Pew Hispanic Center Survey of Mexicans Living in the U.S. on Absentee Voting in Mexican Elections
Appendix 3: Methodology Report
THE PEW HISPANIC CENTER
THE MEXICAN VOTERS SURVEY
In order to fully represent the opinions of Mexico-born Americans living in the United States, ICR conducted interviews with a statistically representative sample of Mexicans so that they could be examined nationally, as well as in target regions of high Mexican concentration.
This survey was conducted by telephone from January 16, 2006 to February 6, 2006 among a nationally representative sample of 987 respondents. The margin of error for total respondents is +/- 4.37%. A total of 62 are registered to vote and 922 are not registered to vote. The margin of error for total registered to vote it is +/- 18.20% and for those not registered to vote it is +/- 4.51%
For this survey, ICR maintained a staff of Spanish-speaking interviewers whom, when contacting a household, were able to offer respondents the option of completing the survey in Spanish or in English. Of the total Mexican interviews, 96% were conducted predominantly in Spanish.
The survey was administered to any male or female age 18 and older that is of Mexican origin or descent.
The field period for this study was January 16 through February 6, 2006. The interviewing was conducted by ICR/International Communications Research in Media, PA. All interviews were conducted using the Computer Assisted Telephone Interviewing (CATI) system. The CATI system ensured that questions followed logical skip patterns and that the listed attributes automatically rotated, eliminating “question position” bias.
A stratified sample via the Optimal Sample Allocation sampling technique was used for this research. Since the incidence of Mexico-born Adults is highly correlated to geography, it was decided early in the process to only interview in areas of reasonable incidence of this population. Covering 100% of this population, that is, giving every Mexico-born American a chance to be selected in the survey, was not at all cost effective. However, given the concentration of this population in certain areas, it was decided that a cover of at least 80% of the population would suffice. As such, states with the highest incidence of this population were selected to be part of the study. These included California, Texas, Arizona, Colorado, New Mexico, Nevada, Florida, New Jersey, New York, Illinois, and Nebraska. This comprised of 86% of all Mexico-born Adults living in the United States.
For this project, telephone exchanges in these states were first divided into high and low incidence strata. Then each strata was run against listed databases to identify households with a Hispanic surname. Then the strata were divided into RDD-no surname and RDDsurname substrata, for a total of four strata overall. The low incidence non-surname strata was not dialed due to extreme low incidence. As is usual with the disproportionate stratified design, high incidence strata were dialed more heavily than lower incidence strata via an optimal allocation method.
Weighting and Estimation
The data was weighted by first using a preweight to correct for the disproportionality across strata. This brought the complete back into proper proportions by region state, and telephone exchange. Then, post-stratification weights were used to weight by age, gender, and education.
The overall response rate for this study was calculated to be 56.6% using AAPOR’s RR3 formula.
Following is a full disposition of the sample selected for this survey: