Details of 2004 Hispanic Media Study Sampling

The researchers requested that the Hispanic Media Study provide not only a nationally representative picture of Latinos, but also an accurate measure of Latinos within the Los Angeles metropolitan area. As such, the study utilized two levels of stratification. Primary stratification was at the regional level, using two basic, census-defined geographies—the Los Angeles metropolitan statistical area (MSA) and the remainder of the country. The second level of stratification was accomplished at the telephone exchange-level (NPA—NXX). Specifically, telephone exchanges were arranged by order of Latino incidence and divided into strata. Overall, then, the study had seven strata; very high, high, medium, and low incidence strata for the non-LA United States, and high, medium, and low strata for the LA MSA. The overall response rate for this study, using the American Association of Public Opinion Research's RR3 response rate formula, was 50.5 percent.

Weighting and estimation were performed independently within the four national strata and three LA oversample strata. The first phase involved the adjustment of the actual final sample sizes to proportionality. Within region, the population totals were determined from the March 2003 Current Population Survey (CPS). An initial weight, or proportionality factor, was then computed for each of the seven segments. Finally, within each of the seven strata, interviews were balanced using a sample balancing routine controlling for age within sex and education. The balancing process was controlled to produce weights scaled to the earlier determined proportionality weights. The design effects based on the disproportionate stratified sampling scheme and post-stratification weighting were 1.6 for the total study and 1.17 for the LA oversample.