Table of
Contents:
What to
consider when evaluating a potential data source on immigrants:
US Decennial Census, 1850-present
US Current Population Survey, November Voting
Supplement, 1994 to present
(Bureau of
Labor Statistics & US Census Bureau)
National Election Studies, 1948 to present
Harvard/Kaiser/Washington Post Latino Survey,
1999
(PIs: Washington
Post, Kaiser, Robert J. Blendon & John M. Benson from Harvard)
National Latino Political Survey, 1989-90
(PIs: de la Garza,
Falcon, Garcia, and Garcia)
Pilot National Asian American Political
Survey, 2000-01
California Statewide Database, 1990-Present
(PI: Bruce
Cain, Institute of Governmental Studies, UC Berkeley)
Cuban and Mexican Immigrants to the US,
1973-1979
General Social Survey, 1972-present
(PIs: Davis,
Smith, Marsden; National Opinion Research Center)
American Citizen Participation Study, 1990
(PIs Sidney
Verba, Kay Lehman Schlozman, Henry E. Brady, Norman Nie)
Social Capital Community Benchmark Survey,
2000
(PI: Robert
Putnam; Roper Center & Saguaro Center at Harvard University)
Other Opinion Polls and Surveys
Public Policy Institute of California
New Americans Exit Poll, New York City 2000
Metropolitan Chicago Information Center
The
empirical study of immigrant groups raises special challenges, touching on the
questions we ask, the groups we seek to cover and the methods employed during
data collection. Such considerations
make it difficult to use secondary data (evidence collected by another
researcher or organization) in the analysis of immigrant-related
questions. Students of immigration
should ask themselves the following when evaluating a potential data source:
http://www.ipums.org/usa/index.html (especially useful for historical
comparison)
or
http://www.census.gov/ (for 1980 census data and after, including
other Census Bureau studies)
A national population count. Immigrant variables are based on the “long form” sample. Use the census web page for descriptive tables (pre-prepared or using the American FactFinder program) from the 1990 or 2000 census [2000 immigration info will be available summer/fall 2002]. For statistical analysis, use the Public Use Microfile Samples available through the Census page (recent censuses only) or through IPUMS (includes most US censuses from 1850 to present).
Best source for repeated cross-sectional citizenship data over the last 100 years. The strong points of census data include: very good coverage of the immigrant population (despite undercounts) thus allowing detailed analysis of smaller immigrant subpopulations; generally high data quality; and the repetition of many questions, allowing comparisons across time. Variables of interest include: place of birth, year/period of entry to the US, race/ethnicity, language ability and citizenship status. Also included are a host of socio-economic variables, household variables and geographic variables allowing for the study of local place effects. The major draw back is the lack of other political variables (only citizenship) and the timeliness of data (only collected every ten years).
NOTE: The census has recently
started a new program, the American Community Survey, which provides yearly
data on most of the standard census questions, including place of birth,
citizenship, ancestry and race. Currently
this survey is only being conducted in select US communities. For more information, visit http://www.census.gov/acs/www/
http://www.bls.census.gov/cps/
(general page)
or
http://www.bls.census.gov/cps/vote/votemain.htm (voting supplement)
Probability
sample of 50,000-60,000 US households.
Primary purpose is to collect data on the US labor force, but since 1994
CPS fields a voting supplement at election time. Information on methodology, variables and downloadable data
(tables and microdata) are available through the web site.
Best
source of national immigrant voting data currently available. Starting in 1994 and taken every two years
to coincide with November elections, the CPS includes key immigration variables
(place of birth, parents’ place of birth, period of entry to the US), key
socio-economic, demographic and geographic variables (including race), and some
key measures of political incorporation (citizenship, voted in last election,
registration). The data quality is
good, though interviews are generally in English (and sometimes Spanish), with
one household member reporting for others.
The large sample size provides a reasonable number of immigrants and,
for large immigrant sub-populations, a sample big enough for specific group
analysis at the national level. For
smaller groups or specific geographic locations, the number of respondents
becomes quite small.
