debriefing the interviewers who administered
the questionnaire, and examining the results for signs of respondent
comfort with the series of questions. 

We conducted our audit work in accordance with generally accepted
government auditing standards between June 1997 and June 1998. 
Preliminary work on the three-card method was conducted earlier.  We
did not conduct an audit of how the INS and Census data were
initially gathered and processed by the agencies. 

We requested comments on a draft of this report from INS, Census, and
the Departments of State, Labor, and Health and Human Services.  On
July 13, 1998, Census provided written comments, and on July 15, INS
provided oral comments at a meeting attended by INS officials
including the Director of the Statistics Branch.  INS' and Census'
comments are discussed in relevant sections of chapter 5 and appendix
III.  INS, as well as State and Labor, provided some technical
comments and suggestions for clarification, which we incorporated as
appropriate.  Health and Human Services reviewed a draft and said
that it had no comments. 


   ORGANIZATION OF THIS REPORT
---------------------------------------------------------- Chapter 1:3

Chapter 2 of this report presents our typology of policy-relevant
statistical information on the foreign-born population (i.e., the set
of demographic categories) and links this typology to policy-related
information needs.  Chapter 3 identifies and assesses current federal
statistical information on immigration flow.  Chapter 4 identifies
and assesses corresponding information on other key demographic
concepts--size of the foreign-born population, emigration, and change
in the size of the foreign-born population.  Chapter 5 discusses
strategies for improvement and makes recommendations to the
Commissioner of INS and the Director of the Bureau of the Census. 

Appendixes provide more detailed information of concern to technical
readers.  Appendix I lists the immigration experts we consulted. 
Appendix II briefly reviews available information on the demographic
characteristics of foreign-born residents.  Appendix III provides
information on the three-card method for collecting survey data on
legal status and our preliminary test of its acceptability to both
respondents and interviewers.  Appendix IV reprints the comments from
the Bureau of the Census. 


POLICY-RELATED INFORMATION NEEDS
============================================================ Chapter 2

A basic typology of policy-relevant statistical information on the
resident foreign-born population\1 can be defined by combining two
dimensions:  one consisting of four demographic concepts that are
relevant to migration (flow, size of the foreign-born population, net
change in size, and emigration) and the other consisting of legal
statuses (legal permanent residents, refugees and asylees, persons
permitted to reside here on a temporary basis, illegal immigrants,
and naturalized citizens).  We developed the typology (or set of
demographic categories) as a tool for sorting and defining different
types of statistical information on the foreign-born population. 

Reviewing specific policy-relevant information needs with reference
to our typology, we found that

  -- Congress has passed laws requiring, either generally or
     specifically, much of the information included in the typology
     and, in some instances, has indicated that the information is
     needed to improve decision-making;

  -- Congressional committees have requested some of the information;
     and

  -- Virtually all of the information is directly or indirectly
     relevant to various congressional activities (e.g., information
     on immigration flow is relevant to establishing or changing
     numerical limits for certain classes of immigrants and temporary
     visas). 


--------------------
\1 In this report, the term foreign-born resident is used to refer to
all foreign-born who are in a permanent legal status or who, if in a
temporary legal status or here illegally, have remained here for
longer than a year. 


   AN INFORMATION TYPOLOGY
---------------------------------------------------------- Chapter 2:1

To build a policy-relevant framework for types of demographic,
statistical information on the foreign-born, we crossed the two
dimensions--demographic concept and legal status of the foreign-born,
as shown in table 2.1.  In doing so, we defined flow with three
columns to distinguish new arrivals, transitions to a new status, and
total flow to specific legal statuses (e.g., all new LPRs).  We
defined each of the other demographic concepts--the size of the
foreign-born population, net change in size, and emigration--with an
individual column.  The rows of the table represent major legal
statuses.  In total, we identified 33 discrete categories, each of
which specifies a distinct type of information. 



                                                                                      Table 2.1
                                                                       
                                                                        Typology of Statistical Information on
                                                                               Foreign-Born Residents\a

                                                                                                    Demographic concept
                                       ---------------------------------------------------------------------------------------------------------------------------------------------
                                                                      Annual flow                                    Population size      Net change in size     Annual emigration
                                       -------------------------------------------------------------------------  ---------------------  ---------------------  --------------------
                                                                                                                  Number of foreign-     Annual increase or
                                       New residents (entered    Transitions to a new      Combined-flow to each  born persons residing  decrease in the size   Foreign-born
                                       U.S. during reference     status (during reference  status (sum of first   here (at given point   of foreign-born        emigrating (during
Legal status of foreign-born           year)                     year)                     two columns)           in time)               population             reference year)
-------------------------------------  ------------------------  ------------------------  ---------------------  ---------------------  ---------------------  --------------------
A. Legal permanent residents           Number who received       Number transitioning to   All new LPRs           Number of LPRs         Increase or decrease   Number of LPRs who
                                       green-card status upon    green-card status;                                                      in the number of LPRs  emigrated
                                       admission to the U.S.     entered the U.S.                                                        residing here
                                                                 earlier
                                                                 (as B, C, D)

B. Refugees and asylees (without LPR   Number who were admitted  Number of new asylees     All new refugees and   Number of refugees     Increase or decrease   Number of refugees
status)\b                              to the U.S. as refugees   who had entered the U.S.  asylees                and asylees (without   in the number of       and asylees (without
                                       or as asylees ("trailing  earlier                                          LPR status)            refugees and asylees   LPR status) who
                                       relatives" of those       (as C, D)                                                               (without LPR status)   emigrated
                                       granted asylum)                                                                                   residing here

C. Students, temporary workers, and    Number of new residents   Number already here who   All residents who      Number of esidents     Increase or decrease   Number of residents
their                                  who entered with student  newly obtained legal      newly achieved r       with student or        in the number of       with legal temporary
families; others temporarily here      or work visa, or other    temporary status;         temporary legal        temporary work visas   residents with         status who emigrated
who                                    temporary form of         entered U.S. earlier      status                 or other temporary     temporary legal
stayed > 1 year\c                      admission                 (as D)                                           legal status           status

D. Illegal immigrants (EWIs and        Number of new illegal     Number who illegally      All new illegal        Number of residents    Increase or decrease   Number of illegal
overstays)                             immigrants who "entered   overstayed; entered U.S.  immigrants             here illegally         in number of illegal   residents who
who stayed > 1 year                    without inspection"       earlier (as C)                                                          residents              emigrated
                                       (EWI)

E. Naturalized citizens                Not an entry status\d     Number who achieved       Same as number in      Number of naturalized  Increase or decrease   Number of
                                                                 citizenship; entered      previous column        citizens               in the number of       naturalized citizens
                                                                 U.S. earlier (as A-D)                                                   naturalized citizens   who emigrated
                                                                                                                                         residing here

Total                                  Total annual flow from    Total not meaningful      Total not meaningful   Total foreign-born     Net change in the      Total foreign-born
                                       foreign countries to      because of double-        because of double-     who reside here = sum  size of the foreign-   who emigrated = sum
                                       U.S. = sum of above       counting; some persons    counting; some enter   of above               born population = sum  of above
                                                                 make more than one        the U.S. and                                  of above
                                                                 transition per year       transition to a
                                                                                           different status
                                                                                           within the same year
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------


Legend

EWI Entry without inspection
LPR Legal permanent resident

Note:  Statistics are not relevant to the shaded categories. 

\a Current data or estimates and trends. 

\b The term "asylees" refers to persons granted asylum. 

\c Other foreign-born residents who were admitted temporarily include
(1) trainees, treaty-traders, and investors as well as their
families; (2) special groups, such as those given temporary protected
status or extended voluntary departure; (3) parolees, that is,
persons allowed to enter the United States on a nonpermanent basis
for urgent humanitarian reasons or when significant public benefit is
likely; and (4) persons entering on tourist or other short-term visas
who nonetheless remain in the United States longer than a year. 

\d In recent years, there has been an average of roughly 300
foreigners per year who--on the basis of having married U.S. 
citizens employed abroad--attained U.S.  citizenship via a special
provision that allows naturalization of applicants who have not
resided in the United States (see INA section 319(b), 8 U.S.C. 
1430(b)).  Some of these persons may enter the United States--for the
first time--as naturalized citizens. 

Source:  GAO conceptualization. 

What is critical for policy analysts and for users of information is
how accurately measures of immigration reflect the actual patterns of
immigration (Kraly and Warren, 1992).  The typology in table 2.1
represents a step in the direction of measuring actual patterns and
reducing confusion.  This is because the typology (or set of
demographic categories) can be used as a tool for sorting existing
statistics and thus determining where gaps exist.  It can also help
in the interpretation of information by clarifying concepts such as
the distinction between two types of flow.  The typology also makes
plain the interrelationship of different kinds of statistical
information and helps clarify which statistics are directly
comparable and which are not. 


   A CONSIDERATION OF
   POLICY-RELATED INFORMATION
   NEEDS
---------------------------------------------------------- Chapter 2:2

A review of major immigration laws indicated that they often include
requirements for federal agencies to report information on flow, the
size of the foreign-born population, net change in size, and
emigration.  Notably, the Immigration Reform and Control Act (IRCA)
of 1986 requires triennial reports to Congress,\2 which are to
describe the number of persons who are legally admitted or paroled
into the United States within a specified interval as well as those
who illegally enter or overstay temporary visas during that
interval.\3 This corresponds to the demographic concept of
immigration flow.  Note that with respect to illegal immigrants, it
is important to distinguish two different types of flow:  those who
"enter without inspection" (EWIs) and those who overstay temporary
visas.  The distinction is important from a policy perspective
because, for example, stronger border controls would not address the
overstay issue. 

IRCA also requires information on a variety of impacts of
immigration, including, for example, the impact on demographics and
population size as well as the impact on social services.  We note
that the impact on population size would logically involve new
entries, but not transitions between legal statuses.  The impact on
social services would occur as a result of both new entries and
transitions of legal status--that is, both types of flow--because the
foreign-born resident's specific legal status determines whether or
not he or she is eligible for specific benefits.  (It is important to
note that although information on both types of flow combined is
relevant for certain policy purposes, data on combined-flow to a
legal status should be treated with caution.  This is because it is
not always clear to what extent each type of flow is represented;
e.g., a change in a combined-flow statistic might reflect a change in
the number of new entries or a change in the number of transitions to
the status in question--or some combination of the two.)

With respect to the other relevant demographic concepts--the size of
the foreign-born population, net change in size, and emigration--the
Immigration Act of 1990 requires information on the alien population
of the United States as well as rates of emigration and an analysis
of trends.\4

Other laws, while not directly requiring statistical information,
nevertheless include mandates that imply such a need.  For example,
the Illegal Immigration Reform and Immigrant Responsibility Act of
1996 mandates an evaluation of the effort to deter illegal entry into
the United States.  Such an evaluation would require a variety of
statistical information, such as trends in the size of the illegal
population (see GAO/GGD-98-21). 

There are other indications of the kinds of information that Congress
has wanted in recent years.  For example, the House Judiciary Report
accompanying IRCA indicated that the requirement for the triennial
reports is intended to enable Congress to review and study
immigration and refugee programs and to consider possible changes to
them with the benefit of reliable and detailed data.\5 Congressional
committees have requested that we provide statistical estimates, such
as projections of legal immigration, to help in decisions regarding
numerical limits.\6 Another example would be the congressional
request for estimates of the number of certain nonimmigrant workers
who transitioned to green-card status (see GAO/PEMD-92-17). 

Among the ongoing or recurring congressional activities for which the
types of information shown in table 2.1 might be useful are

  -- periodic revisions of numerical limits for LPRs, annual setting
     of levels of refugee inflow, periodic resetting of limits for
     certain temporary visas, and annual prioritization of funding
     and special programs intended to reduce illegal immigration;

  -- periodic redefinitions of (1) the conditions under which illegal
     immigrants and others can (or cannot) adjust to LPR status and
     (2) ceilings on the number of asylees who may transition or
     adjust to LPR status; and

  -- periodic revisions of public benefits available to persons in
     different immigration statuses, which can in turn influence
     personal decisions about changing one's immigration status. 

In addition, congressional committees have indicated that they wanted
information to address issues such as the impact of foreign workers
on the U.S.  economy and on the working conditions of Americans (see,
e.g., GAO/PEMD-92-17 and GAO/HEHS-98-20).  There has also been some
interest in the trends in the numbers of naturalized citizens,
because they have the right to bring in certain relatives.\7


--------------------
\2 P.L.  99-603, section 401, codified at 8 U.S.C.  1364. 

\3 Parole is permission to enter the United States and remain for an
undefined temporary period for humanitarian or public interest
reasons.  Information required for the triennial reports includes (1)
the number and classification of aliens (including nonimmigrants) who
were admitted, paroled, or granted asylum during the relevant period
and (2) a reasonable estimate of the number of aliens who entered the
United States during the period without visas or who became
deportable during the period under [general classes of deportable
aliens named in] section 237 (formerly, 241) of the Immigration and
Nationality Act, 8 U.S.C.  1237. 

\4 P.L.  101-649, section 142, codified at 8 U.S.C.  1103(d). 

\5 H.  Rept.  99-682(I) at 99 (1986). 

\6 In 1988 and 1989, we responded to congressional requests for
projections of levels of future legal immigration (GAO/PEMD-88-7,
GAO/T-PEMD-89-1, GAO/PEMD-89-12, GAO/PEMD-90-5).  In deciding to
change the annual numerical limits in the Immigration Act of 1990,
Congress relied heavily on this work (see Congressional Record, July
12, 1989, 14297-312). 

\7 Academic and government interest in this topic has been addressed
in a series of articles in International Migration Review.  (For an
overview of these articles, see Goering, 1989.)


IMMIGRATION FLOW
============================================================ Chapter 3

INS' annual Statistical Yearbook includes several statistics on
immigration flow--particularly, statistics on LPR flow (persons with
new green-card status), refugees and asylees, and naturalized
citizens.  But various quality problems limit the utility of these
data for policy purposes. 

  -- Conceptual problems make a key trend difficult to interpret and
     valid comparisons of certain reported data difficult to make;
     confused reporting compounds the conceptual problems. 

  -- Administrative data undercount persons granted asylum as well as
     those attaining naturalized citizenship. 

  -- Data gaps occur for key statistics, such as the number of
     foreign-born persons who take up residence here each year. 

The relevant statistics and descriptive quality ratings (together
with the reasons for our ratings) are presented in table 3.1. 

The INS Yearbook includes several statistics that seemingly fit our
typology for information on immigration flow.  As shown in table 3.1,
these include, for example, figures for the number of new LPRs who
entered the United States in fiscal year 1996 (421,405), the number
of persons already here who attained new green-card status (494,495),
and the total number of new LPRs (915,900).  Similar kinds of data
are reported for refugees and asylees and for naturalized citizens
(see rows B and E of table 3.1).  The INS Yearbook also discusses the
sources of these data and some of their limitations. 

Few, if any, relevant data on flow are reported outside the Yearbook. 



                                         Table 3.1
                          
                              Published Federal Statistics on
                             Immigration Flow, With Descriptive
                                Quality Rating and Reason\a

                                        Demographic concept: annual flow
                      ---------------------------------------------------------------------
                      New residents         Transitions to a new    Combined-flow to each
Legal status of the   (entered U.S. during  status (during          status (sum of first
foreign-born          reference year)       reference year)         two columns)
--------------------  --------------------  ----------------------  -----------------------
A. Legal permanent    421,405 new LPRs      494,495 new LPRs        915,900 new LPRs
residents

                      Overcount: Some       Undercount: Some        Conceptual problem: The
                      persons who           persons counted as new  majority of new LPRs
                      transitioned to LPR   residents belong here;  were already living in
                      status (and belong    also, processing        U.S., so the count does
                      in the next col.)     slowdown lowered the    not measure flow into
                      were counted here.    count.                  the U.S.; in 1995 and
                                                                    continuing into 1996, a
                                                                    processing slowdown
                                                                    affected the count.

                                                                    Confused reporting: INS
                                                                    Yearbook highlights
                                                                    trends as if
                                                                    meaningful; some
                                                                    statements in the
                                                                    Yearbook imply that new
                                                                    LPRs represent new
                                                                    entries.

B. Refugees and       74,791 new refugees   18,556 new asylees      93,347 new refugees and
asylees\b                                                           asylees

                      Undercount:           Undercount: Persons     Undercount: Some
                      "Trailing relatives"  granted asylum on       asylees are not
                      who enter the U.S.    appeal and some         counted.
                      as asylees are not    "trailing relatives"
                      counted in the INS    (who were already in
                      Yearbook.             the U.S.) are not
                                            counted in the INS
                                            Yearbook.

C. Students,          No published          No published estimate.  No published estimate.
temporary workers,    estimate.
and their families;
others temporarily
here who stayed > 1
year\c

                      Gap: No data          Gap: No data identify   Gap: No data identify
                      identify the number   the number new to this  the number who stayed >
                      new to the U.S. who   category who were       1 year.
                      stayed > 1 year.      already in the U.S.

D. Illegal            No published          No published estimate.  No published estimate.
immigrants (EWIs and  estimate.
overstays) who
stayed for > 1 year

                      Gap: INS has not      Gap: INS has not        Gap: INS estimates net
                      estimated illegal     estimated illegal       change in the illegal
                      flow.                 flow.                   population, but has not
                                                                    estimated illegal flow.

E. Naturalized        Not applicable.       1,044,689 new citizens  1,044,689 new citizens
citizens

                Not an entry          Undercount: Most minor  Undercount: Most minor
                      status\d              children are not        children are not
                                            counted.                counted.

Total flow            No published          Total not meaningful    Total not meaningful
                      estimate.             because of double-      because of double-
                                            counting.               counting.

                      Gap: Because of gaps  Some persons make more  Some persons enter the
                      for some categories,  than one transition in  U.S. and then
                      the total cannot be   the same year.          transition to a
                      tallied.                                      different status--
                                                                    within the same year.
-------------------------------------------------------------------------------------------
Legend

EWI Entry without inspection
LPR Legal permanent resident

Note:  Statistics are not relevant to the shaded cells. 

\a Fiscal year 1996. 

\b The term "asylees" refers to persons granted asylum. 

\c Other foreign-born residents admitted temporarily include (1)
trainees, treaty-traders, and investors as well as their families;
(2) special groups, such as those given temporary protected status or
extended voluntary departure; (3) parolees, that is, persons allowed
to enter the United States on a nonpermanent basis for urgent
humanitarian reasons or when significant public benefit is likely;
and (4) persons entering on tourist or other short-term visas who
nonetheless remain in the United States longer than a year. 

\d In recent years, there has been an average of roughly 300
foreigners per year who--on the basis of having married U.S. 
citizens employed abroad--attained U.S.  citizenship via a special
provision that does not require residing in this country (see INA
section 319(b), 8 U.S.C.  1430(b)).  Some of these persons may enter
the United States--for the first time--as naturalized citizens. 

Sources:  Published federal statistics on immigration flow are from
INS 1996 Yearbook, for LPRs, p.  52; for refugees and asylees, p.  85
(INS printed Department of State figures); for naturalized citizens,
p.  147.  Ratings and reasons are based on interviews with INS,
State, and EOIR staff, information obtained from agency datasets,
detailed information provided in the INS Yearbook, and GAO analysis. 


   CONCEPTUAL PROBLEMS AND
   CONFUSED REPORTING
---------------------------------------------------------- Chapter 3:1


      TRENDS IN THE NUMBER OF NEW
      LPRS
-------------------------------------------------------- Chapter 3:1.1

INS' administrative count of the number of new LPRs combines the flow
of (1) new LPR entries to the United States and (2) transitions (or
adjustments) to LPR status.  But there is a conceptual problem;
namely, this statistic represents two different measures, each of
which can vary independently.  This makes results--particularly for
trends--difficult to interpret. 

The majority of new LPRs are in category 2, transitions to LPR
status.  That is, they are not new to the United States; as indicated
in tables in the INS Yearbook, they have already been living here for
years--typically either illegally or as long-term temporary
residents.  Various factors can raise or lower the number of green
cards authorized for such persons, independently of trends in new
entries.  Two instances show how this can--and has--happened: 

  -- The late 1980s amnesty (through IRCA) for illegal immigrants who
     had lived here for more than 5 years created a sudden major
     upswing in the trend line for new LPRs because a large group of
     illegal residents became eligible to apply for green-card
     status.  This increase was unrelated to any change in the number
     of persons entering the United States. 

  -- A recent change in law (1994) allowed illegal immigrants living
     in the United States who qualified for green cards to transition
     to LPR status without leaving the United States--thus shifting
     the processing of thousands of cases from the Department of
     State to INS.  Because INS could not immediately handle the
     additional workload, there was a logjam, or slowdown, in issuing
     the cards to persons already living here.  The logical effect of
     a slowdown is a decrease in the number of cards authorized
     (i.e., a downturn in the trend line), independent of any change
     in the number of persons newly taking up residence in the United
     States.  Subsequently, as INS' capacity to handle the new
     workload improves, a speed-up in processing would increase the
     number of cards authorized for persons already living here,
     creating an upswing in the combined-flow trend line. 

INS statistical staff told us that annual trends in the number of new
LPRs do not convey a meaningful indication of any demographic
concept.\1 They also said it is unclear how to disentangle the
effects of processing logjams and catch-ups. 

Despite these problems, INS has repeatedly highlighted annual trends
in "immigrants admitted" (what we term the combined-flow LPR
statistic in table.  3.1).  The 1995 and 1996 INS Yearbooks lead off
their introductions with the first highlights of current findings, as
follows: 

  -- "720,461 persons were granted legal permanent residence status . 
     .  .  a decrease of nearly 84,000 from the year before" (INS
     1995 Yearbook, p.  11). 

  -- "915,900 persons were granted legal permanent resident status . 
     .  .  an increase of more than 195,000 over the year before"
     (INS 1996 Yearbook, p.  11). 

In each case, the same page of the Yearbook interprets these trends
as either a "decline in immigration to the United States" (1995) or a
"rise in immigration to the United States" (1996).  The introduction
itself does not mention the recent processing problems or that any of
the new immigrants counted were already living here. 

Readers who turn to the body of the Yearbook will find caveats. 
However, although tables in the INS Yearbook indicate that the
majority of new LPRs were already living in the United States, the
Yearbook text indicates only that some were already here.\2

INS statistical staff maintain that the agency never claims that the
count of newly authorized green cards represents a proxy for all new
residents, but the use of the term "admitted" may be confusing to
some readers.  Indeed, the 1996 Yearbook makes the confusing
statement that

     "The majority of immigrants [LPRs] enter the United States as
     immediate relatives of U.S.  citizens or through the preference
     system, consisting of family-sponsored and employment-based
     immigrants.  These categories combined accounted for 78 percent
     of all admissions in 1996." (p.  18, emphasis added.)

A nonexpert might infer that "enter the United States" means exactly
that, not realizing that the majority of new LPRs were already here. 
(In other words, some readers might not realize that the statement
quoted above is supposed to refer to persons entering LPR
status--regardless of whether they are already living here, as the
majority are.)


--------------------
\1 One policy-relevant interpretation of this statistic would be the
number of persons newly qualifying for benefits accorded to LPRs. 

\2 One sentence--in the 1996 Yearbook section, "Limitations of Data"
(p.  24)--does mention the fact that those who are transitioning or
adjusting have often lived here for years. 


      FLOW OF LEGALS VERSUS NET
      CHANGE IN ILLEGALS
-------------------------------------------------------- Chapter 3:1.2

Another conceptual problem arises because the Yearbook presents
statistics relevant to flow for legal immigration--but an estimate
for net change in the illegal population.  That is, the 1996 Yearbook
reports that nearly 1 million persons achieved green-card status in
fiscal year 1996 and that the population of illegals is increasing,
on average, by 275,000 per year.  These two figures are not
comparable, however, because flow is a very different concept than
net change.  Briefly, the difference in concepts is as follows: 

  -- The flow of illegal immigrants refers to new illegal EWIs and
     overstays who resided here for more than a year.  (As shown in
     table 3.1, these two types of flow can be described
     separately--and a combined number can be provided.)

  -- Net change in the size of the illegal population is calculated,
     mathematically, as the difference between (1) the flow of
     illegal residents and (2) legalizations and other exits
     (emigration, death) from the entire illegal population, as shown
     in figure 3.1. 

  -- Net change may be a positive or negative number.  When positive,
     net change measures the extent to which the flow exceeds
     legalizations, emigration, and deaths--on the part of the entire
     population of illegals who were already living here.  When
     legalizations, emigration, and deaths outnumber the entries, net
     change is negative. 

Legal flow cannot be validly compared to net change in the illegal
population.  The fact that the INS Yearbook reports statistics
relevant to flow for legal immigration but net change for the illegal
population is compounded by the fact that the Yearbook does not
discuss demographic concepts.  The Yearbook does not clarify the
meaning of immigration flow or net change in population size--or the
distinction between the two.  Reporting flow for legals and net
change for illegals, without clearly distinguishing between flow and
net change, could lead to misinterpretations; that is, it might
invite invalid comparisons.  If a comparison of legal and illegal
immigration is to be made, the same demographic concept should be
used for data on legals and illegals; for example, legal flow should
be compared to illegal flow. 

   Figure 3.1:  Net Change in the
   Size of the Illegal Population

   (See figure in printed
   edition.)

   Source for net-change equation: 
   Bogue et al., 1993.

   (See figure in printed
   edition.)

This is important because numerically, the difference (illegal flow
versus net change in illegals) could be great.  Large numbers of
illegal immigrants transition to green-card status (over 120,000 of
those authorized to receive green cards in fiscal year 1996 admit to
having entered the United States as EWIs), and the emigration of
illegal immigrants may also be large.  Thus, the flow of illegal
immigrants might--in some years--be considerably larger than the
reported net change in the size of the illegal population. 


   UNDERCOUNTS
---------------------------------------------------------- Chapter 3:2


      ASYLEES ARE UNDERCOUNTED
-------------------------------------------------------- Chapter 3:2.1

We found that INS Yearbook tallies of persons granted asylum are
limited to cases processed through INS' RAPS (Refugee, Asylum and
Parole System) data system.  Both RAPS and the Yearbook omit asylees
whose cases were approved by the Executive Office of Immigration
Review (EOIR) rather than INS.\3 Both also omit asylees who enter
from abroad after their paperwork is approved by INS--trailing
relatives of persons granted asylum (i.e., family members "following
to join" a principal asylee)--and processed abroad either by the
Department of State or INS.\4 Also omitted were other trailing
relatives whose cases were processed in the United States by INS (460
during the first four months of fiscal year 1996).\5

Table 3.2 summarizes the count of asylees published in the INS
Yearbook and the additional counts that we were able to identify by
talking with various staff at Department of State, EOIR, and INS and
by requesting tabulations from various data systems.  INS staff told
us that in addition to the 460 cases noted above, they estimate that
approximately 1,000 more were processed in the United States during
fiscal year 1996; however, the number of trailing relatives approved
and processed overseas by INS during fiscal year 1996 could not be
estimated. 



                               Table 3.2
                
                   Persons Granted Asylum and Whether
                  Included in Yearbook Count (FY 1996)

                                Number new asylees  Agency data source
Included in count?              ------------------  ------------------
Yes                             18,556              INS

No                              5,096               EOIR

                                3,652\a             State

                                1,460 + an          INS
                                additional unknown
                                number\b

======================================================================
Subtotal                        10,208 + an         INS, State, EOIR
                                additional unknown
                                number

======================================================================
Total                           28,764 + an         INS, State, EOIR
                                additional unknown
                                number
----------------------------------------------------------------------
Legend

EOIR = Executive Office of Immigration Review, Department of Justice
INS = Immigration and Naturalization Service

\a These 3,652 cases were approved by INS and processed overseas by
the Department of State for admission to the United States with
asylum status.  State told us that in most instances, a case
represents an individual person, but there may be exceptions (i.e.,
instances where a single case represents multiple persons). 

\b INS approved 460 trailing relatives in the United States during
the first 4 months of fiscal year 1996, and estimates that
approximately 1,000 more were approved during the remaining 8 months. 
The additional unknown number represents an unknown number of
trailing relatives whose cases INS processed overseas for admission
to the United States with asylum status. 

Source:  GAO analysis of INS, State Department, and EOIR data for
fiscal year 1996. 

In sum, the count of asylees published in the INS Yearbook should be
increased--probably by more than 50 percent (i.e., from 18,556 to
28,764 or perhaps to an even higher figure).  The undercount is not
fully described in the Yearbook, apparently because the
administrative processes are complex and it is difficult to identify
all cases in which asylum was granted.  By contrast, however, we
found no biases in the Department of State counts of refugees
reported in the INS Yearbook. 


--------------------
\3 EOIR is within the Department of Justice, but is separate from
INS.  Asylum cases denied or referred by INS are reviewed by EOIR
immigration judges.  EOIR provides a daily electronic transfer of
records to RAPS for administrative purposes.  The INS Yearbook notes
that asylees processed by EOIR are not counted in the Yearbook's
tallies. 

\4 On behalf of INS, the Department of State processes "Visa 92"
cases in areas where INS does not have facilities.  These cases
include spouses and children who are in another country at the time
when a member of the family was granted asylum inside the United
States (see 8 U.S.C.  208(b)(3)).  Based on an INS-approved petition,
State issues travel papers to these spouses and children, allowing
them to come to the United States; once here, they are admitted with
asylum status.  We believe that, while some of these spouses and
children may not have actually come to the United States, most of
them likely came--because all of them were processed by the
Department of State or INS (albeit in an overseas location).  They
have not been included in any count reported in INS or State
Department publications--nor are they mentioned as omitted from
relevant counts in the INS Yearbook. 

\5 Trailing relatives processed by INS are included in INS' Computer
Linked Application Information Management System (CLAIMS).  Although
INS could identify 460 trailing relatives that were processed in the
United States during the first 4 months of fiscal year 1996, INS was
not able to identify other cases of asylee approval that were
included in the CLAIMS system.  The 1996 Yearbook does not report
that its tallies omit these asylees. 


      NATURALIZED CITIZENS ARE
      UNDERCOUNTED
-------------------------------------------------------- Chapter 3:2.2

The number of naturalizations reported by INS (row E of table 3.1) is
an undercount of new foreign-born U.S.  citizens because the tally
excludes most minor children.  That is, as explained in the INS
Yearbook, minor children automatically receive citizenship (by
derivation) when their parents naturalize.\6 A separate form is not
required for these children, and they are not listed on their
parents' forms.  INS counts persons listed on forms; a complete count
cannot be obtained without revising the existing form.  (We also note
that trends in naturalization are affected by processing speed-ups
and slowdowns similar to those discussed above for LPRs.  As of July
1998, the logjam for naturalization applications was estimated to be
between 1.6 and 2 million unprocessed applications.\7 )


--------------------
\6 Under section 320 of the Immigration and Naturalization Act (8
U.S.C.  1431), a child automatically becomes a naturalized citizen if
he or she is under 18 years old and is living in the United States
pursuant to lawful admission for permanent residence at the time of
naturalization. 

\7 The logjam occurred because of an unprecedented increase in
applications for citizenship.  The process of rolling out new
fingerprinting procedures may have also contributed to the slowdown. 


      TRANSITIONS TO LPR STATUS
      MAY BE UNDERCOUNTED
-------------------------------------------------------- Chapter 3:2.3

Transitions to LPR status may be undercounted.  INS statistical staff
told us that in 1996, transitions to LPR status apparently continued
to be deflated (to some extent).  That is, even after the change in
law allowing illegal U.S.  residents to transition to green-card
status without leaving and reentering, some continued to exit and
reenter--preferring the travel to paying the fee required for
transitioning without leaving.  Although this number may be
relatively small, it represents a subtraction from the transitions
column and an addition to the new entries column--thus distorting
what is reported about patterns of flow to some extent. 


