This article is only available in the PDF format. Download the PDF to view the article, as well as its associated figures and tables.
To the Editor:—
From my experience as a statistical consultant to research physicians, it appears that most (research physicians) consider the P value to be the main concept of statistics. Perhaps this is due to the semantic attractiveness of the phrases "statistical significance" and "statistical tests." "Significance" in medicine means something like "importance" while tests are the most objective procedures a physician performs. To many nonstatisticians the statement "My data are statistically significant" sounds like "If I took my data to a statistician he would agree with my conclusions." "Statistical significance," ironically, is merely a technical phrase to a statistician having little to do with importance or his total judgment; "tests" to a statistician are no more objective than anything else he does.As semantically attractive as are the aforementioned phrases about significance and tests, "estimation" has a very unscientific connotation to most nonstatisticians, suggesting off-hand guessing. In fact, the
Boen JR. P Values Versus Means and Standard Errors In Reporting Data. JAMA. 1969;208(3):535–536. doi:10.1001/jama.1969.03160030109025