20.7 Survey Studies
Statistical concepts, such as the margin of error in a public opinion poll or the probability of rain or snow, appear in everyday conversation. But, just as one may understand how the heart functions and how blood circulates but not be able to perform a cardiac catheterization, an understanding of statistical concepts does not enable one to perform the work of a statistician. Although the concepts may be familiar, the tools of statistics may be misapplied and the results misinterpreted without a statistician’s help.
In medical research, the quality of the statistical analysis and clarity of presentation of statistical results are critical to a study’s validity. Decisions about statistical analysis are best made at the time that the study is designed and generally should not be deferred until after the data have been collected. Even the most sophisticated statistical analysis cannot salvage a fundamentally flawed study. Regardless of the statistician’s role, authors (who may include statisticians) are responsible for the appropriate design, analysis, and presentation of the study’s results.
Many excellent statistical texts are available, and a comprehensive approach is far beyond the scope of this chapter. However, authors, editors, and manuscript editors should have a general understanding of study designs, statistical terms and concepts, and the use of statistical tests and presentation. Although few rules exist to guide how statistics should be presented, presenting statistics briefly but completely and consistently should improve the reader’s understanding of the analysis.