Survey Nonresponse Bias, HIPAA, and Mixed Data Collection Modes

DESCRIPTION (provided by applicant): Large sample surveys have long offered critical information on the occurrence and determinants of health and disease in the population. However, participation in general population surveys has been declining over time, potentially threatening the quality of the data that surveys provide. The true impact of the downward canting of response rates is unclear, as the relationship between response rates and response bias which had traditionally been assumed to be strong has recently been called into question. This proposal seeks funding to conduct a systematic analysis of survey non-response bias using an unprecedented amount of health-related information on both respondents and non-respondents to determine how non-response bias might be affected by the inclusion of a Health Insurance Portability and Accountability Act (HIPAA) Authorization Form (HAF) and the use of a mixed-mode data collection design. Our specific aims are to: 1). Investigate the effect of including a HIPAA Authorization Form (HAF) on survey non-reresponse bias. There has been speculation that HIPAA has muted participation in research and introduced bias, but little direct evidence to substantiate those claims exists. 2). Assess whether a mixed-mode survey design (mail and telephone) reduced bias and enhanced the post-stratification weighting methods commonly used to correct for non-response bias. Recently, mixed mode surveys have become an increasingly popular strategy for maximizing response rates but their impact on reducing non-response bias and improving post-survey adjustments is unclear. The major goals of the research supported by the National Institutes of Health (NIH) are to lead the way toward important medical discoveries, investigate ways to prevent disease, and identify the causes, treatments, and cures for common and rare diseases: surveys provide vital information in each of these areas. An underlying goal of our work is to make practical recommendations about best practices in survey data collection that could help improve survey response and ultimately, data quality, in a time when conducting surveys is becoming more difficult. PUBLIC HEALTH RELEVANCE: Surveys have long offered critical information on health and disease, but recently fewer and fewer people are willing to complete these questionnaires. If only a few people complete surveys, the surveys may not provide useful information about diseases. In this study, we will determine whether poor participation in a health survey results in collection of bad data. The goal of this work is to provide practical recommendations about best practices in survey data collection that could help improve survey response and ultimately, data quality.