• Presentation

Does Context Really Matter? Results from a Spanish Language Advance Letter Pilot


Carley-Baxter, L., Link, M. W., Roe, D., & Quiroz, R. S. (2006, May). Does Context Really Matter? Results from a Spanish Language Advance Letter Pilot. Presented at American Association for Public Opinion Research 61st Annual Conference, Montréal, Canada.


The Spanish-speaking population in the United States has grown substantially over the past several decades, yet survey participation levels among Spanish-speakers have not kept pace. In random-digit-dialed (RDD) surveys in particular, Spanish-speakers are often under-represented, potentially limiting the validity of and increasing the bias associated with survey estimates. While the use of advance letters has been shown to improve overall response rates in telephone surveys, their utility within the Hispanic community is unclear. Typically, advance letters contain a Spanish language translation of a letter originally developed in English, but rarely is the letter content initially developed for a non-English speaking population. Here, we pilot the use of tailored Spanish language advance letters for persons in areas thought to be primarily Spanish-speaking and compare the results to a group who received a standard English language letter with Spanish translation and a control group who did not receive a letter. A pilot study was conducted as part of the Behavioral Risk Factor Surveillance System (BRFSS) in four states (Arizona, Texas, Florida, and New York) during 2005. The content of the tailored letter was developed based on information obtained from focus groups conducted with speakers of different Spanish dialects. For the survey, likely Spanish-speaking households were sub-sampled from the regular BRFSS monthly samples in each state based on either reverse matching telephone numbers with a Hispanic surname list or telephone numbers in a telephone exchange in which more than half of the households were believed to be Hispanic based on Census information. These telephone numbers were then randomly assigned to one of three groups: tailored Spanish language letter, English with Spanish translation letter, or no letter. In the analysis, we compare response rates, respondent demographics, and selected survey estimates obtained across these three groups.