• Presentation

Revisiting the Use of Tailored Spanish Language Advance Letters in RDD Surveys

Citation

Carley-Baxter, L., Link, M., Roe, D., & Quiroz, R. S. (2007, May). Revisiting the Use of Tailored Spanish Language Advance Letters in RDD Surveys. Presented at American Association for Public Opinion Research Conference, Anaheim, CA.

Abstract

Sustained growth in the Spanish-speaking population in the United States makes it increasingly important for researchers to attain representative participation in surveys from this group. Underrepresentation could lead to bias in survey estimates and call into question the validity of survey findings. Research has shown that use of advance letters increases overall response rates to telephone surveys; however, the utility of this approach for Spanish speakers is still in question and little research has looked at tailoring the content of the letter for the specific concerns of this population. In a prior round of this work, we demonstrated that use of a standard English advance letter with Spanish translation increased response rates more than did a Spanish language letter tailored to address concerns about survey participation typically voiced by Hispanics. We hypothesize that over-emphasis on immigration issues may have limited the utility of the tailored letter. Here we report on the second round of this pilot, in which the content of the tailored letter was revised substantially. The pilot was conducted as part of the Behavioral Risk Factor Surveillance System (BRFSS) in four states (Arizona, Texas, Florida, and New York) during fall 2006. The content of the revised tailored letter was developed based on focus groups of Spanish speakers with different dialects. For the survey, likely Spanish-speaking households were sub-sampled from the ongoing BRFSS in each state based on either reverse matching telephone numbers with a Hispanic surname list or telephone numbers in exchanges in which more than half of the households were believed to be Hispanic. These telephone numbers were then randomly assigned to one of three groups: tailored Spanish language letter, English letter with Spanish translation, or no letter. In the analysis, we compare response rates, respondent demographics, and selected survey estimates obtained across these three groups.