NOTE: Some Census Bureau
publications using the CPS voting data are available at http://www.census.gov/prod/www/abs/vote.html. More general information about the foreign-born, including
citizenship, is available from http://www.census.gov/prod/www/abs/for-born.html.
http://www.umich.edu/~nes/index.htm (data, methodology, bibliography)
National
probability samples conducted every two years (Presidential and mid-term
elections), focusing on public opinion and political participation. Presidential elections include pre- and
post-election surveys. Sample N usually
ranges from 1,200 to 2,200. Sample
population restricted to adult citizens.
Key
source of political opinion and participation information in the United
States. There are a wide range of
questions on contemporary policy issues, the current election of that year,
party identification and participation.
Most standard demographic and socio-economic status variables are
included. Usually includes variables on
race, ethnicity and place of birth.
Prior to the 1980s, usually included a question on parents’ place of
birth. No variables on length of
residence in the United States, if born outside the US.
Not
a great resource for immigration researchers due to: the small sample size (the
potential immigrant sub-sample is very small), lack of information on
period/year of immigration and inconsistent questions over the years on
variables such as ancestry and parents’ nativity. Samples can be pooled, and those interested in voting behavior in
the 1960s and ‘70s might find that this is their only solid resource. Those interested in using NES should
carefully go over the codebook for a specific election year since questions
vary year to year.
http://www.kff.org/content/2000/3023/LatinoFullToplineFinal.PDF (for summary of study and descriptive tables, including some
cross-tabulated by national origin or race)
http://data.fas.harvard.edu/guide-bin/ropunzip?R00454 (for codebook and raw data
through the Harvard-MIT Data Center, Harvard affiliates only; also available
from Roper, USICR1999-5925)
Telephone opinion poll of
4,614 adults in California, Florida, Illinois, New York, Texas and the District
of Columbia aimed at comparing Latino attitudes on values, politics, race
relations and social policies to those of non-Latino whites and blacks. Latino N=2,417 (1,443 citizens, 974
non-citizens) and non-Latinos N=2,197 (1,802 whites and 285 blacks).
A great resource for those
interested in Latino immigrants. The
large sample size provides significant samples of first and second generation
Latinos with good representation among the major national-origin groups (818
Mexicans, 318 Puerto Ricans, 312 Cubans & 593 Central or S.
Americans). “Latino” respondents were
identified by a question asking, “Are you, yourself, of Hispanic or Latin
origin or descent…?” Includes an interesting
and diverse array of questions, including on naturalization (and reasons
for/against), on voting and other political activities, on political opinions
and attitudes toward major social issues and even some transnational activities
(e.g., ‘Did you vote in a home country election since coming to the US?’). Includes most standard demographic and SES
backgrounds variables, plus religion, and critical immigration questions
(country of birth, length of residence, citizenship, etc.). Conducted in English or Spanish between June
30 and August 30, 1999.
Past the 2nd
generation, the sample becomes limited, and it is not a perfectly
representative national sample (the five states and DC area account for about
70% of the US Latino population), though results are weighted to national
proportions using CPS data. Response
rate is 53%.
http://www.icpsr.umich.edu:8080/ABSTRACTS/06841.xml?format=ICPSR
(data, methodology)
National
area probability sample of adults of Cuban, Mexican or Puerto Rican origin
(from immigrant to 3rd generation) focused on political attitudes
and behaviors. Includes a comparative
sample of non-Latinos (defined as anyone not considered Latino). N=3,415 (Mexican=1546, Cuban=682, Puerto
Rican=589, and non-Latino=598).
Methodology, codebook, questionnaires (English and Spanish) and
micro-data available through ICPSR (or Harvard-MIT data center).
For
those interested in one of the three groups sampled, one of the best data
sources available. The survey aimed for
85% coverage of these Latino groups, using the 1980 census as a basis for
selected area probability sampling.
Respondents were interviewed if they had at least one parent, or two
grandparents, solely of Mexican, Cuban or Puerto Rican origin. Respondents could be interviewed in English
or Spanish. Variables cover the
respondent’s family history, citizenship, organizational memberships, political
participation, voting practices, preferences on policy issues, views toward
political parties and political candidates/leaders, sources of political
information such as the media, feelings about political trust and efficacy,
perceptions of the relationship between government and Latino groups, and
degree of concern about international issues and social problems. Standard demographic and socio-economic
variables are included. The Latino
response rate was 72%, the non-Latino rate was 56%. For those who don’t want to manipulate the micro-level data
themselves, the Principal Investigators published a book of key survey results:
Latino Voices, Westview Press, 1992.
