This analysis was conducted using data from the 2004 Hispanic Media Study. The fieldwork for the Hispanic Media Study was conducted between February 11 and March 11, 2004, by International Communications Research (ICR) of Media, Pennsylvania, which used a disproportionate stratified sampling design to complete nationally representative interviews of 1,316 Latinos. The survey, which was designed, analyzed, and sponsored by the Pew Hispanic Center, gathered data on respondents' family origins, media use, language use by media, political attitudes on immigration, acculturation and assimilation, language proficiency, and demographics.

The telephone numbers for the completed interviews were run through a number of databases to filter them for listed households with Latino surnames. Forty-one percent of the respondents were found to have Latino surnames (43.7 percent unweighted, N = 570).

To investigate the differences between these and non-Latino surname respondents, twelve demographic variables were selected from the survey, as well as five media-consumption variables, five corresponding language-of-media-consumption variables, and two Latino identity variables.

The demographic variables included gender, years living in the United States, citizenship, marital status, number of people in household, employment status, education, age, income, region, country of origin, voter registration, and party identification.

The news media consumption variables included questions on whether respondents got their news from network television, local television, newspapers, the radio, or the internet, and the media language consumption questions asked whether respondents got their news in Spanish from each of these respective sources.

Finally, the Latino identity variables were comprised of a self-identification item in which respondents were asked whether they called themselves Latino or Hispanic, and an acculturation measure.

Data analysis was then conducted using independent samples t-tests of surname status on all the aforementioned variables, as well as Spearman r nonparametric correlations.