   DATA GAPS
---------------------------------------------------------- Chapter 3:3


      GAP FOR FLOW OF LONG-TERM
      RESIDENTS ADMITTED WITH
      LEGAL TEMPORARY STATUS
-------------------------------------------------------- Chapter 3:3.1

There is a gap in flow statistics for residents who are admitted with
temporary visas (i.e., nonimmigrant visas, such as student visas,
temporary work visas, and so forth).  The gap occurs because the
relevant INS data system does not distinguish newly admitted persons
from readmissions of the same person.  For example, a foreign student
living here for 4 years but visiting his parents briefly every 6
months would be represented as eight short stays--for up to eight
individuals--rather than one long stay for a single person.  Thus,
although durations of stay have been calculated (INS, 1996; Lowell,
1996), they do not correspond to the flow of long-term residents who
were admitted with legal temporary status.  Moreover, problems at INS
or with INS contractors (including computer-processing problems and
lack of agreement on what constitutes a valid match) have, since
1992, prevented calculations involving matched entries and exits. 
And with respect to this and other data systems, INS statistical
staff told us that there had been no recent audits or evaluations to
test the level of error in data processing and data archiving.  Such
work might reveal, for example, whether double-counting or lost cases
occurred. 

Department of State information on issuances of temporary visas
cannot fill this gap for two reasons.  First, in some cases, a visa
allows multiple entries and exits, whereas other visas allow only a
single entry so that a new visa is required for each entry and exit. 
There is no direct correspondence, then, between the number of visas
issued and the number of new residents with visas.  Second, a single
person can, if qualified, be issued multiple visas in different
categories--for example, a temporary work visa and a tourist visa
could be issued to the same person on the same day--and it would be
very difficult to cross-reference records for statistical purposes. 

We believe this gap is important because the number of long-term
residents living here with temporary visas may be quite large.  One
analyst (Woodrow, 1998) puts the flow of nonimmigrant residents into
the United States at approximately 1 million, perhaps more, in 1990. 
Under alternative assumptions, possible figures for fiscal year 1996
might range from very roughly 500,000 to 1 million, but the true
dimensions are unknown.\8


--------------------
\8 Our analysis--listing all sources of nonimmigrant flow, using what
data were available for 1996 and, where necessary, making assumptions
concerning minimum numbers--suggests that the annual flow of
nonimmigrant residents could easily be over 500,000 and might be
considerably higher.  However, unsubstantiated assumptions were
sometimes necessary; e.g., in making minimum and maximum
calculations, we assumed that between 35 and 80 percent of the
student visas issued by the State Department represented a new
resident who remained for over a year. 


      GAP FOR TOTAL FLOW OF
      FOREIGN-BORN RESIDENTS INTO
      THE UNITED STATES
-------------------------------------------------------- Chapter 3:3.2

From the perspective of charting the overall impact of immigration on
American society, the most important statistic for which there is a
gap may be the total number of new entries who take up residence in
the United States each year (total for col.  1 in table 3.1).  To
fill this gap would require dealing with the gaps and biases in the
various estimates for new entries in the various legal status
categories:  LPRs; refugees and asylees; foreign students, temporary
workers, and others here legally for a nonpermanent stay; and illegal
residents.  Indeed, approximate figures developed for lower and upper
bounds could differ by as much as one million persons--leaving
policymakers still without a proper information base.\9


--------------------
\9 We attempted to develop approximate figures by listing all
relevant categories, using what data were available for each and,
where necessary, making assumptions for lower and upper bounds.  We
did not report the specific figures because of their speculative
nature. 


THE RESIDENT FOREIGN-BORN
POPULATION:  SIZE, NET CHANGE, AND
EMIGRATION
============================================================ Chapter 4

The Bureau of the Census provides decennial census and intercensal
survey data on the size of the foreign-born population and change in
size, and it estimates emigration.  Although there are no directly
relevant administrative records, INS provides some additional
information through "composite estimates."\1 Overall, information on
the size of the foreign-born population, change in size, and
emigration is limited, as summarized in table 4.1.  There are two
main reasons for this: 

  -- Decennial census and survey estimates apply only to total
     foreign-born and naturalized citizens (rows E and total in table
     4.1).\2 For specific legal statuses, there are gaps and
     uncertain estimates (rows A through D in table 4.1).  Aside from
     a question on U.S.  citizenship, the census and intercensal
     surveys do not ask about legal status.  (One reason is that
     questions on the respondent's legal status are very sensitive
     and might result in biased answers or affect responses to other
     questions.) INS efforts to fill data gaps without additional
     data collection efforts have resulted in uncertain estimates. 

  -- The data provided by Census for total foreign-born and
     naturalized U.S.  citizens have not been rigorously evaluated. 
     The unevaluated estimates are at least somewhat uncertain
     because of the questions about adequate coverage of the
     foreign-born population and other quality issues that have been
     raised by a number of analysts.  Moreover, problems have cropped
     up in estimating emigration--and also in reliably quantifying
     net change. 

We also found that the information in table 4.1 cannot be accessed by
referring to just one or two publications.  Rather, policymakers and
other information consumers must first identify and access a variety
of sources (including the Internet), then piece results
together--with no guide to their comparability. 



                                        Table 4.1
                         
                          Published Federal Statistics for Three
                            Demographic Concepts, Descriptive
                            Quality Rating and Reason, 1996\a

                                              Demographic concept
                      -------------------------------------------------------------------
                         Population size      Net change in size      Annual emigration
                      ---------------------  ---------------------  ---------------------
                      Number of foreign-     Annual increase or     Number of foreign-
                      born persons residing  decrease in the size   born residents who
Legal status of       here (at given point   of foreign-born        left the U.S.
foreign-born          in time)               population             (calendar 1996)
--------------------  ---------------------  ---------------------  ---------------------
A. Legal permanent    10,525,000 LPRs\b      No published           No published
residents                                    estimate.              estimate.

                      Uncertain: Composite   Gap: No                Gap: No
                      estimate requires      administrative or      administrative or
                      major assumptions.     survey data.           survey data.

B. Refugees and       No published           No published           No published
asylees               estimate.              estimate.              estimate.
(without LPR
status)\c

                      Gap: No comprehensive  Gap: No                Gap: No
                      administrative data;   administrative or      administrative or
                      questions about legal  survey data.           survey data.
                      status deemed too
                      sensitive to ask
                      survey respondents.

C. Students,          No published           No published           No published
temporary workers,    estimate.              estimate.              estimate.
and their families;
others here
temporarily who
stayed >1 yr\d

                      Gap: No usable         Gap: No                Gap: No
                      administrative data;   administrative or      administrative or
                      questions about legal  survey data.           survey data.
                      status too sensitive
                      to ask in a survey.

D. Illegal            5 million EWIs and     275,000 more illegal   No published
immigrants (EWIs and  overstays\e            residents\f            estimate.
overstays) who
stayed
> 1 year

                      Uncertain: Composite   Uncertain: Composite   Gap: No
                      estimate requires      estimate requires      administrative or
                      major assumptions.     major assumptions.     survey data.

E. Naturalized        7.9 million            1.1 million more       No published
citizens              naturalized            naturalized citizens   estimate.
                      citizens\g             (3/97 versus 3/
                                             96)\h

                                             300,000 more
                                             naturalized citizens
                                             (3/96 versus 3/95)\i

                Unevaluated: Survey    Uncertain: These       Gap: No
                      estimate               survey estimates       administrative or
                      is disputed in         involve time lag in    survey data.
                      literature owing       reporting.
                      to possible
                      overreporting of
                      citizenship not
                      firmly resolved.

Total                 24.6 million foreign-  1.2 million more       No published
                      born residents\j       foreign-born           estimate.
                                             residents (3/97
                                             versus 3/96)\k

                                             0.1 million more
                                             foreign-born
                                             residents (3/96
                                             versus 3/95)\l

                      Unevaluated: Analysts  Uncertain: Survey      Gap: Census first
                      have raised questions  estimates of year-     said estimate of
                      about survey           to-year change are     emigration (195,000)
                      estimates'             imprecise.             referred to all
                      underrepresentation                           foreign-born; Census
                      of newcomers, etc.                            now says it refers to
                                                                    emigration of legal
                                                                    residents only.
-----------------------------------------------------------------------------------------
Legend

EWI Entry without inspection
LPR Legal permanent resident

\a The three demographic concepts are size of the foreign-born
population, net change in size, and emigration. 

\b As of Apr.  1996; range:  10.175 to 10.875 million LPRs. 

\c The term asylees refers to persons granted asylum. 

\d Other foreign-born residents admitted temporarily include (1)
trainees, treaty-traders, and investors as well as their families;
(2) special groups, such as those given temporary protected status or
extended voluntary departure; (3) parolees, that is, persons allowed
to enter the United States on a nonpermanent basis for urgent
humanitarian reasons or when significant public benefit is likely;
and (4) persons entering on tourist or other short-term visas who
nonetheless remain in the United States longer than a year. 

\e As of Oct.  1996; range:  4.6 to 5.4 million EWIs and overstays;
best estimate for EWIs (only), 2.9 million; best estimate for
overstays (only), 2.1 million.  (Note:  These INS estimates of EWIs
and overstays include some persons who were illegal immigrants but
attained a special category status, such as extended voluntary
departure.)

\f Annual average increase (1992-96) in EWIs and overstays combined. 
(Note:  These INS estimates of EWIs and overstays include some
persons who were illegal immigrants but attained a special category
status, such as extended voluntary departure.)

\g As of Mar.  1996; 90-percent confidence interval:  7.5 to 8.3
million naturalized citizens. 

\h 90-percent confidence interval:  680,000 to 1.6 million more
naturalized citizens. 

\i 90-percent confidence interval:  240,000 fewer to 800,000 more
naturalized citizens. 

\j As of Mar.  1996; 90-percent confidence interval:  23.9 to 25.3
million foreign-born residents. 

\k 90-percent confidence interval:  460,000 to 1.99 million more
foreign-born residents. 

\l 90-percent confidence interval:  720,000 fewer to 840,000 more
foreign-born residents. 

Source:  Published federal statistics on the foreign-born are from
the INS Web page; INS Yearbook, pp.  197-98; Current Population
Reports, Bureau of the Census, Mar.  1997; Schmidley and Robinson,
1998, table A3.  Additionally, the confidence intervals in notes to
this table were provided to us by the Bureau of the Census.  The
ratings are based on agency interviews, review of the literature, and
logical analysis. 


--------------------
\1 We define "composite estimates" as those made in the absence of
direct data.  Such estimates draw on what data are available--often a
combination of different kinds of data and/or different kinds of
estimates--plus major assumptions.  (By major assumptions, we mean
assumptions that can have important impacts on the estimates.)

\2 Census and survey estimates for total foreign-born would logically
include residents in all major legal statuses, including those with
temporary legal status as well as illegal immigrants.  We believe
that such estimates of total foreign-born are probably generally
consistent with our definition of foreign-born residents--that is,
persons living here who either (1) are in a permanent legal status or
(2) if in a temporary legal status or here illegally, remain in the
United States for more than a year. 


   THE AVAILABLE DATA
---------------------------------------------------------- Chapter 4:1

The Alien Address Report Program (an annual registration system
maintained by INS) was discontinued in 1981.  Until its demise, that
system was a source of administrative data on the population of
legally resident aliens (i.e., noncitizens).  Since all resident
aliens were required to register, the number living here would be
represented by the number registering--provided that all legal aliens
complied.  The registration system was discontinued partly for
budgetary reasons, but also because not all aliens reported, and the
value of the information was unclear.  Likewise, administrative
systems to record emigration by aliens and U.S.  citizens, begun
early in this century, were discontinued (in the 1950s) partly
because they were believed to underestimate permanent departures. 

After the Alien Address Report Program was discontinued, the
decennial census represented, until recently, the only remaining
regularly scheduled collection of data on the size of the
foreign-born population.  Occasional supplements to the Current
Population Survey (CPS) included questions on nativity and
citizenship.  Then, starting in 1994, such questions were added to
the CPS on a regular basis, providing information on the foreign-born
population in intercensal years.\3


--------------------
\3 Concurrent with the addition of these questions to the CPS on a
regular basis, cognitive interviewing was conducted to evaluate and
improve the questions on nativity and citizenship.  (Cognitive
interviewing is a method of testing how respondents interpret survey
questions; e.g., a test respondent is asked to "think aloud" while
coming up with an answer to a question.  The results are used to
revise questions, as needed, in order to more clearly communicate
with respondents.)


   GAPS AND UNCERTAIN ESTIMATES
---------------------------------------------------------- Chapter 4:2


      LACK OF CENSUS AND SURVEY
      DATA ON LEGAL STATUS
-------------------------------------------------------- Chapter 4:2.1

Neither the census nor the CPS asks about the legal status of
noncitizens--or whether they are, in fact, here illegally.  There are
good reasons for this:  such questions fall under the heading of
"threatening" survey questions (Bradburn and Sudman, 1979); many
respondents might not answer these questions truthfully; and others
might avoid participating altogether if they hear that such questions
will be asked.  In addition, the Bureau of the Census is concerned
about privacy invasion issues.\4


--------------------
\4 The Bureau is concerned that its role not include obtaining (and
maintaining) potentially damaging information from survey respondents
or persons who cooperate in a decennial census. 


      UNCERTAIN COMPOSITE
      ESTIMATES
-------------------------------------------------------- Chapter 4:2.2

In an effort to fill data gaps, INS developed "composite estimates"
for the number of illegal residents and for the number of legal
permanent residents.  Although these estimates represent a step
forward (because they provide some information that would otherwise
not be available), they are necessarily uncertain.  That is, it is
difficult to determine whether the figures might be underestimates or
overestimates and to judge what the magnitude of misestimation might
be. 

INS' composite estimate of the current number of illegals residing
here is based mainly on the following three calculations--each of
which is characterized by uncertainty: 

  -- First, INS calculates an estimate of illegal
     "overstays"--persons who entered legally on a temporary basis
     and failed to depart.  These estimates are uncertain for several
     reasons:  INS' data system does not track many legal entries by
     Mexicans and Canadians (Department of Justice, 1997); so if such
     persons overstay, they would not be counted.  Although the
     system records other persons' legal entries and departures, a
     substantial portion of the departure data is missing each year,
     and the assumptions INS uses to differentiate missing departure
     data from actual overstays are controversial.\5 Moreover, INS
     made its current (1996) estimate by projecting old overstay
     estimates forward (Department of Justice, 1997).  (Data
     collected after 1992 are deemed not usable because of computer
     processing problems and lack of agreement on what constitutes a
     valid match.)

  -- Second, INS calculates the total number of Mexican illegals by
     comparing administrative data on Mexican legal immigrants to CPS
     data on total Mexican foreign-born.  Here, uncertainty derives
     from not knowing survey underrepresentation of illegals and from
     questions about the estimate of emigration (see INS Yearbook).\6

  -- Third, INS estimates the number of non-Mexican residents who
     "entered without inspection" (i.e., EWIs from other countries
     around the world) based on a variety of data, including data
     from the late 1980s amnesty (that IRCA provided) as well as more
     recent data on trends in apprehensions.  Translating such data
     into an estimate of the number of current residents necessarily
     involves assumptions and uncertainty.\7

Turning to the size of the LPR population, this INS estimate is based
on an indirect method that includes subtracting the estimated number
of illegals (just discussed) from the number of foreign-born aliens
(i.e., noncitizens) estimated in the CPS.  Thus, the uncertainties
regarding the estimate of illegals are necessarily carried over to
the estimate of the number of foreign-born residents with green
cards. 

With respect to the INS estimate of net change in the size of the
illegal population from year to year (275,000), we note that this
estimate is derived by comparing composite estimates for two points
in time (October 1992 and October 1996) and dividing the total change
into equal amounts of change for each year.\8 Hence, this estimate is
marked by the uncertainty of the composite estimates of the size of
the illegal population.  It also reflects a general level of change
rather than depicting current trends. 


--------------------
\5 For a critique of INS' method of calculating overstays, see
GAO/PEMD-95-10. 

\6 The Mexican overstay data are used to divide the estimate of total
Mexican illegals into overstays and EWIs. 

