|
Abstract:
|
Independent surveys that measure the same quantity typically produce different estimates. While sampling error is one obvious explanation, there are many non-sampling error sources that could contribute to an even greater variation across different studies. In this paper, the authors compare important tobacco-use estimates from several national surveys. They explore variations in the definitions, timing, instrument wording, interview modes, and other methodological dissimilarities that might contribute to those differences. The authors' objective is to inform users of these tobacco-use estimates about notable differences between the surveys and to show the impact these differences can make on the resulting estimates.
|
