We describe the feasibility of audio-enhanced personal digital assistants (ADPAs) for data collection with 60 Latino migrant farmworkers. All participants chose to complete APDA surveys rather than using paper-and-pencil. No one left the study prematurely: two (3%) data cases were lost due to technical difficulties. Across all data .27% missing data were observed: nine missing responses on eight items. Participants took 19 minutes on average to complete the 58-question survey. The factor most influential for completion was education level. APDA methodology enabled both English-and Spanish-speaking Latino migrant farmworkers to become active research participants with minimal loss of data.
Keywords migrant farmworker; data collection; data qualityThe Latino population in the US is the most rapidly growing ethnic minority with 28 million American residents with Mexican nativity or parents with Mexican birth (U.S. Census Bureau, 2006). An estimated 3-5 million migrant farmworkers are in the US; 90% are of Latino ethnicity, with the majority originating from Mexico (U.S. Department of Labor [USDoL], 2005). As a group, migrant farmworkers display worse health status indicators than the majority population (Kandula, Kersey, & Lurie, 2004;Rubalcave, Teruel, Thomas, & Goldman, 2008; United States Department of Health and Human Services [USDHHS], 2002a). Levels of education influence reading ability, and although the highest grade completed varies by place of birth, foreign-born migrant farmworkers on average complete the sixth grade in contrast to completion of the 11th grade by U.S.-born migrant farmworkers (USDoL). In addition, 81% of the migrant farmworkers speak predominantly Spanish (National Center for Farmworker Health, 2009 When collecting data from low-literacy or non-English speaking populations, interviewers are often needed to read or translate questions (Aday & Cornelius, 2006). Face-to-face interviews provide the researcher with opportunities to clarify study questions for participants, answer questions, and improve data quality through knowledgeable navigation of the survey (Czaja & Blair, 1996). These benefits should be weighed against the intense time and staffing resources required for face-to-face interviews, as well as the potential introduction of social desirability bias Tourangeau & Smith, 1996). Potential sources of error can also be introduced when an instrument that is designed to be self-administered is read to subjects with low levels of literacy (Al-Tayyib, Rogers, Gribble, Villarroel, & Turner, 2002;Lange, 2002).To address the high cost of face-to-face interviews, as well as the difficulties of reading and navigating paper-based questionnaires, researchers have developed audio computer-assisted self-interviewing (ACASI), which has been used successfully in various populations and shown to improve data quality and standardize survey administration through faster data entry and fewer non-responses (Ramos, Sedivi, & Sweet, 1998). The use of audio recording has been shown to reduce iss...