When is a co-authorship warranted? (April 4, 2006), Category: Human side to statistics

I am co-author on over 60 papers and have helped with the publication of many more papers. What does it take to get a co-authorship? I don't quibble a lot about this, but it seems everyone has their own standard. When I am asked, I tend to discourage listing me as a co-author if all I did was perform a routine (routine to me, anyway) data analysis. If the analysis is very difficult and/or uses new and uncommon approaches, then I would tend to seek co-authorship. Also, if I helped with writing a substantial section of the paper itself, co-authorship is probably warranted.

When you list yourself as a co-author, you are making an implicit guarantee about the paper. You are willing to vouch for the credibility of the study and you have to have some level of familiarity with all of the aspects of the research. There's a reasonable limit to this of course, or no statistician would ever be listed on a medical journal article. Still, it is important to keep in mind that the medical journals have had to deal with several controversies over honorary authorships where someone not directly involved in the research is included, either as a reward for their diligence in supervising a lab or in an attempt to gain extra credibility for the research.

There are also opposite abuses in a recent ghost writing scandal. Some papers were penned almost entirely by a professional writer hired by a pharmaceutical firm and then offered to prominent academic researchers to be published under their names. This hides the obvious conflict of interest that a paid professional writer would have and allows a drug company to get favorable statements published that their marketing representatives can use to promote their products.

There's a lot that has been written recently about abuses of co-authorship, including the recent cloning scandal in South Korea. When I get time, I want to include a few links and references about this.

07/08/2008.