Catching up on reading over the holidays, I came across
David L. Vaux's piece in the December 13th issue of
Nature magazine titled,
Know when your numbers are significant. The subhead summarizes the article this way, "Experimental biologists, their reviewers and their publishers must grasp basic statistics, or sloppy science will continue to grow."
I'm sharing this article because I'm sure the message resonates with the statistical community. I like that the messenger is a non-statistician. On the other hand, I notice some criticism of the statistics in the reader comments.
If a statistician were to write such a piece for scientists of a specific discipline, what would you write? I'd love to see your comments in the comment space below or in an email to
me.
With 2013 being the
International Year of Statistics, this is a good opportunity for statisticians to be more active in such guidance. Already, Carl Bialik - the WSJ's Numbers Guy - has compiled general guidance for his readers in his December 28 column,
Statistical Habits to Add, or Subtract, in 2013.
Indeed, guidance specific to scientists might complement American Statistical Association President Marie Davidian's initiative to "advance the visibility of statistics and statisticians among our fellow scientists, targeting the American Association for the Advancement of Science (AAAS), the world’s largest general scientific society" (as she writes in her
Amstat News column, "
Not Just a New Year … An International Year of Statistics! The December
Amstat News column,
AAAS + Statistics = U, has more about this initiative.)
In his Nature article, Vaux notes the "incidence of papers in cell and molecular biology that have basic statistical mistakes is alarming." He then goes on to "spell out the basics that every experimental biologist should know." He even has a statistical glossary that includes standard deviation, standard error of the mean, confidence interval, independent data, replicate data, and sampling error.
In order to achieve improved understanding and use of elementary statistics, Vaux says young researchers should be taught the basics of statistics as soon as they get the results of their first experiment. To influence established researchers, Vaux has several recommendations for journals. Vaux closes with an observation on the influence Strunk and White's
The Elements of Style has had for English grammar and suggesting experimental biologists could use a similar booklet on statistics.
[2/21/13: See also,
Statistically speaking, we don’t use calculus much, an op-ed by Biology Professor Dave Gammon praising the utility of statistics.]