\7 We also note that the INS estimate of illegal immigrants includes
some persons in special categories (such as temporary protected
status) whom we have included as long-term temporary residents. 

\8 The INS estimates are for all illegals; the Bureau of the Census
estimates annual net change for the counted population of illegals. 


   EXISTING CENSUS AND SURVEY
   DATA:  LACK OF EVALUATION
---------------------------------------------------------- Chapter 4:3

Valuable as the census and CPS data are--or can be--for estimates of
the size of the foreign-born population and other demographic
concepts, various analysts have raised questions about quality.  The
Bureau of the Census believes that the foreign-born are less likely
to be enumerated than the native-born, but a rigorous evaluation has
not been conducted. 


      QUESTIONS ABOUT
      UNDERREPRESENTATION OF THE
      FOREIGN-BORN IN THE CENSUS
-------------------------------------------------------- Chapter 4:3.1

The Bureau of the Census has carefully evaluated the quality of 1990
census data in the Post Enumeration Survey (PES), but the PES does
not distinguish foreign-born residents from native-born.  A wide
variety of factors have been hypothesized--or identified in indirect
analyses and qualitative studies--as contributing to
underrepresentation of the foreign-born (see table 4.2).  However, it
is also possible that reverse errors could occur (Jasso and
Rosenzweig, 1987; Schmidley and Robinson, 1998), and in the absence
of a rigorous evaluation targeting the foreign-born, the level of net
underrepresentation is unknown.  For the 1990 census as a whole,
however, a rigorous evaluation indicated that undercounting was the
more important factor.  (For a discussion of the net undercount in
the 1990 census and the gross levels of overcounting and
undercounting that occurred, see GAO/GGD-91-113.)



                               Table 4.2
                
                   Reasons for Underrepresentation of
                 Foreign-Born in a Census or Survey and
                            Type of Evidence

                                                    Type and source of
Reason                                              evidence
--------------------------------------------------  ------------------
Some long-term residents (e.g., those here on       Hypothesis
temporary visas) may not identify the United        (Hollman, 1997;
States as their usual place of residence--and thus  Passel and Clark,
would be omitted from the census.                   1998)

In ad hoc households of recently arrived illegal    Qualitative
immigrants with no special ties to each other,      studies (de la
omissions on the household roster are more likely;  Puente, 1993;
in extreme cases, more than 10 men might share a    Velasco, 1992)
single unit--working and sleeping in shifts.
Indeed, the person speaking with the enumerator
may not really know who else lives there.

Some illegals intentionally avoid enumeration.      Qualitative study
                                                    (de la Puente,
                                                    1993)

Recent immigrants are likely to be boarders in the  Small qualitative
homes of a more established immigrant family--and   studies (McKay,
thus may be omitted from household rosters. Also,   1993; McKay et
Mexicans from villages do not recognize the         al., 1996)
Spanish word for boarders ("inquilinos"), which is
used in surveys; they use the "Spanglish" term
"renteros."

Some immigrant households are hidden from           Qualitative
enumerators, such as when a newly arrived family    studies (Gabbard
rents a back bedroom in an apartment--in violation  et al., 1993; de
of the lease--and the lease holder omits the extra  la Puente, 1993;
persons when enumerating the household. Other       Davies, 1995;
examples include families in hidden makeshift       Chapa and del
shacks, converted garages, or colonias              Pinal, 1993)
(shantytowns that lack streets, electricity, and
other services).

Some Mexican immigrants may falsely claim U.S.      Demographic
birth.                                              analysis (Warren
                                                    and Passel, 1987;
                                                    Woodrow, 1991;
                                                    Woodrow and
                                                    Passel, 1990)\a
----------------------------------------------------------------------
\a As an example of demographic analysis, Woodrow used the number of
Mexican-origin U.S.-born persons reported in, e.g., 1980 and data on
subsequent U.S.  births to derive an expected level of increase in
the Mexican-origin U.S.-born population as of, e.g., 1989.  The
expected increase was compared to survey data collected in 1989.  An
unexpectedly high number of Mexican-origin persons claimed U.S. 
birth in the 1989 survey--a pattern that might be explained by
immigration combined with false reports of U.S.  birth. 

Underrepresentation is thought to be concentrated in certain groups
of the foreign-born, such as newcomers and illegals, who currently
constitute perhaps one-fifth of all foreign-born residents. 
Unofficial estimates by Census staff put the undercount of illegal
immigrants at about 33 percent in the 1980 census; Passel (1986) has
suggested a range between 33 and 50 percent.  For the 1990 census,
various anaylses put the figure at roughly 20-30 percent (Woodrow,
1991; Van Hook and Bean, 1997; Woodrow- Lafield, 1995.)


      POSSIBLE UNDERREPRESENTATION
      OF THE FOREIGN-BORN IN THE
      CPS
-------------------------------------------------------- Chapter 4:3.2

Post Enumeration Survey results are used to adjust the Current
Population Survey for misrepresentation of specific groups defined by
age, sex, race or Hispanic origin, and state of residence.  However,
it is not known whether the PES adjustments sufficiently improve
representation of the foreign-born.  If there were no differences
between the coverage of foreign-born and native-born persons of the
same age, sex, race, and so forth, the PES adjustments should produce
accurate CPS estimates of the foreign-born.  But as delineated in
table 4.1, foreign-born persons may be less likely to be found or
identified as residents than native-born persons are, and it is
possible that some foreign-born persons may also falsely claim U.S. 
birth.  Thus, despite the PES adjustments, the CPS data could
underrepresent the foreign-born.\9

A different issue could contribute to added underrepresentation of
the foreign-born in the CPS:  survey nonresponse.  As in any survey,
some sampled CPS households do not respond to the CPS.\10 CPS
response rates are calculated (and adjustments to correct for
nonresponse are applied) to geographically large areas--on average,
about five areas per state.\11 If foreign-born residents are as
likely to respond to the CPS as native-born persons who reside in the
same area, there is no problem.  But this may not be the case;
possible reasons for lower response rates among certain groups of
foreign-born include interviewer problems communicating with
non-Hispanic immigrants; possible distrust of government or strangers
among certain groups (those illegally here, asylees from repressive
countries); and for some groups of new immigrants, less familiarity
with polling.  Thus, higher rates of nonresponse among the
foreign-born may contribute to underrepresentation.  (An analysis of
nonresponse in subareas where foreign-born are concentrated would
settle the issue.)


--------------------
\9 When using CPS data on Mexicans to estimate illegals, the
Commission on Immigration Reform and Secretaria de Relaciones
Exteriores (1997) used an additional adjustment to correct for
underrepresentation. 

\10 The CPS nonresponse rate is less than 10 percent overall,
including foreign-born and native-born.  It is known to differ by
geographic area; e.g., Los Angeles has a much higher nonresponse rate
than Seattle. 

\11 To illustrate how weights compensate for nonresponse, data from
three areas with response rates of 100, 90, and 50 percent would
receive relative weights of 1, 1.1, and 2, respectively. 


      QUESTIONS ABOUT THE
      ESTIMATION OF NATURALIZED
      CITIZENS
-------------------------------------------------------- Chapter 4:3.3

Turning to estimates of the number of naturalized citizens, the main
concern is not with undercoverage (because citizens are likely to be
counted) but rather with the potential for overrepresentation because
of false claims of citizenship.  This issue is currently being
disputed in the literature (Passel et al., 1998; Schmidley and
Robinson, 1998).  We also note that the CPS, which reinterviews
respondents over a 16-month period, has thus far asked the
citizenship question only at the first interview.  Consequently, the
CPS estimates omit some of the most recent naturalized citizens. 
Currently, this is important because of large numbers of
naturalizations--over 1 million persons were naturalized in fiscal
year 1996.  The Census Bureau is now considering asking the
citizenship question in every CPS interview. 


      IMPRECISION AND POSSIBLE
      VOLATILITY IN ESTIMATES OF
      YEAR-TO-YEAR CHANGE
-------------------------------------------------------- Chapter 4:3.4

Census staff have stated that "The CPS nativity data provide a
reliable basis for tracking change in the size of the total
foreign-born population at the national level" (Schmidley and
Robinson, 1998, p.  17).  But various Census staff told us that
year-to-year change in the size of the total foreign-born population
could--and alternatively that it could not--be reliably measured by
CPS data.  Current estimates of year-to-year net change in the size
of the foreign-born population (see table 4.1, total row) seem
imprecise; indeed, the confidence intervals are so broad that the
estimates might be deemed too imprecise for policy-making purposes. 
For example, as shown in table 4.1, the 90-percent confidence
interval for 1996-97 net change in the size of the foreign-born
population ranges from an increase of under a half million to an
increase of about 2 million.  In other words, the foreign-born
population (24 million) may have increased by as little as 2 percent
or as much as 8 percent--in a single year. 

At our request, Census staff prepared an "annual averages" estimate
for the 1996 to 1997 change.\12 The annual averages estimate is,
again, an increase of 1.2 million; the confidence interval (800,000
to 1.7 million) is smaller, but still seems imprecise. 

It is also somewhat troubling that the trends in year-to-year change
appear to be volatile--going from near zero in one year to an
increase of over a million the next.  It is not known whether some
degree of real change occurred, whether an artifact caused the
result, or whether the very large difference in estimates is simply
from sampling error. 


--------------------
\12 Such an estimate is more stable because it draws on larger
samples--i.e., it includes the responses of all persons interviewed
throughout the 12 months of each year.  A separate estimate of the
size of the foreign-born population is prepared based on data for
each month of the relevant year, and then an average is taken across
all 12 months.  This is done separately for each year, and the
difference between the years is calculated. 


      PROBLEMS IN ESTIMATING
      EMIGRATION
-------------------------------------------------------- Chapter 4:3.5

In using 1980 and 1990 census data to estimate emigration of
foreign-born persons, analysts at the Bureau of the Census
encountered inconsistent results--apparently because of coverage
problems.  Bureau of the Census analysts attempted to estimate
emigration by tracking arrival cohorts--for example, Mexicans who
came to the United States to live during the 1970s--across the 1980
and 1990 censuses.  The logic was that by observing the extent to
which the size of a cohort dwindled between 1980 and 1990 (while
accounting for deaths), one could infer the level of emigration.\13
The approach is logical and reflects basic procedures of demographic
analysis.  But no dwindling was apparent for Mexico and several other
countries; instead, cohorts that arrived in the 1970s appeared to
grow between 1980 and 1990, thus yielding negative estimates of
emigration--a logical impossibility.\14

Census staff determined that data for certain countries were
unusable, and emigration rates were calculated only for residents
from countries with usable data (such as Spain).  These rates were
then extrapolated to countries with unusable data (such as
Mexico).\15

The result was an estimate of 195,000 emigrants each year.  But a
number of uncertainties are involved; notably, emigration rates may
differ for legals and illegals, and if so, extrapolation from
countries such as Spain (with mostly legal entries) to countries like
Mexico (with large numbers of illegal entries) would be
inappropriate.  Census staff have recently determined that--in
hindsight--the 195,000 is best interpreted as an estimate of
emigration on the part of legal foreign-born residents only.\16 One
possible problem with this interpretation is that in calculating the
195,000, the Bureau extrapolated emigration rates to all foreign-born
residents--a group that includes some illegals.\17


--------------------
\13 An estimate of deaths is subtracted out. 

\14 A similar pattern appears to hold for the 1990s.  That is, a
recent paper by Census staff indicates that for Mexico and other
countries in this hemisphere, the number of persons who say they
arrived before April 1990 appears to have grown slightly (rather than
dwindled) between the 1990 census and the 1997 CPS (see table A.10 of
Schmidley and Robinson, 1998). 

\15 Specifically, it was assumed that the emigration rate for
residents from countries such as Mexico was half that for countries
such as Spain or Argentina (chiefly because economic conditions in
countries like Mexico make returning there less attractive). 

\16 The Census paper containing the 195,000 figure does not state
this qualification, which emerged only after we questioned staff
about their figure for net migration.  The estimate of 195,000 is
used in calculations of net migration that are used in intercensal
estimates of the total population (and are control totals for the
CPS).  In those calculations, the 195,000 estimate is treated as an
estimate of emigration by legals (see Hollman, 1998). 

\17 Census staff agreed that a small number of emigrating illegals
were included in the 195,000; they believed that the resultant
overestimation of emigrating legals was probably small. 


   REPORTING DATA ON SIZE OF THE
   FOREIGN-BORN POPULATION, CHANGE
   IN SIZE, AND EMIGRATION
---------------------------------------------------------- Chapter 4:4

A policymaker or interested member of the general public would have
to access four disparate sources to obtain the estimates shown in
table 4.1:  (1) the 1996 INS Statistical Yearbook, (2) the INS Web
page, (3) an issue of Current Population Reports, and (4) a Bureau of
the Census "working paper" on trends and methodological issues.  In
no case does any one publication refer to all the others, and we
could find no central source pointing the interested person to all
four.  In some cases, less widely distributed publications are needed
to understand the methodological bases of the figures.  The lack of a
central publication relating the results found in one source to those
found in another means that readers must gauge the relative quality
and comparability of the various estimates, and their
interrelationship, for themselves. 

In some cases, reporting has not provided the complete information
that readers need to judge the quality or stability of the estimates. 
For example, the Bureau of the Census working paper, entitled "How
Well Does the Current Population Survey Measure the Foreign-born
Population in the United States," reports that the nonresponse rate
for the CPS is about 6.5 percent, but does not let readers know
whether the level of response for communities or areas dominated by
foreign-born residents is roughly the same as for other areas.  To
cite another example, the description of the estimates of illegal
overstays in the INS 1996 Statistical Yearbook fails to inform
readers that data on overstays have not been available since 1992 and
that the 1996 estimates of illegals were achieved by projections of
data from earlier years. 


IMPROVING INFORMATION:  CURRENT
INITIATIVES AND NEW STRATEGIES
============================================================ Chapter 5

INS has some initiatives underway to fill data gaps, including an
attempt to develop a measure of illegal flow into the United States. 
In addition, we have identified strategies to improve census and
survey data on the size of the foreign-born population.  These
include (1) eliminating data gaps by collecting survey information on
immigration status through a less sensitive form of questioning--the
three-card method--and (2) achieving greater certainty for estimates
of total foreign-born and naturalized citizens through evaluative
analyses, and where needed, corrective adjustments. 


   CURRENT INITIATIVES FOR
   IMPROVING INFORMATION
---------------------------------------------------------- Chapter 5:1

A number of new initiatives are underway at INS.  Notably, INS is
attempting to improve its data on foreign students--a key group of
long-term temporary residents.  INS is also attempting to develop a
measure of illegal flow into the United States and a new measure of
the size of the illegal population.  These initiatives do not address
all of the problems identified in chapters 3 and 4, however. 

  -- For foreign students, INS has created a new approach to
     record-keeping that should allow estimates of the number of new
     students who come to reside here each year, the size of the
     student population, and net change in size.  A unique high-tech
     identification card (with digitally encrypted identifying
     information) is prepared for each student in advance and issued
     to him or her upon arrival in the United States.  The special
     card is used for each entry and exit.  INS is pilot-testing that
     approach by tracking a group of 10,000 students--that is,
     following them individually across time--from prearrival through
     years of schooling, including exits from the United States and
     reentries.\1 (Roughly a half-million students resided here
     during the 1995-96 school year; U.S.  Department of Education,
     1997.)

  -- To fill the information gap for illegal flow, INS is currently
     working to develop a new composite estimate.  The new estimation
     method will utilize CPS data on the number of foreign-born
     residents and their reported year of entry to the United States. 
     It will also utilize INS administrative records for data on the
     number of new legal residents.  Assuming that the method is
     successful, it will produce estimates of flow and trends in flow
     that include both EWIs and overstays who remain in the United
     States longer than a year.  (The new method will not provide
     separate estimates for the flow of EWIs and the flow of
     overstays, however.) The approach also produces an estimate of
     the size of the illegal population without relying on the usual
     overstay data. 

With respect to improving the reporting of information on flow, INS
statistical staff told us that, intermittently, they have discussed
plans for an analytic report that is oriented to policy-information
needs.  However, due to lack of staff time, such plans have always
been shelved. 

