2008 Collaborative
Multi-racial Post-election Study (CMPS)
The 2008 CMPS is a
national telephone survey of registered voters, with
comparably large samples of African Americans, Asian
Americans, Latinos, and Whites. The survey is a
collaboration between professors at Texas A&M; UCLA; UC
Riverside; UNM; USC; and UW-Seattle.
Here are links to the survey toplines by the four major
racial/ethnic groups. We also provide more details about
the survey’s design and methods below.
Survey toplines (weighted by age,
gender, and education for each racial group)
Thank you for your interest in the CMPS.
Sincerely,
Lorrie Frasure Matt Barreto
Sylvia Manzano Ange-Marie
Hancock
Ricardo Ramirez Karthick
Ramakrishnan
Gabriel Sanchez Janelle Wong
CMPS
Survey Design and Methods
This telephone survey—conducted between November 9, 2008
and January 5, 2009—is the first multiracial and
multilingual survey of registered voters across multiple
states and regions in a presidential election. In contrast
to the 2008 ANES which oversampled Black and Latino voters,
and was available in Spanish, the CMPS was available in six
languages and contains robust samples of the four largest
racial/ethnic groups: Whites, Latinos, Blacks, Asians . The
CMPS contains 4,563 respondents who voted in the November
2008 election and that self-identified as Asian, Black,
Latino, and White, and was available in English, Spanish,
Mandarin, Cantonese, Korean, Vietnamese and respondents
were offered the opportunity to interview in their language
of choice.
There are six states in the country where representative
studies will yield robust samples of all four major racial
groups. These states include California, Texas, New York,
Florida, Illinois, New Jersey, and our statewide samples
range from 243 to 669 cases. In order to arrive at more
nationally representative samples of each minority group,
we added two supplemental states per racial group,
including Arizona and New Mexico (Latinos), North Carolina
and Georgia (Blacks), Hawaii and Washington (Asians). Of
these 12 states, 3 were considered political battlegrounds
in the 2008 Presidential electorate – New Mexico, Florida,
and North Carolina. In order to examine multi-racial
politics in competitive and non-competitive environments,
we supplemented our sample with six additional diverse
battleground states : Colorado, Michigan, Nevada, Ohio,
Pennsylvania, and Virginia. As of the 2008 election,
two-thirds of the national electorate is concentrated in
these 18 states. For Latinos, 92% of all registered voters
reside in these states; 87% of Asian Americans; and 66% of
Blacks, and 61% of Whites.
The sample was drawn of registered voters using the
official statewide databases of registered voters,
maintained by elections officials in each of the 18 states.
For voters without listed phone numbers, records were
enriched using a combination of public and private sources
of consumer information by Catalist Data Services. The
vendor classified the sample by racial/ethnic group based
on a combination of variables: first and last name,
population density, and consumer information. In 2008,
Catalist had the most comprehensive database of registered
voters and is particularly useful in studying multi-racial
populations because of their classification methodology.
All interviewers were conducted by live interviewers, via
telephone to landlines and mobile phone numbers when they
were directly supplied by registered voters on the voter
list. The survey contains an overall sample of 4,563 with a
margin of error of 1.5% for the national sample, 2.5% for
the Latino sample, and 3.2% for the White, Black and Asian
samples. Overall, the AAPOR response rate-1 was 11.4% and
response rate-3 was 41.9% The average survey length was
24.8 minutes.
Post-stratification weights were applied to correct for any
discrepancies for age, gender, and education which may
accompany telephone surveys. The November 2008 CPS provides
estimates of the registered voter population by race, age,
gender, and education level which we applied to our sample,
by racial group, so that our distributions match those of
the Census on these important demographic categories.
There are 51 items dealing with sociopolitical attitudes,
mobilization and political activity. Additionally, there
are 21 items that capture demographic information,
including: age, ancestry, birthplace, education, ethnicity,
marital status, number in the household, religiosity,
gender, media usage and residential context.