December 22, 2009

Latinos Online, 2006-2008: Narrowing the Gap

Appendix A: Methodology

Data Regarding Hispanics

The analysis of Hispanics is based upon the merging of three datasets derived from landline telephone surveys conducted by International Communications Research:

All three surveys used similar methodologies. Surveys were conducted by telephone, and targeted U.S. Latinos ages 18 and older, who had the option to respond in Spanish, English, or a combination of the two languages.

The samples for each survey were independently drawn using random digit dialing (RDD). The 2006 National Survey of Latinos (NSL) Survey and the 2006 Hispanic Religion Survey samples were stratified according to the density of Hispanic population and ancestry groups. The 2008 Post-Election Survey sample was stratified according to the density of the Hispanic population.

Following data collection, the data from each of the three surveys were corrected for the disproportionality of the stratification schemes. For the 2006 NSL, data were also weighted based upon age, sex, education, ancestry, years since entering the U.S., and nativity using 2005 Current Population Survey data. For the 2006 Hispanic Religion Survey, data were also weighted based upon age, sex, education, ancestry, years since entering the U.S., nativity, and religion using 2006 Current Population Survey data, as well as past Pew Hispanic Center survey results. Results for the 2008 Post-Election Survey were post-stratified using national estimates of age, gender, education, ancestry, years since entering the U.S., nativity, and region obtained from the 2008 Current Population Survey.

Data Regarding Non-Hispanics

The data regarding blacks and whites are based upon landline interviews from five surveys conducted by Princeton Survey Research Associates for the Pew Internet & American Life Project from 2006 to 2008:

All five surveys used similar methodologies. Surveys were conducted by telephone, and targeted U.S. adults ages 18 and older. The samples for each survey were independently drawn using random digit dialing (RDD).

Following data collection, post-stratification weights were applied to the data. The weights were based upon population parameters for age, gender, education, race/ethnicity, and region. For the 2006 surveys, these weights were derived from the 2005 Current Population Survey, and for the 2008 surveys the weights were derived from the 2007 Current Population Survey. In addition, a county population density factor was calculated based on the 2000 Census. This parameter was then compared with the sample characteristics to construct the sample weights.