We know of no new initiatives or special analyses being conducted or
planned at the Bureau of the Census.  The agency told us that its
budget and mission do not allow substantial resources to be devoted
to estimates of the foreign-born.  However, Census also told us that
if the planned American Community Survey is implemented, it will
include questions on nativity and U.S.  citizenship.  The resulting
estimates of the foreign-born population would be more stable than
CPS estimates because a very large sample is planned for this survey. 
(Statistically, the larger the sample, the more likely it is that the
resulting estimate will fall close to the value that would be
obtained in a complete census.)


--------------------
\1 Improved tracking is not only a statistical issue.  Tracking of
persons here on student visas is also of concern from a law
enforcement and intelligence perspective (see Hearing before the
Senate Subcommittee on Technology, Terrorism, and Government
Information, Committee on the Judiciary, Feb.  24, 1998).  Tracking
of tourist entries and exits is also needed to monitor the Visa
Waiver Pilot Program, which allows tourists from several foreign
countries to visit the United States without a visa.  By law, the
waiver for an individual country must be revoked if the number of
overstays exceeds 2 percent of tourist entries.  (See GAO/PEMD-95-20
and U.S.  Department of Justice, 1997.)

INS also told us that it (1) is pilot-testing an automated
entry-and-exit tracking system for all classes of nonimmigrants at
two airports (and is planning to expand this test to 10 airports) and
(2) is continuing its attempts to solve the computer and other
problems that have prevented reliable matches of entries and exits
since 1992.  However, even if successful, these efforts would not
address the issue of distinguishing reentries from new entries--which
is crucial for estimating the number of foreign-born persons who
newly take up residence here each year. 


   STRATEGIES FOR IMPROVING CENSUS
   AND SURVEY DATA ON THE
   FOREIGN-BORN
---------------------------------------------------------- Chapter 5:2

Strategies for improving census and survey information include (1)
research on a new survey technique that asks less sensitive questions
about immigration status (the three- card method) and (2) low-cost
evaluations of existing census and survey data on the foreign-born. 


      SURVEY DATA TO FILL GAPS FOR
      LEGAL STATUS:  THE
      THREE-CARD METHOD
-------------------------------------------------------- Chapter 5:2.1

To confirm or improve existing uncertain estimates--and to fill data
gaps--we determined that a new approach to asking survey respondents
about their legal status might be helpful.  Previously developed
methods for reducing question threat and protecting respondent
privacy (see app.  III) do not seem suitable for use with the
foreign-born population.  We therefore devised a new method that
builds on earlier techniques but requires no unusual interview
procedures.  We discussed this method (which we term the three-card
method) with INS staff and with officials and staff at the Bureau of
the Census.  We also discussed it with a number of statistical and
survey experts.  And we conducted a preliminary qualitative test of
its acceptability to respondents and interviewers. 

The three-card method is designed to encourage truthful responses
concerning legal status--yet no foreign-born person need reveal
whether he or she (or anyone else) is an illegal immigrant.  Indeed,
no one could ever discover, even through a series of deductions,
whether a respondent was here illegally.  Yet when all data are
combined, survey-based estimates are available for all legal
statuses--including illegal immigrants. 

The key to the approach is to use three random subsamples of persons
from the foreign-born population (different individuals in each
subsample).  The following explanation uses a hypothetical example
that will be carried through for each of the three cards.  The
hypothetical example helps explain how an estimate of illegal
immigrants is obtained, once information from all responses is
gathered and combined. 

Respondents in the first subsample are shown a card with three boxes
such as that shown in figure 5.1.  The card in figure 5.1 is in
Spanish because we pretested the three-card method in Spanish.  The
English translation is: 

  -- Box A:  legal permanent resident (with an official, validly
     obtained green card);

  -- Box B:  U.S.  citizen; student, work, or tourist visa;
     undocumented; and refugee or asylee;

  -- Box C:  some other category, not in Box A or Box B (specify). 

Respondents are asked which box applies to them and are told that if
it is Box B, we do not want to know which specific category applies
to them.  For respondents who pick Box B, it is the end of the
interview.  However, for respondents who pick Box A, a number of
follow-up questions are possible.  The purpose of interviews with the
first subsample is to obtain a valid estimate of the percent of
foreign-born who have officially obtained green cards.  (Hypothetical
example:  35 percent of the resident foreign-born population have
official green cards.)\2

   Figure 5.1:  Card 1 of the
   Three-Card Method

   (See figure in printed
   edition.)

Respondents in the second subsample (completely different
individuals) are shown a different card (see fig.  5.2).  It also has
three boxes.  The difference is that Box A shows the status of a
(naturalized) U.S.  citizen, whereas Box B includes legal permanent
resident (LPR) along with other immigration statuses.  Respondents in
this subsample are also asked to pick the box that applies to
them--and told that if it is Box B, we do not want to know which
category applies to them.  The purpose of interviewing this sample
with this card is to obtain a valid estimate of the percentage of the
foreign-born who are naturalized citizens.  (Hypothetical example: 
30 percent are naturalized citizens.)

   Figure 5.2:  Card 2 of the
   Three-Card Method

   (See figure in printed
   edition.)

Respondents in the third subsample (again, different persons) are
shown yet another card (see fig.  5.3).  It also has three boxes, but
this time, Box A features refugees and persons granted asylum, as
well as residents here legally with temporary visas.  Those who pick
Box A are asked follow-up questions to determine their exact legal
status.  Respondents are told that if they are in Box B, we do not
want to know which specific category applies to them.  The purpose of
interviewing this sample is to get estimates of the percentages in
the categories in Box A.  (Hypothetical example:  5 percent are
refugees or asylees; an additional 5 percent are here on temporary
visas.)

   Figure 5.3:  Card 3 of the
   Three-Card Method

   (See figure in printed
   edition.)

Assuming that the categories are mutually exclusive and exhaustive,
it is possible to obtain an estimate of illegals.  That is, extending
the hypothetical examples above, we would estimate that 75 percent of
the foreign-born are in the major legal statuses (35% + 30% + 5% + 5%
= 75%).\3 Suppose also that 1 percent picked Box C (some other
category).  Subtracting these hypothetical estimates from 100 percent
yields

          100% - 75% - 1% = 24%. 

In this hypothetical example, an estimated 24 percent did not claim
to be in any legal status, and the implication is that 24 percent of
the foreign-born population are illegal immigrants. 


--------------------
\2 Respondents who pick Box C are asked what their specific
immigration status is.  In our preliminary test, no respondents
picked Box C. 

\3 As worded in figs.  5.1-5.3, the category for refugees and asylees
may overlap with the LPR category.  (A refugee who has attained
green-card status may identify as both a refugee and a person with a
valid green card.) Thus, we suggest that in future field tests, the
cards be modified, adding the phrase "without a green card" to the
refugee-asylee category. 


      PRELIMINARY TEST OF THE
      THREE-CARD METHOD
-------------------------------------------------------- Chapter 5:2.2

We conducted a preliminary test of the acceptability of the card
shown in figure 5.1--that is, its acceptability to interviewers and
respondents.  Our test population consisted of foreign-born
farmworkers, chosen because it is thought to have a high percentage
of illegals and because almost all speak Spanish.  This population
also has few naturalized citizens--only about 3 percent of
foreign-born respondents claimed to be naturalized citizens in the
1994-95 National Agricultural Workers Survey (Mines et al., 1997). 
Very few farmworkers are here on legal temporary visas.\4 Nearly one
million farmworkers acquired valid green cards under the IRCA amnesty
of the late 1980s, although many of them have now moved to other
types of work.  Thus, we expected that most foreign-born farmworkers
would either be illegals or have valid green cards. 

The specific questions for use with the cards and the icons shown on
the cards were developed through a series of pretests that we
conducted with farmworkers and other foreign-born Hispanics.  We
conducted the pretests using staff who are fluent in both Spanish and
English. 

The preliminary test consisted of a contract with the survey firm
that conducts the National Agricultural Workers Survey for the
Department of Labor.  Under this contract, 81 interviews featuring
the card shown in figure 5.1 were added to the 1997-98 National
Agricultural Workers Survey.  Each interview was with a foreign-born
farmworker.\5 All chose to be interviewed in Spanish.  None of those
interviewed refused to answer the question on immigration status,
using the card shown in figure 5.1.  Of the 81 respondents, 30 picked
Box A, claiming to have a valid green card.  The remaining 51
respondents chose Box B, which contains the illegal category as well
as various legal statuses.  None chose Box C. 

In debriefing sessions, the interviewers told us that overall, the
method was acceptable to them and to respondents.  Respondents did
not voice suspicion of the 3-box card.  Rather they appeared to
accept it at face value.  No special explanations were needed about
why we were asking the questions.  Criticisms focused on points such
as improving the icons (symbolic pictures) used on the cards.  (More
details concerning the three-card method are included in app.  III.)


--------------------
\4 According to our recent report, in fiscal year 1996 only about
15,000 farmworkers were brought into the United States with H-2A
visas for temporary agricultural work, and this represents less than
1 percent of the agricultural field labor force (GAO/HEHS-98-20). 

\5 The contractor did not provide the identities of farm employers to
us. 


      POSSIBLE USE OF THE METHOD
      IN CURRENT SURVEYS
-------------------------------------------------------- Chapter 5:2.3

We also considered how the three-card method might be used with
foreign-born respondents in a more conventional federal household
survey.  There are two constraints.  First, as currently developed,
the technique is used in a face-to-face personal interviewer survey,
rather than a telephone survey.  Second, federal agencies are
generally concerned that the sensitivity of a question on legal
status (even using the three-card method) might affect responses and
so would prefer that it be the last question asked--or that it be
used in a linked follow-back survey.\6 The latter is
expensive--costing perhaps as much as $300 per interview, according
to Census staff.  Eventually, if the three-card method were
demonstrated to be nonthreatening, it might be possible to
incorporate it as part of a regular federal survey, which would be
much less expensive. 

However, we believe more testing is needed before the technique is
used in a large-scale survey.  Such testing should be done under
conditions similar to those of a federal household survey, and not
only farmworkers should be tested.  The tests should further explore
the acceptability of the technique to respondents and should involve
cognitive testing to ensure that questions are worded and the cards
are designed for optimal communication with respondents.  There
should also be a validity test to estimate whether, or to what
extent, respondents tell the truth when answering. 


--------------------
\6 A follow-back survey would interview foreign-born respondents
identified in a larger federal survey, such as the CPS or the
proposed American Community Survey.  The follow-back interviews would
be conducted within a few months of the main survey interview. 


      ACHIEVING MORE CERTAIN
      ESTIMATES THROUGH EVALUATIVE
      ANALYSES
-------------------------------------------------------- Chapter 5:2.4

Evaluations of the representation of the foreign-born population and
key groups within that population, such as illegals, would allow more
confident (and through adjustments, potentially better) estimates of
the size of the foreign-born population, net change in size, and
emigration.  There are a range of possible methods for evaluating
data and trends.  Some methods might be costly, while others would be
relatively low cost. 

  -- High-cost approaches would involve new data collection.  For
     example, a survey of self-identified foreign-born workers might
     be based on a sampling frame that is independent of the typical
     census or household survey approach.  That is, it might be
     possible to draw a sample of work places and conduct or schedule
     interviews with foreign-born employees as they enter or exit
     those work places.  Data on foreign-born persons interviewed in
     this way could then be compared to census results. 

  -- A more moderate-cost approach might involve adding a nativity
     question to the census short form--but only for households in
     areas where the Post Enumeration Survey will be conducted.  This
     would allow separate estimates of undercoverage for foreign-born
     and native-born persons within sex, age, and race (and Hispanic
     origin) groups.  Although no new interviews would be required,
     some cost is associated with adding questions. 

  -- Less costly methods involve no expansion of data collection. 
     For example, a pilot survey of new green-card holders, sponsored
     by INS and the National Institutes of Health, draws its sample
     from recent INS administrative records rather than mail lists or
     neighborhood enumerations.  The pilot survey takes extensive
     immigration histories, so if the full survey is implemented on
     an on-going basis, records for respondents in the survey could
     be matched to results of the decennial census--with a nonmatch
     indicating a likely lack of coverage.  A similar matching
     approach might be used for persons new to other legal statuses
     (although we recognize that less information would be available
     on them).  Turning to coverage of illegals, it might be feasible
     to access records of interior apprehensions just after the
     decennial census.  Again, records could be matched to check
     coverage. 

  -- Another example of a relatively low-cost approach would involve
     checking survey claims of U.S.  birth against state birth
     records.  The checking might be limited to a few key states,
     such as California, New York, Florida, and Texas.  It might be
     concentrated among (or begin with) respondents deemed to be at
     high risk of providing a false answer--perhaps because they live
     in areas with a high concentration of foreign-born or even areas
     where a high concentration of illegal immigrants seems likely,
     based, for example, on data on all respondents in the area
     regarding country of origin, recency of arrival in the United
     States, and occupation. 

  -- Other low-cost analyses are also possible.  For example, the
     possibility of underrepresentation of foreign-born respondents
     because of high nonresponse rates would be ruled out if response
     rates separately calculated for areas with high concentrations
     of foreign-born were similar to those for other areas. 
     Alternatively, if response rates are shown to be different,
     compensatory adjustments might be used.  Other data analyses
     could help investigate apparent trends (year-to-year change). 
     These would include technical analyses such as, for example,
     checking whether increases in the number who reported recently
     arriving correspond to trends showing increases in the estimated
     number of foreign-born residents--and vice versa. 

The high- and moderate-cost methods have the advantage of more
generalizable results; for example, they are not limited to persons
who recently received green cards.  Some of the less costly methods
require cooperation and data-sharing between, for example, INS and
the Bureau of the Census; but besides lower cost, these have the
advantage of allowing checks on valid reporting of nativity as well
as checks for coverage.  Equally important, the data-sharing methods
can provide information on specific legal-status groups. 

While we have outlined some of the possible approaches, we realize
that other strategies might also be possible. 


   CONCLUSIONS
---------------------------------------------------------- Chapter 5:3

For informed decisions on immigration issues, policymakers need
information on immigration flow, by legal status.  Separate
information is needed on the two different types of flow--new entries
into the United States and transitions to new legal
statuses--because, for example, Congress sets levels of funding for
programs to deter illegal immigrants from coming into the United
States and also defines conditions for allowing illegals to
transition to legal status. 

Policymakers also need information on the size of the foreign-born
population--again by legal status.  And information on emigration
helps to gauge the meaning of statistics on immigration flow and on
population sizes; that is, it balances information on entries with
information on exits, and it indicates the amount of turnover in the
resident population. 

INS records that are maintained for administrative purposes describe
the number of new legal permanent residents (green-card holders), new
refugees and asylees, and new naturalized citizens.  As reported in
the Yearbook, however, these statistics are limited by (1) conceptual
problems and confused reporting, (2) undercounts, and (3) information
gaps. 

  -- Annual trends in the number of green cards issued--a potentially
     key trend in legal immigration flow--is difficult to interpret
     because of conceptual problems, and the way it is reported in
     the Yearbook can confuse readers.  Similarly, invalid
     comparisons of legal flow to data on illegals may occur--again
     because of conceptual problems exacerbated by confused
     reporting. 

  -- The number of new asylees is an undercount, because the Yearbook
     tally omits certain categories of persons, such as those who are
     granted asylum on appeal.  The number of persons who newly
     attained citizenship is also an undercount. 

  -- Federal statistics are not available for some categories of
     immigration flow, such as the number of long-term temporary
     residents who come to the United States each year.  Perhaps most
     importantly, there is no estimate of the total number of
     foreign-born persons who take up residence in the United States
     each year. 

Turning to relevant demographic concepts other than flow, statistics
are reported in a more scattered fashion; indeed, a variety of INS
and Bureau of the Census publications, including the INS Web page,
must be accessed. 

The Bureau of the Census provides information on the size of the
resident foreign-born population, annual net change in size, and
emigration.  However, decennial census and survey data on the
foreign-born have not been evaluated with respect to coverage,
misreporting of nativity, and nonresponse.  Moreover, there are no
separate data for legal permanent residents, illegal immigrants, or
most other statuses.  (Neither the decennial census nor surveys that
target the general population ask questions about foreign-born
respondents' legal status.  This is so, in part, because such
questions are very sensitive and might result in problems, such as
distorted answers to the legal-status question or to other items on
the questionnaire.)

The inability to differentiate between key subgroups of the
foreign-born population is important from a policy perspective
because virtually all laws on immigration are based on specific legal
statuses.  INS has made efforts to fill gaps for some legal statuses
by using the limited data that are available and creating
assumption-based models.  The resulting estimates are necessarily
uncertain because assumptions and judgments are substituted for data. 

We identified or developed strategies that might improve immigration
statistics.  Specifically, we devised a new method for collecting
survey data on the legal status of foreign-born respondents.  The
"three-card method" asks questions that are less sensitive than a
direct question requiring the respondent to state his or her specific
legal status.  It ensures absolute privacy of response and requires
no unusual interview procedures.  Yet this method allows
statistically unbiased survey estimates for all major legal statuses. 
A preliminary qualitative test of the new method indicated that no
one refused to answer the questions.  The test population consisted
of farmworkers, and although the test was not designed to make
statistical estimates, farmworkers' answers were consistent with an
interview population that contains a high proportion of illegal
immigrants.  Thus, the new method appears to show promise and to
merit further testing and development. 

We also identified strategies for evaluating survey data on the
foreign- born.  For example, if a household does not respond to a
survey, it is not known whether the residents are foreign-born. 
Nevertheless, levels of nonresponse can be compared across
communities or areas that are known to differ in terms of nativity
(based on decennial census data or on the nativity of those who did
participate in the survey). 


   RECOMMENDATIONS
---------------------------------------------------------- Chapter 5:4

To help correct undercounts, eliminate conceptual problems, and where
possible, fill gaps for information on immigration flow, we recommend
that the Commissioner of INS (1) evaluate and, where feasible,
improve data on flow and (2) utilize an effective information
typology (either the one put forward in table 2.1 or an alternative
designed by INS) to clearly distinguish different demographic
concepts and to determine which statistics can fairly be compared to
others. 

To eliminate confused reporting of data and estimates concerning
immigration flow, we recommend that the Commissioner of INS more
clearly report information about trends in legal immigration flow and
about the difference between the concepts of flow and net change in
the INS Yearbook--or develop a new reporting format that communicates
effectively to policymakers and interested members of the general
public. 

To reduce the uncertainty associated with statistical estimates of
relevant demographic concepts other than immigration flow, fill
information gaps for specific legal statuses, and address fragmented
reporting, we recommend that the Commissioner of INS and the Director
of the Bureau of the Census together

  -- devise a plan of joint research for evaluating the quality of
     census and survey data on the foreign-born;

  -- further develop, test, and evaluate the three-card method that
     we devised for surveying the foreign-born about their legal
     status; and

  -- either publish a joint report or closely coordinate reports that
     present information on population size, net change, and
     emigration. 


   AGENCY COMMENTS
---------------------------------------------------------- Chapter 5:5

INS indicated that it is currently working to improve the clarity of
statistical reporting in the Yearbook and that it finds the typology
very useful.  INS also indicated that attempts to fill certain
information gaps may be limited by inherent difficulties and cost
considerations.  In response to this concern, we added the phrase
"where feasible" to the relevant recommendation. 

With respect to our recommendation concerning the three-card method,
INS made two comments: 

  -- First, INS suggested that because it is not an expert in survey
     methodology, its appropriate role would be limited to providing
     support and consultation to Census in that agency's efforts to
     develop, evaluate, and test the new method.  We believe that, as
     stated above, the recommendation for joint INS-Census work
     allows latitude for INS and Census to determine their
     appropriate roles. 

  -- Second, INS indicated that it would need an independent
     evaluation of the three-card method before committing funds to
     the method's development.  We agree that INS' obtaining an
     independent evaluation of the method before proceeding with
     further development would be prudent. 

The Bureau of the Census provided written comments, which raised no
objections to our findings on data gaps and the quality of federal
statistics on immigration.  The Census Bureau also did not object to
our recommendation that Census improve reporting and further evaluate
existing data on the foreign-born. 

However, the Census Bureau stated a concern about its involvement in
a survey designed to obtain information on the legal status of the
foreign-born.  Specifically, Census is concerned that even with the
privacy protections of the three-card method, such data collection
might compromise the trust and cooperation of the public.  Our
recommendation is only that the Census Bureau be involved in the
development, testing, and evaluation of the new method--not
necessarily in any resulting survey.  We believe that Census would
bring essential expertise to designing and overseeing this work. 
Testing--even large-scale testing--need not involve data collection
by the Census Bureau.  We have not revised our recommendation, but in
the interest of clarity, we modified appendix III to indicate that
contractors or other federal agencies might be used for actual data
collection involving the three-card method.\7


--------------------
\7 The full text of the Census Bureau's written response is
reproduced in app.  IV. 


EXPERTS ON IMMIGRATION WHOM WE
CONSULTED
=========================================================== Appendix I


   EXPERTS ON IMMIGRATION
--------------------------------------------------------- Appendix I:1

Charles B.  Keely, Center for Immigration Policy and Refugee
Assistance and Chair, Department of Demography, Georgetown
University, Washington, D.C. 

Daniel B.  Levine, Westat, Inc., Rockville, Md. 

Demetrios G.  Papademetriou, Senior Associate and Director,
Immigration Policy Program, Carnegie Endowment for International
Peace, Washington, D.C. 

Jeffrey S.  Passel, The Urban Institute, Washington, D.C. 

Michael S.  Teitelbaum, Alfred P.  Sloan Foundation, New York, N.Y.\1


--------------------
\1 Mr.  Teitelbaum was Vice Chair of the President's Commission on
Immigration Reform. 


DEMOGRAPHIC CHARACTERISTICS
========================================================== Appendix II

This appendix summarizes the available published data on demographic
characteristics of foreign-born persons with respect to current
(1996) flow and "stock"--that is, the size of the total foreign-born
population.  Major demographic characteristics include country of
origin or birth, age or age group, sex, race and Hispanic origin, and
marital status.  Other variables of interest include area or state of
U.S.  residence and economic characteristics such as occupation,
employment status, homeownership, poverty status, benefit receipt,
and so forth.  Not all demographic and other characteristics
identified above were available for all legal statuses of
foreign-born persons. 


   FLOW OF FOREIGN-BORN POPULATION
-------------------------------------------------------- Appendix II:1

This section discusses the available demographic information for
categories of flow for which we were able to identify relevant
published data--legal permanent residents (LPRs), refugees and
asylees, and naturalized citizens (see ch.  3, table 3.1).\1


--------------------
\1 As explained in chapter 3, relevant flow data are not available
for (1) foreign students, persons with temporary work visas, and
others admitted temporarily who stay for longer than a year, and (2)
illegal residents, i.e., EWIs and overstays who stay longer than a
year.  We also were unable to identify relevant flow data for the
total number of foreign-born residents into the United States. 
Therefore, demographic characteristics are not discussed for these
groups. 


      LEGAL PERMANENT RESIDENTS
------------------------------------------------------ Appendix II:1.1

The 1996 INS Statistical Yearbook reports only one demographic
characteristic separately for new entries, transitions
(adjustments),\2 and combined-flow (total) legal permanent residents: 
the country of birth.  Other demographic characteristics--age group,
sex, marital status, and state of intended residence--are reported
only for combined-flow LPRs, which totaled 915,900 in fiscal year
1996.  For one group of new LPRs--employment-based principal
immigrants--the 1996 INS Yearbook also reports occupation.\3
Occupational detail is shown for just 261,000 of the remaining
864,000 LPRs; over 551,000 of these LPRs are students, children,
homemakers, retirees, or unemployed workers. 

More detailed tabulations for the 915,900 LPRs can be achieved by
analyzing public use files.\4 Education and income are not included
in these administrative records.  When fully implemented, the new
green-card survey (now in pilot stage) should provide a wealth of
demographic information about new LPRs (Jasso et al., 1997).  Such
information will, no doubt, prove useful in policy-related analyses,
such as assessing potential labor market impacts. 


--------------------
\2 As noted elsewhere in this report, aliens who transition to LPR
status have often already been living in the United States for years;
therefore, the demographic characteristics enumerated should be
understood in that context, rather than interpreted as uniformly
applying to the group of aliens newly arriving in the United States
in 1996. 

\3 Employment-based principal immigrants are persons who qualify for
green-card status based on their labor market skills.  Employers
wishing to sponsor an alien for permanent residence in order to fill
a job must first apply for labor certification from the Department of
Labor.  Labor certification is awarded when there are insufficient
numbers of U.S.  workers available to undertake the employment sought
by an applicant and when the alien's employment will not have an
adverse effect on the wages and working conditions of U.S.  workers
similarly employed. 

\4 The public use files contain information on aliens granted legal
permanent resident status.  They are available on magnetic tapes or
cartridges from the National Technical Information Service.  These
files contain information on demographic characteristics, which may
be broken out separately for new entries and transitions by the
computer-oriented analyst.  More information about the public use
files, which are currently available for fiscal years 1972-96, is
listed on p.  201 of the 1996 INS Yearbook. 


      REFUGEES AND ASYLEES
------------------------------------------------------ Appendix II:1.2

The 1996 INS Yearbook reports the country of birth for 93,347 newly
admitted refugees and approved asylees in fiscal year 1996.  No other
demographic data on refugees or asylees are reported in the 1996 INS
Yearbook, other than for those who became LPRs that year (there is no
requirement for a refugee or asylee to become an LPR).  The
Department of Health and Human Services annually reports country of
citizenship and state of initial resettlement (in the United States)
for newly admitted refugees.\5


--------------------
\5 The fiscal year 1996 data are reported in "Table 4:  Amerasian,
Entrant, and Refugee Arrivals by Country of Citizenship and State of
Initial Resettlement" (U.S.  Department of Health and Human Services,
1998, p.A-8).  The report also contains certain overall 1996
demographic characteristics (such as employment rate, labor force
participation rate, and unemployment rate, by sex, and English
proficiency, hourly wages, and homeownership at time of arrival). 
However, these data refer to refugees age 16 and older in a
"five-year sample population consisting of Amerasians, Entrants, and
Refugees of all nationalities who arrived in the years 1991-1996."
Thus, demographic characteristics for 1996 refugees are not reported
separately, but the capacity to report them separately may exist,
depending upon the characteristics of the sampling frame and whether
the sample of refugees would be sufficiently large to obtain reliable
statistical results. 


      NATURALIZED CITIZENS
------------------------------------------------------ Appendix II:1.3

The 1996 INS Yearbook reports the sex, age group, marital status,
state of residence, major occupational group, region, and country of
former allegiance for 1,044,689 aliens who became naturalized
citizens in fiscal year 1996.  As noted in chapter 2, these persons
were already living in the United States. 


   FOREIGN-BORN POPULATION
   ("STOCK")
-------------------------------------------------------- Appendix II:2

This section discusses the available demographic characteristics for
categories of the resident foreign-born population for which we were
able to identify relevant published data--legal permanent residents,
illegal residents, naturalized citizens, and the total foreign-born
(see ch.  4, table 4.1).\6


--------------------
\6 As explained in ch.  4, we were unable to identify published
statistics on the size of (1) the current population of refugees and
asylees or (2) the current population of foreign students, legal
temporary workers, and others legally here for a temporary period who
have stayed longer than a year.  We do not discuss demographic
characteristics for these groups. 


      LEGAL PERMANENT RESIDENTS
------------------------------------------------------ Appendix II:2.1

On its Internet site, the INS reports the estimated state of
residence for its estimated 10,525,000 legal permanent residents
(plus or minus 350,000, and ranges for each state) as of April
1996.\7 No other demographic data are reported.  The state-level
distributions of LPRs are based on separate calculations for each
state as follows:  (1) adding post-1990 LPRs to the 1990 census count
of noncitizens, (2) subtracting 1990 estimates of illegal aliens and
nonimmigrants, (3) subtracting estimates of post-1990 illegal aliens
and nonimmigrants, and (4) adjusting for emigration and mortality. 
The ranges are based on adjustments to the 1990 Census.  The ranges
in the estimates, INS stated, resulted from adjustments made to the
number of noncitizens counted in the 1990 census.\8


--------------------
\7 The Internet address is .  Some
details of the estimation procedures are not stated there, however. 

\8 Specifically, "the census undercount of non-citizens was assumed
to range between 5 and 7 percent, while the percentage of aliens who
reported that they were citizens but who were actually non-citizens
was assumed to range between 1 and 5 percent."


      ILLEGAL ALIENS
------------------------------------------------------ Appendix II:2.2

The 1996 INS Yearbook reports two demographic characteristics--the
top 20 countries of origin and the top 20 U.S.  states of
residence--for 5 million illegal aliens as of October 1996.\9 No
other demographic data are reported.  A separate publication (Warren,
1997) provides estimates for all states (with upper and lower ranges)
and point estimates for 97 countries. 

The 1996 state-level distributions of illegals in the 1996 INS
Yearbook were constructed from separate estimates of the distribution
of EWIs and overstays.\10 For EWIs who entered the United States from
1988 to 1996, the totals were distributed to states "using INS
statistics for the early 1990s on the destination of the
beneficiaries of aliens who legalized" under the Immigration Reform
and Control Act (IRCA) of 1986.  Estimates of overstays who arrived
during 1988-96 were distributed to state of residence "based on
annual estimates of overstays by state of destination for 1986 to
1989" (emphasis added).  The state-level data for 1996, therefore,
reflect the distributions of illegal immigration for earlier years or
legal (rather than illegal) immigrants and thus may not represent the
actual current distributions. 


--------------------
\9 The countries are Mexico, El Salvador, Guatemala, Canada, Haiti,
Philippines, Honduras, Dominican Republic, Nicaragua, Poland,
Bahamas, Colombia, Ecuador, Trinidad and Tobago, Jamaica, Pakistan,
India, Ireland, Korea, and Peru.  The states are California, Texas,
New York, Florida, Illinois, New Jersey, Arizona, Massachusetts,
Virginia, Washington, Colorado, Maryland, Michigan, Pennsylvania, New
Mexico, Oregon, Georgia, District of Columbia, Connecticut, and
Nevada.  The information in the 1996 INS Yearbook is also listed on
the INS Internet site at . 

\10 As explained in ch.  4, these estimates were made by combining
estimates of illegal aliens from two major groups.  The first group
is aliens who entered surreptitiously across land borders, usually
between official ports of entry, who often are referred to as EWIs
(entries without inspection).  The second group is "nonimmigrant
overstays"--aliens who were legally admitted to the United States
temporarily and stayed beyond the specified period of admission.. 


      NATURALIZED CITIZENS
------------------------------------------------------ Appendix II:2.3

In the Bureau of the Census Current Population Reports, the
characteristics reported for naturalized citizens include age group,
sex, race, year of entry to the United States, educational
attainment, labor force status, income, receipt of means-tested cash
benefits (Aid to Families With Dependent Children, Supplemental
Security Income, and general welfare), poverty status, and
homeownership. 


      TOTAL FOREIGN-BORN
------------------------------------------------------ Appendix II:2.4

Also in its publication Current Population Reports, Census reports
age group, sex, race, citizenship (naturalized citizen or not a
citizen), period of entry to the United States, educational
attainment, labor force status, income, receipt of means-tested cash
benefits (Aid to Families With Dependent Children, Supplemental
Security Income, and general welfare), poverty status, and
homeownership for 24,557,000 foreign-born persons in 1996.\11 It
reports, for selected states, the percentage of state populations
that were foreign-born.\12 It also reports the number and percentage
of foreign-born from selected regions of the world and countries of
birth.\13


--------------------
\11 The Bureau of the Census has placed numerous publications,
including those pertaining to the foreign-born population, on its
Internet site at . 

\12 The states are those with the highest level of concentration: 
California, New York, Hawaii, Florida, New Jersey, Nevada, Texas,
Arizona, and Rhode Island. 

