The 13th World Conference on Tobacco OR Health

Objective: Accurate population surveillance of tobacco use is needed for planning and evaluation of tobacco control efforts. Over the past decade, decreases in telephone response rates have led researchers to want to consider web-based data collections as low-cost alternatives to compliment these survey efforts. This study evaluated the comparability of the two survey methods and assessed the possibility that combining both survey methods in a mixed-mode survey could generate solid estimates.

Methods: We gathered data through an Adult Tobacco Survey of 2,400 adults in a random-digit dialed (RDD) survey and 6,000 adults in an internet based data collection. The internet sample was adjusted by propensity scoring for web access and likelihood of response. The study compared unweighted data to understand underlying differences in the raw distributions based on the two samples. It compared weighted distributions for the two samples to identify potential sources and extent of bias. The study developed and tested methodologies for combining the two sets of weights using design- and model-based strategies for the estimation process including variance calculations and significance testing.

Results: The surveys conducted in the fall of 2005, allowed comparison of RDD and web based survey estimates. Estimates were within a few percentage points of one another. We developed innovative efforts to combine both the RDD and the web-based estimates resulting in more accurate measures of the tobacco use.

Conclusions: These findings highlight the value of both data collection methods and suggest ways that web-based surveys to compliment RDD survey estimates.