|
Abstract:
|
The Center for Population Sciences at the University of Tennessee Health Science Center currently has five NIH/DOD funded grant projects that are addressing tobacco, alcohol abuse, and obesity prevention in partnership with the US Air Force (USAF). These and future research projects inherently face certain statistical challenges resulting from environmental, policy and behavioral constraints that are present at various stages of the research process and are not in control of the research team. For example during the design stage we cannot have a true control group, or ask alcohol use related questions from underage Airmen. When estimating sample size and power needed for studies we have large number of Airmen clustered on multiple levels such as training flights within squadrons within limited number of USAF bases. We collect outcome measures that tend to be zero skewed, such as intent to use tobacco products or consume alcohol in the future, which requires attention during data analysis. This paper will present several examples and solutions regarding statistical challenges researchers face at various stages, and have to consider while planning and conducting studies in the USAF.
|