\13 The regions are Central America, Caribbean, South America,
Europe, and Asia.  The countries are Mexico, Canada, El Salvador,
Cuba, Dominican Republic, Jamaica, Germany, Great Britain,
Philippines, China, India, Vietnam, and Korea. 


THE THREE-CARD METHOD
========================================================= Appendix III

This appendix presents the logic of the three-card method and its
roots in the literature; a summary of the development and initial
field test of the method; the statistical expression of the
three-card estimator, together with its variance and a discussion of
the "technique effect" (i.e., variance costs); and finally, a
discussion of how the method might be applied. 


   LOGIC OF THE METHOD
------------------------------------------------------- Appendix III:1

The three-card method involves two processes:  First, separate
estimates of each legal status--that is, each status except that of
illegal immigrant--are obtained through subsampling (i.e., by drawing
three random nonoverlapping samples of foreign-born respondents) and
asking each subsample a different legal-status question.  Second, an
estimate of the percentage who are here illegally is derived through
subtraction.  Specifically,

  -- As explained in chapter 5, the first subsample of respondents is
     shown the 3-box card in figure 5.1, which features the status
     "legal permanent resident" in Box A.  The percent of subsample 1
     who choose Box A represents an estimate of the percentage of the
     foreign-born population who are legal permanent residents. 

  -- The second subsample (completely different persons) is shown the
     3-box card in figure 5.2, which features the status "U.S. 
     citizen" in Box A.  The percent of subsample 2 who choose Box A
     represents an estimate of the percentage of the foreign-born
     population who are U.S.  citizens. 

  -- The third subsample (again, completely different persons) is
     shown the 3-box card in figure 5.3, which features two
     statuses--temporary visas and refugees or asylees (who have not
     obtained a green card) in Box A.  The percentage of subsample 3
     who choose Box A represents an estimate of the percentage of the
     foreign-born who are in these statuses. 

  -- Box C on each card allows an estimate of the percentage who fall
     into categories not covered in Boxes A or B.\1

No respondent is ever asked about illegal status--nor is any specific
respondent ever identified as illegal.  Each respondent is shown one
(and only one) immigration status card, so no one could ever deduce
whether a particular respondent is here illegally.  However, an
estimate of the percentage of foreign-born who are here illegally can
be obtained by subtraction, that is, by subtracting from 100 percent
each of the estimates of legal statuses detailed above. 

While the three-card method is unique, previously reported survey
techniques also involved subsampling.  These techniques include
"randomized response" (Warner, 1965; Greenberg et al., 1969; Locander
et al., 1976), "aggregated response" (Boruch and Cecil, 1979), and
"item count" (Droitcour et al., 1991).  The stratagem used in the
three-card method of obtaining an estimate of illegals by subtracting
estimated legals from total foreign-born is a "residual method" that
derives from demographic techniques (Schryock, Siegel, and
Associates, 1980; Warren and Passel, 1987). 

Depending upon the sample design, the three-card method can be used
to obtain a separate estimate of illegals for a sizable geographic
area (such as California), simply by separately analyzing the data
for the area of interest.  This would require the sample design to
include enough respondents from a particular subgroup or geographic
area.  In this way, estimates could potentially be obtained for
several key states--for example, California, New York, and Texas--and
the main characteristics of the illegal population could be
described.  Similarly, a separate estimate of illegals can be
obtained for a particular demographic subgroup (such as women of
child-bearing age or persons below the poverty line). 

A unique element of the three-card method consists of a set of
follow-up questions addressed to respondents who choose Box A.  These
questions are essential because they can help confirm the validity of
answers for respondents choosing Box A or, alternatively, provide the
information needed to reclassify certain respondents into Box B or
C.\2


--------------------
\1 We expect that most respondents choosing Box C would be recoded as
Box A or B, depending on their explanation of what their status is. 
A very small percentage of respondents should fall into immigration
statuses not covered by the categories shown on the cards. 

\2 For example, some subsample 1 respondents who initially choose Box
A in fig.  5.2 (U.S.  citizenship) may actually have only applied for
U.S.  citizenship.  By a line of questioning that asks when the
respondent applied for citizenship and whether he or she has as yet
actually received a notification and attended a swearing-in ceremony,
those who have only applied for citizenship should be identified. 
Such respondents would then be recoded into Box B or C as
appropriate. 


   DEVELOPMENT AND INITIAL FIELD
   TEST
------------------------------------------------------- Appendix III:2

We developed a bilingual (English-Spanish) questionnaire through a
series of pretests and then contracted for an initial, qualitative
field test of the technique.  Only one immigration-status card was
tested:  the card with "legal permanent resident" in Box A (see fig. 
5.1).  The field test was intended to gauge the acceptability of the
3-box card and the question series to both respondents and
interviewers. 

The pretests to develop the three-card survey method were conducted
by four GAO staff who are fluent in Spanish.\3 The pretest
respondents consisted of Hispanic immigrants.  GAO staff introduced
themselves to prospective respondents and gained their cooperation at
four locations--a "drop-in" center in Los Angeles; a legal clinic in
Arlington, Virginia; and farms and farm-related facilities in
Colorado and Pennsylvania.\4 In all, 27 pretest interviews were
conducted by GAO staff. 

Throughout the pretests and questionnaire revisions, our aims were to
emphasize respect for the respondent; make sure the respondent
understood the questions; and minimize "question threat." For
example, to reassure the respondents, just before being shown the
immigration status card, they were told that the topic of the card
would be immigration status.  They were also told that the categories
were arranged in such a way that the questions would not zero in on
anything they might not want to tell us. 

For respondents who picked Box A (legal permanent resident),
follow-up questions were designed to help us judge the validity of
the response.  For example, respondents who picked Box A were asked
to state the specific program through which they obtained their green
card.  No follow-up questions were asked of anyone who picked Box B. 

We subjected the three-card survey method to a preliminary test of
respondent (and interviewer) acceptance.  Specifically, we contracted
with Aguirre International, the survey organization that conducts the
Department of Labor's (DOL) National Agricultural Workers Survey
(NAWS), to administer our survey in interviews that would be added on
to the regular NAWS data collection effort.\5 A total of 81
interviews were completed by four interviewers at multiple sites in
six states (Arizona, California, Florida, Georgia, Kentucky, and
Texas) between November 1997 and February 1998.  Consistent with NAWS
procedures, farmworker respondents were paid $10 cash as an incentive
to participate. 

Interviewers reported encountering varying degrees of
apprehensiveness on the part of respondents.  For example, in
Florida, the interviewer's impression was that all respondents were
illegal, fearful of the Immigration and Naturalization Service (INS),
and nervous about the survey.  At some locations, there were
instances where respondents voiced hostility, seemed to identify the
survey with INS, or initially mistook the interviewer for a tax
collector.  At other locations, respondents appeared to be relaxed,
and some even volunteered to the interviewer that they were here
illegally. 

All 81 respondents answered the immigration questions.  Interviewers
in Arizona, California, and Kentucky specifically mentioned, in
debriefings, that respondents did not find the 3-box question on
legal status too intrusive. 

Of the 81 respondents, 30 selected Box A, that is, claimed to have a
valid green card.  Each respondent who selected Box A was able to
identify a specific category or program under which he or she
obtained a green card.  Most of these (23 out of 30) said they had
obtained their green cards through the amnesty (Immigration Reform
and Control Act/Seasonal Agricultural Workers) program or through the
Family Unity program. 

Fifty-one respondents selected Box B, which contained the
undocumented worker category and three other legal immigration
categories.  Most of those who selected Box B were probably working
without legal authorization, based on NAWS and our prior work; that
is, the population of foreign-born farmworkers is thought to contain
few naturalized U.S.  citizens and very few workers here with
temporary work visas.\6

The interviewer working in Florida said that some respondents
hesitated to select a box.  However, the data indicate that all
Florida respondents did make a selection--and all picked Box B. 

No one selected Box C--"Other categories not included in Box A or B."
No one was coded as "unsure" about which box applied to him or her. 

Overall, the general approach--including the 3-box card, the
explanation leading up to the card, and the specific questions
asked--appeared to be acceptable to respondents.  However, three
issues emerged, as follows: 

  -- Although most respondents did not appear to be confused about
     which box to choose, a few respondents who told interviewers
     that they had "fake" green cards or had "border crossing cards"
     were unsure which box to pick.  There was also one respondent in
     Arizona who was initially not sure which of two legal categories
     applied to him.  Interviewers had been trained to help
     respondents, when necessary, by restating (paraphrasing)
     questions and providing explanations.  In the future, to help
     protect the privacy of illegal immigrants (by avoiding the need
     for them to ask a question that reveals their status), more
     specific instructions might be voiced to all respondents at the
     outset.  For example, "Persons with a 'fake' green card--not a
     valid one--belong in Box B."

  -- Although some respondents in Florida said they found the icons
     (pictures) on the cards useful, others voiced criticisms. 
     Interviewers reported that some respondents thought the icons
     were childish and unnecessary, and even that the card was
     unnecessary because they had understood the verbal instructions
     and felt that they did not need help.  The California
     interviewer indicated that respondents with less education
     tended to like the card more than those with more
     education--perhaps reflecting reactions to the icons.\7

  -- There were also criticisms of some specific icons:  The icon for
     refugees shows a person running toward the United States, and
     some respondents interpreted the icon as being an undocumented
     worker running away from U.S.  law enforcement officers.  The
     icon for student, worker, or tourist visa is a suitcase with
     various stickers, and some felt that a suitcase was not clearly
     linked to an immigration status. 

The reactions to the icons suggest there is a need for focus groups
or other qualitative work aimed at evaluating and possibly revising
the icons and, if possible, finding a way to present them that is
more acceptable to those respondents who do not need them. 

More generally, as stated in chapter 5, we believe that extensive
testing--including testing to explore the acceptability of the
technique to respondents and to estimate the validity of
responses--should be conducted before the technique is used in a
large-scale survey. 


--------------------
\3 Three are bilingual Hispanics; the fourth had lived and worked in
Hispanic countries for a total of 5 years. 

\4 We identified these locations through the help of charitable
organizations and health clinics, and by directly contacting farms in
the vicinity of our Denver Field Office. 

\5 The contractor did not provide the identities of farm employers to
GAO. 

\6 Only about 3 percent of farmworkers claim to be naturalized
citizens in National Agricultural Workers Surveys (Mines et al.,
1997).  According to a recent GAO report, in fiscal year 1996, only
about 15,000 farmworkers entered the country through the H-2A
nonimmigrant guestworker program (i.e., temporary agricultural work
visas); this "represents less than 1 percent of the agricultural
field workforce" (GAO/HEHS-98-20, p.  18). 

\7 In addition, one might reason that not all respondents would
appreciate the 3-box card.  Specifically, those respondents who do
have their green cards would not need the 3-box card; i.e.,
logically, they might just as soon answer a direct question.  By
contrast, illegal immigrants might appreciate the card.  (The Florida
interviewer, who said that some respondents found the 3-box card
useful, thought that all respondents there were working illegally;
all chose Box B.)


   STATISTICAL EXPRESSION OF THE
   ESTIMATOR AND VARIANCE
------------------------------------------------------- Appendix III:3

The statistical expression of the three-card estimator of the percent
illegal, its variance, and the "technique effect" are shown in figure
III.1.  Clearly, there are variance costs associated with using an
indirect method.  First, each estimate of legal status is based on a
subsample, rather than the full sample.  Second, the indirect
estimate of illegals is affected by the variance of each of the
estimates that is included in its calculation.  Assuming that the
three subsamples are of equal size and that for each card 25 percent
of respondents belong in Box A, the confidence interval for a
three-card method estimate of illegals would be three times as large
as a corresponding direct estimate.  However, it must be kept in mind
that variance costs depend heavily upon (1) the distribution of
immigration status in the population surveyed and (2) the relative
sizes of the three subsamples.  Two examples are presented below to
give the reader a flavor of what "real world" precision might be: 

  -- Example 1:  In a population of foreign-born agricultural
     workers, the distribution of immigration status might be 55
     percent illegal; 36 percent legal permanent residents; 3 percent
     U.S.  citizens; and 6 percent temporary workers, refugees, or
     asylees.  Assuming this distribution and a total sample size of
     only 1,000, allocated with 100 respondents to answer the card
     with U.S.  citizen in Box A; 200 to answer the card with
     temporary visas, refugees, or asylees in Box A; and 700 to
     answer the card with legal permanent resident in Box A, the
     95-percent confidence interval for an estimate of 55 percent
     illegal would be 49-61 percent. 

  -- Example 2:  In the residential foreign-born population of the
     United States, taken as a whole, the distribution might be 22
     percent illegal; 30 percent U.S.  citizens; 38 percent legal
     permanent residents; and 10 percent temporary workers, refugees,
     or asylees (without green cards).  Assuming this distribution
     and a total sample size of 13,000--the approximate number of
     foreign-born in the March Current Population Survey (CPS)
     supplement--with 6,000 respondents allocated to answer the card
     with legal permanent resident in Box A; 5,500 allocated to
     answer the card with U.S.  citizen in Box A; and 1,500 to answer
     the card with temporary workers, refugees, or asylees in Box A,
     the 95-percent confidence interval would be 20-24 percent. 

   Figure III.1:  Statistical
   Expression of the Three-Card
   Estimator, Its Variance, and
   Technique Effect

   (See figure in printed
   edition.)

We note that in the foregoing examples, the sizes of the three
samples were chosen according to the general statistical principle of
"optimal allocation."\8 The confidence intervals in the examples
would have been larger if three equal-sized samples had been used. 


--------------------
\8 Optimal allocation minimizes the variance of an estimate by
assigning a greater proportion of the sample where the variance is
highest--and a smaller proportion where the variance is lowest. 


   POSSIBILITIES FOR
   IMPLEMENTATION
------------------------------------------------------- Appendix III:4

This section considers possibilities for implementing the three card
method--assuming that further research proves its viability.  There
are two key points:  First, the three-card method is designed for use
in a face-to-face survey in which the questionnaire is administered
by a professional interviewer.  It cannot be used in a telephone
survey, because respondents could not see the cards.\9 Second,
because the relevant population consists of the foreign-born, an
appropriate survey to utilize the method would be either (1) a survey
that targets populations with high concentrations of foreign-born
(e.g., NAWS or a survey of a community with high concentrations of
foreign-born) or (2) an ongoing survey conducted for other purposes
that is large enough to include a sizable foreign-born group. 

The CPS is one example of a relatively large survey.  The proposed
American Community Survey is much larger.  Neither survey conducts
all interviews in person.  (The CPS, e.g., repeatedly interviews the
same households, but of eight total interviews, only the first and
the fifth are conducted in person.) A special plan would be required
if the three-card method were to be used with surveys that do not
conduct all interviews in person.  Various other federal surveys
conducted by agencies other than Census--or surveys conducted by
contractors--might also be considered. 

Another issue is that--at least initially--the sponsors of ongoing
surveys would probably deem the three-card method too sensitive to
include.  Therefore, one possibility that was suggested in
conversations with staff and officials at the Bureau of the Census is
a "follow-back" survey.  For example, a large ongoing survey might be
used to identify a sample of foreign-born respondents who would be
"followed back" within, for example, a few months of the survey
interview; a brief reinterview would then be conducted, including the
three-card method.\10

It would be more costly to conduct house-to-house canvasing to
identify foreign-born respondents, but the specific costs would
depend on the extent to which canvasing efforts could be targeted,
based on existing census and survey data on the locations where
foreign-born persons are concentrated. 


--------------------
\9 It is not known at this time whether the three-card method could
be modified for use in a mail survey. 

\10 If such a plan were acceptable to an ongoing survey, the cost
might be roughly $300 per completed follow-back interview.  This
would put the cost of 12,000 interviews at roughly $3.5 million.  Of
course, if the three-card method were eventually accepted as an
add-on to the questions asked in an ongoing survey, the cost per
interview would be a small fraction of $300. 