Because
of the gap in time between the 1980 census and the NLPS sampling procedures,
coverage of the populations of interest might be different from the desired 85%
and tends to be concentrated in certain regions. For those interested in the immigrant experience, Puerto Ricans
represent a difficult case since they are US citizens at birth. Examination of generational differences is
hard in certain groups due to a low number of cases.
http://www.apa-politics.org/ (summary report and descriptive statistical
tables)
Inspired
by the National Latino Political Survey, this is a pilot study of
Asian-Americans’ political behaviors and attitudes. Data are not yet publicly available, but a summary report and
descriptive statistics can be found at the web site above. Total N=1,218, with coverage of six groups
in five cities: Chinese, Korean, Japanese, Vietnamese, Filipino and South
Asians in LA, New York, San Francisco, Honolulu and Chicago.
Probably
the most comprehensive study of Asian-Americans’ political attitudes and
behaviors to date. Includes questions
on political opinions, policies, political participation (e.g. voting) and
organizational involvement (in home country and US). Also covers standard demographic and socio-economic
variables. Respondents could be
interviewed in English or their own language.
Until
the data becomes public-use, it is difficult to provide a better evaluation of
this data source. The summary report
and descriptive statistics available are mostly geared to media
consumption. Given the large number of
groups covered, intensive analysis of any one group would be difficult.
http://swdb.berkeley.edu/info/info.html
Housed
in Institute of Government Studies, UC Berkeley, this is a non-partisan project
that supplies data on California’s electoral districts. The database merges
voter registration and election returns (who is elected and results on ballot
questions) to tract level census data.
The Registrar of Voters or County Clerks collects voter registration data
and electoral returns in precincts in each of the 58 California Counties. The researchers at IGS have aggregated
voting and registration data up to a common geographical level, and then broken
them down to match census data.
The
value of this database lies in the work done merging census information with
registration and voting data. The data
can provide descriptions of political districts or be used in multi-level
statistical analyses. Variables
include: race, ethnicity (selected), age, gender, occupation, income,
education, poverty status and household language. Researchers have coded registration and voting data by some
ethnic backgrounds, such as Chinese, Vietnamese and Korean based on surname
analysis. The data can be read into
spreadsheet and word processor programs.
Because
these data are not micro-level files, beginning researchers should take care
about making causal inferences from the aggregated data. It does not appear that immigration status
or citizenship information from the census has been merged to the district
information, but apparently one could request this information.
http://opr.princeton.edu/archive/cmd/latjourn/ (for information on the study)
http://opr.princeton.edu/archive/cmd/index.html#stata (for data and codebook)
A
study of immigrant adaptation and worker incorporation into the US. In 1973-74, 822 adult male Mexicans arriving
in the US through border crossings at Laredo and El Paso, and 590 adult male
Cubans entering in Miami were interviewed (N=1,412). The immigrants were re-interviewed in 1976 and 1979. (Response rate in 1976 was 70% of the original;
the second follow-up was lower.)
While
the project looked primarily at economic incorporation, it is one of the few
longitudinal immigrant datasets available and one of the few providing some
citizenship and civic engagement data from the 1970s. In addition to demographic and economic variables, the study
asked respondents their citizenship status, their intention to become citizens
and it asked questions about association memberships, participation and the
ethnic composition of those membership groups.
The major limitation of the data is the relative lack of political
variables, and the short time space between migration and the last follow-up
interview. (The last interview was held
six years after entry, but citizenship generally requires a residency period of
five years, thus few had naturalized by the project’s end.)
For
research design and results from the study, see A. Portes and R. Bach. 1985.
Latino Journey: Cuban and Mexican Immigrants in the United States. Berkeley: University of California
Press. Portes and associates have also
published articles on the political variables from the survey.
http://www.icpsr.umich.edu/GSS/index.html (information, methodology,
codebooks, data)
National probability sample
of diverse “social indicators.” Since
1972 over 38,000 respondents have answered 3,260 different questions, many
repeated over the years. For
information on methodology and, especially, questions asked, visit the GSS web
site. Researchers can produce
cross-tabulations (and simple statistics) interactively on the website as well
as download micro-data.