   AGENCY COMMENTS
------------------------------------------------------- Appendix III:5

The Census Bureau indicated a concern that appendix III gives an
"overly optimistic evaluation" of the three-card method.  To ensure
clarity, we added a sentence to appendix III, which summarizes the
caveat stated in chapter 5--that is, that extensive testing is needed
before including the new method in a large-scale survey. 




(See figure in printed edition.)Appendix IV
COMMENTS FROM THE BUREAU OF THE
CENSUS
========================================================= Appendix III



(See figure in printed edition.)


MAJOR CONTRIBUTORS TO THIS REPORT
=========================================================== Appendix V

GENERAL GOVERNMENT DIVISION,
WASHINGTON, D.C. 

Eric M.  Larson, Evaluator-in-Charge
Barry Reed, Technical Adviser
Elizabeth W.  Scullin, Communications Analyst

DENVER FIELD OFFICE

Maria Vargas, Evaluator

OFFICE OF THE GENERAL COUNSEL,
WASHINGTON, D.C. 

Ann H.  Finley, Senior Attorney

OFFICE OF THE CHIEF ECONOMIST,
WASHINGTON, D.C. 

Venkareddy Chennareddy, Referencer




We would like to thank Juan Gobel, Los Angeles Field Office, and
Nilsa Perez, General Government Division, for helping with interviews
with Hispanic immigrants. 


BIBLIOGRAPHY
============================================================ Chapter 1

Ahmed, Bashir, and J.  Gregory Robinson.  "Estimates of Emigration of
the Foreign-born Population:  1980-1990" (Technical Working Paper No. 
9).  Washington, D.C.:  Bureau of the Census, 1994. 

Bogue, Donald J., et al.  "Components of Change in Size:  The
Demographic Bookkeeping Equation," Readings in Population Research
Methodology, Vol.  1:  Basic Tools.  Chicago:  U.N.  Population Fund,
Social Development Center, 1993. 

Boruch, Robert, and Joe S.  Cecil.  Assuring the Confidentiality of
Social Research Data.  Philadelphia:  University of Pennsylvania
Press, 1979. 

Brackstone, Gordon J.  "Discussion," 1988 Proceedings of the Section
on Survey Research Methods, pp.  77-78.  Paper presented at the
Annual Meeting of the American Statistical Association, New Orleans,
Aug.  1988. 

Bradburn, Norman M., and Seymour Sudman.  Improving Interview Method
and Questionnaire Design:  Response Effects to Threatening Questions
in Survey Research.  San Francisco:  Jossey-Bass, 1979. 

Chapa, Jorge, and Jorge del Pinal.  "Enumeration, Housing
Characteristics and Sampling Rates in the Colonias of the Texas
Border Area:  A Perspective on Census Data," Proceedings of the 1993
Research Conference on Undercounted Ethnic Populations, Richmond, May
5-7, 1993.  Washington, D.C.:  Department of Commerce, 1993. 

Davies, Christopher S.  "Colonia Settlements:  Working-Class Refuge
Stations along the Texas-Mexico Border." Planning Forum, Vol.  1
(1995), pp.  33-53. 

de la Puente, Manuel.  "Why Are People Missed or Erroneously Included
by the Census:  A Summary of Findings from Ethnographic Coverage
Reports," Proceedings of the 1993 Research Conference on Undercounted
Ethnic Populations, Richmond, May 5-7, 1993.  Washington, D.C.: 
Department of Commerce, 1993. 

Droitcour, Judith, et al.  "The Item Count Technique as a Method of
Indirect Questioning:  A Review of Its Development and a Case Study
Application," Measurement Errors in Surveys, pp.  185-210, Paul P. 
Biemer et al.  New York:  John Wiley & Sons, 1991. 

Edmonston, Barry (ed.).  Statistics on U.S.  Immigration:  An
Assessment of Data Needs for Future Research.  Washington, D.C.: 
National Academy Press, 1996. 

Fernandez, Edward, and J.  Gregory Robinson.  "Illustrative Ranges of
the Distribution of Undocumented Immigrants by State" (Technical
Working Paper No.  8).  Washington, D.C.:  Bureau of the Census,
Population Division, 1994. 

Gabbard, Susan, Edward Kissam, and Philip L.  Martin.  "The Impact of
Migrant Travel Patterns on the Undercount of Hispanic Farm Workers,"
Proceedings of the 1993 Research Conference on Undercounted Ethnic
Populations, Richmond, May 5-7, 1993.  Washington, D.C:  Department
of Commerce, 1993. 

Goering, John M.  "The 'Explosiveness' of Chain Migration:  Research
and Policy Issues:  Introduction and Overview," International
Migration Review, 23 (1989), pp.  797-812. 

Greenberg, Bernard G., et al.  "The Unrelated Questions Randomized
Response Model:  Theoretical Framework," Journal of the American
Statistical Association, 64 (1969), pp.  520-39. 

Groves, Robert M.  Survey Errors and Survey Costs.  New York:  Wiley,
1989. 

Hearing Before the Senate Judiciary Subcommittee on Technology,
Terrorism, and Government Information, "Foreign Terrorists in
America:  Five Years After the World Trade Center," Feb.  24, 1998. 

Hearing Before the House Judiciary Subcommittee on Immigration and
Claims, "Legal Immigration Projections." 104th Congress, 2nd Session,
Serial No.  104, May 16, 1996. 

Hoaglin, David C., et al.  Data for Decisions:  Information
Strategies for Policymakers.  Cambridge, Mass.:  Abt Books, 1982. 

Hollman, Frederick W.  "Derivation of Independent Population
Controls." Unpublished paper, Bureau of the Census, 1997. 

Hollmann, Frederick W., et al.  "U.S.  Population Estimates by Age,
Sex, Race, and Hispanic Origin:  1990 to 1997" (PPL-91R). 
Washington, D.C.:  Bureau of the Census, 1998. 

Jasso, Guillermina, and Mark R.  Rosenzweig.  "Using National
Recording Systems for the Measurement and Analysis of Immigration to
the United States," International Migration Review, 21 (1987), pp. 
1212-44. 

Jasso, Guillermina, et al.  "The New Immigrant Survey (NIS) Pilot
Study:  Preliminary Results." Revised version of paper presented at
the 1997 Joint Meeting of the Public Health Conference on Records and
Statistics and the Data Users Conference.  Washington, D.C.:  July
1997. 

Kraly, Ellen Percy, and Robert Warren.  "Estimates of Long-Term
Immigration to the United States:  Moving U.S.  Statistics toward
United Nations Concepts," Demography, 29 (Nov.  1992), pp.  613-26. 

Levine, Daniel B., Kenneth Hill, and Robert Warren (eds.). 
Immigration Statistics:  A Story of Neglect.  Washington, D.C.: 
National Academy Press, 1985. 

Locander, William., et al.  "An Investigation of Interview Method,
Threat and Response Distortion," Journal of the American Statistical
Association, 71 (1976), pp.  269-75. 

Lowell, B.  Lindsay, ed.  Temporary Migrants in the United States. 
Washington, D.C.:  U.S.  Commission on Immigration Reform, 1996. 

McKay, Ruth B.  "Undercoverage of Hispanics in household surveys,"
Monthly Labor Review, 116:9 (1993), pp.  38-42. 

McKay, Ruth B., et al.  "Translating Survey Questionnaires:  Lessons
Learned," in "Advances in Survey Research," by Marc T.  Braverman and
Jana Kay Slater (eds.).  NewDirections for Evaluation, 70 (1996), pp. 
93-104. 

Mines, Richard, et al.  A Profile of U.S.  Farmworkers: 
Demographics, Household Composition, Income and Use of Services. 
Washington, D.C.:  Department of Labor, Office of the Assistant
Secretary for Policy, 1997. 

Passel, Jeffrey S.  "Undocumented Immigration," The Annals of the
American Academy of Political and Social Science, Vol.  487 (Sept. 
1986), pp.  181-200. 

Passel, Jeffrey S.  "Immigration Statistics:  No Longer Neglected,
But Still Inadequate,"Statistical Policy Working Paper 26, Federal
Committee on Statistical Methodology, pp.  27-42.  Washington, D.C.: 
OMB, Aug.  1997. 

Passel, Jeffrey S., and Rebecca L.  Clark.  Immigrants in New York: 
Their Legal Status, Incomes, and Taxes.  Washington, D.C.:  The Urban
Institute, 1998. 

Pollard, A.H., et al.  "The Balancing Equation," pp.  10-11. 
Demographic Techniques, 2nd ed.  Sydney, Australia:  Pergamon Press,
1981. 

Schmidley, A.  Dianne, and J.  Gregory Robinson.  "How Well Does the
Current Population Survey Measure the Foreign Born Population in the
United States?" (Technical Working Paper No.  22).  Washington, D.C.: 
Bureau of the Census, 1998. 

Schryock, Henry S., Jacob S.  Siegel, and Associates.  The Methods
and Materials of Demography (4th printing, rev.).  Washington, D.C.: 
U.S.  Government Printing Office (GPO), 1980. 

U.S.  Bureau of the Census.  "The Foreign-Born Population:  1996."
Current Population Reports, P20-494, Mar.  1997. 

U.S.  Bureau of the Census.  "The Foreign-Born Population in the
United States:  March 1997." Current Population Reports, P20-507,
Mar.  1998. 

U.S.  Commission on Immigration Reform.  U.S.  Immigration Policy: 
Restoring Credibility.  1994 Report to Congress.  Washington, D.C.: 
GPO, 1994. 

U.S.  Commission on Immigration Reform and Secretaria de Relaciones
Exteriores.  Binational Study on Migration Between Mexico and the
United States.  Mexico:  Editorial y Litografa Regina de los
Angeles, S.A., 1997. 

U.S.  Department of Commerce.  Statistical Policy Handbook. 
Washington, D.C.:  GPO, 1978. 

U.S.  Department of Education.  Digest of Education Statistics, 1997
(NCES-98-015).  Washington, D.C.:  NCES, 1997. 

U.S.  Department of Health and Human Services.  Refugee Resettlement
Program:  FY 1996.  Report to Congress.  Washington, D.C.:  GPO,
1998. 

U.S.  Department of Justice.  "Immigration and Naturalization
Service:  Monitoring of Nonimmigrant Overstays." Office of the
Inspector General, Report No.  1-97-08.  Sept.  1997. 

U.S.  General Accounting Office.  Immigration:  Data Not Sufficient
for Proposed Legislation (GAO/PEMD-89-8).  Washington, D.C.:  GAO,
Dec.  1989. 

U.S.  General Accounting Office.  Immigration Statistics:  Guidance
on Producing Information on the U.S.  Resident Foreign-Born
(GAO/GGD-98-155).  Washington, D.C.:  GAO, July 1998. 

U.S.  General Accounting Office.  Immigration Statistics:  Status of
the Implementation of the National Academy of Sciences'
Recommendations (GAO/GGD-98-119).  Washington, D.C.:  GAO, June 1998. 

U.S.  General Accounting Office.  Illegal Immigration:  Southwest
Border Strategy Results Inconclusive; More Evaluation Needed
(GAO/GGD-98-21).  Washington, D.C.:  GAO, Dec.  1997. 

U.S.  General Accounting Office.  Criminal Aliens:  INS' Efforts to
Identify and Remove Imprisoned Aliens Need to Be Improved
(GAO/T-GGD-97-154).  Washington, D.C.:  GAO, July 1997. 

U.S.  General Accounting Office.  1990 Census:  Reported Net
Undercount Obscured Magnitude of Error (GAO/GGD-91-113).  Washington,
D.C.:  GAO, Aug.  1991. 

U.S.  General Accounting Office.  H-2A Agricultural Guestworker
Program:  Changes Could Improve Services to Employers and Better
Protect Workers (GAO/HEHS-98-20).  Washington, D.C.:  GAO, Dec. 
1997. 

U.S.  General Accounting Office.  Illegal Immigration:  INS Overstay
Estimation Methods Need Improvement (GAO/PEMD-95-20).  Washington,
D.C.:  GAO, Sept.  1995. 

U.S.  General Accounting Office.  Immigration and the Labor Market: 
Nonimmigrant Alien Workers in the United States (GAO/PEMD-92-17). 
Washington, D.C.:  GAO, Apr.  1992. 

U.S.  General Accounting Office.  Immigration Reform:  Major Changes
Likely Under S.  358 (GAO/PEMD-90-5).  Washington, D.C.:  GAO, Nov. 
1989. 

U.S.  General Accounting Office.  Immigration:  Projected Immigration
Under S.  448 and Recent Trends in Legal Immigration
(GAO/PEMD-89-12).  Washington, D.C.:  GAO, Apr.  1989. 

U.S.  General Accounting Office.  Immigration:  S.  355 Would Change
the Distribution of Immigrant Classes (GAO/T-PEMD-89-1).  Washington,
D.C.:  GAO, Mar.  1989. 

U.S.  General Accounting Office.  Immigration:  The Future Flow of
Legal Immigration to the United States (GAO/PEMD-88-7).  Washington,
D.C.:  GAO, Jan.  1988. 

U.S.  Immigration and Naturalization Service.  1996 Statistical
Yearbook of the Immigration and Naturalization Service.  Washington,
D.C.:  GPO, 1997. 

U.S.  Immigration and Naturalization Service.  "Duration of Stay of
Nonimmigrants Departing the United States." Washington, D.C.:  INS,
1996. 

U.S.  Immigration and Naturalization Service.  Statistical Yearbook
of the Immigration and Naturalization Service, 1995.  Washington,
D.C.:  GPO, 1997. 

Van Hook, Jennifer, and Frank D.  Bean.  "Estimating Underenumeration
Among Unauthorized Mexican Migrants to the United States: 
Applications of Mortality Analysis," Migration Between Mexico and the
United States:  Binational Study.  Vol.  2:  Research Reports and
Background Materials, pp.  551-69.  Washington, D.C.:  U.S. 
Commission on Immigration Reform, 1998. 

Velasco, Alfredo.  "Ethnographic Evaluation of Behavioral Causes: 
Undercount in the Community of Sherman Heights, California."
Ethnographic Evaluation of the 1990 Decennial Report Series.  Report
No.  22.  Washington, D.C.:  Bureau of the Census, Center for Survey
Methods Research, Oct.  1992. 

Warner, Stanley.  "Randomized Response:  A Survey Technique for
Eliminating Evasive Answer Bias," Journal of the American Statistical
Association, Vol.  60 (1965), pp.  63-69. 

Warner, Stanley.  "The Linear Randomized Response Model," Journal of
the American Statistical Association, Vol.  66 (1971), pp.  884-88. 

Warren, Robert.  "Estimates of the Undocumented Immigrant Population
Residing in the United States:  October 1996." Paper presented at the
Joint Statistical Meetings, Anaheim, Calif., Aug.  13, 1997. 

Warren, Robert, and Ellen Percy Kraly.  "The Elusive Exodus: 
Emigration From the United States," Population Trends and Public
Policy (No.  8).  Washington, D.C.:  Population Reference Bureau,
1985. 

Warren, Robert, and Jeffery S.  Passel.  "A Count of the Uncountable: 
Estimates of Undocumented Aliens Counted in the 1980 Census,"
Demography, 24 (1987), pp.  375-93. 

Woodrow, Karen A.  Preliminary Research and Evaluation Memorandum No. 
75.  Demographic Analysis Evaluation Project D2:  Preliminary
Estimates of Undocumented Residents in 1990.  Washington, D.C.: 
Bureau of the Census, Population Division, Oct.  22, 1991. 

Woodrow, Karen A., and Jeffrey S.  Passel.  "Post-IRCA Undocumented
Immigration to the United States:  An Assessment Based on the June
1988 CPS," Undocumented Migration to the United States:  IRCA and the
Experience of the 1980s.  Frank D.  Bean et al.  (eds.).  Washington,
D.C.:  The Urban Institute, 1990. 

Woodrow-Lafield, Karen A.  "An Analysis of Net Immigration in Census
Coverage Evaluation," Population Research and Policy Review 14
(1995), pp.  173-204. 

Woodrow-Lafield, Karen A., "Undocumented Residents in the United
States in 1990:  Issues of Uncertainty in Quantification,"
International Migration Review, 32 (1998), pp.  145-73. 

*** End of document. ***