Very good data quality, and
a viable resource for those who want simple cross-tabulations of foreign-birth
with answers to various opinion questions.
However, because the GSS aims to serve a wide audience of social science
researchers, there are limited questions on political incorporation and no
special focus on immigrant respondents.
Questions related to politics, including voting data and opinions on
such things as the effectiveness of democracy and government, can be found
under the subject heading “politics.”
Standard demographic and socio-economic variables appear across the
surveys. A question asking whether the
respondent, his/her parents and his/her grandparents were born in the US is
included in all surveys, as are question on race and ethnicity. One can therefore identify three immigrant
generations, but country of birth is not specified. No question on length of residence in the US.
In any one survey, the
number of foreign-born respondents is small, usually just over 200. A cumulative GSS file is available, and combining
all datasets from 1972 produces about 2,100 foreign-born individuals. Unfortunately, many of the political opinion
questions were only used in a single GSS survey, negating the advantage of
combining data sets. The lack of
information on country of birth is a handicap, but can be compensated by
questions on ethnicity coded in detail.
In the end, probably of limited utility, but the web site is very
user-friendly and worth visiting for those interested in social attitudes and
public policy opinions.
http://www.icpsr.umich.edu/cgi/ab.prl?file=6635 (information and data; also Harvard data
center)
Study
of political and nonpolitical civic participation in the United States. Clustered and stratified probability sample,
with over-sampling of political activists, blacks and Latinos. N=2,517, drawn from an original national screening
sample of 15,053.
Great
data set for broad questions of political behavior, interest and civic
participation. Questions include:
interest in politics, party identification, voting activity, participation in
community politics, campaign activities, and opinions on national and social
problems in the United States.
Respondents also provided information about their own personal
experiences with government programs.
Background variables include education (self and parents), occupation,
income, church activity and religious affiliation, race and ethnicity, age,
gender, and union membership.
Critically, the survey also asked whether the respondent and his/her
parents were born in the US. Immigrant
status can be determined, but not country of origin. Also includes a question on US citizenship.
The
major problem is the small number of foreign-born in the sample—238—and the
lack of country of birth information.
The ethnicity variable can be used in lieu of place of birth since it is
coded in detail (i.e., 46 separate responses are allowed). There is no question on length of residence
in the US. Second generation
respondents can be identified, a major advantage, but again the numbers are
small. In some cases, the quality and
breadth of data collected might compensate for the small immigrant sub-sample.
Results
from the survey have been published in Voice and Equality: Civic Voluntarism
in American Politics. Cambridge, MA: Harvard University Press, 1995 and
various articles by Verba and colleagues.
http://www.ropercenter.uconn.edu/scc_bench.html (for data and methodology)
or
http://www.cfsv.org/communitysurvey/ (for survey results)
National
probability sample (N=3,003) with over-sampling of blacks and Hispanics, and
samples from 41 communities across the US (N=26,230) [with different sampling
strategies across various communities].
Information on methodology, variables and downloadable data (tables and
microdata) available through the web site.
Response rates range from 15-45%.
The survey seeks to measure “various manifestations of social capital
and its suspected correlates.”
Great
resources for measures of social capital, civic engagement and some political
behaviors. Includes questions on trust
(including of other ethnic & racial groups), involvement in organizations
and associations, voting, registration, interest in politics and protest
activity. Includes basic demographic
and background variables related to socio-economic and family status and some
variables related to ethno-racial background (includes race and selected ethnic
codes for Hispanic and Asian groups).
Includes citizenship status, but does not measure migration status
(either through a direct question or a question on place of birth).
The
most significant problem with this dataset for those interested in immigration
is the lack of a specific variable to distinguish immigrants from the
native-born. Potential follow-up
surveys might include this variable.
Interviews were primarily done in English, with some Spanish and some
Cantonese interviews in California.
The
list below enumerates various opinion polls and other surveys that include
questions on political attitudes and participation and which over-sample
immigrant and minority groups. In each
case, information on the sponsoring organization is provided. In most cases, the raw data are not publicly
accessible, but these sites usually include reports draw from the original
data. Those interested in pursuing
further analysis of these polls and surveys should contact the sponsoring
agency or Principal Investigator to gain access.