Early career advice for Jr. Researcher on selectivity of authorship credit for collaboration in areas outside one's specialization

Question

I have a problem...Once folks learn my academic specialty within the medical sciences and of my strong background in statistics, they often want to form a collaborating - mostly they want me to conduct analyses, develop the study methods, write methods, results and discussion sections of their manuscripts. All this in exchange authorship credits and no actual pay. More often than not, these offers come from clinicians with little experience in human subject research and statistics, who specialize in research topics vastly different than my area of focus.

These days I am weary of getting involved in these projects because when I did do so a few years ago, I became very worried about the integrity of the data and the scientist with whom I worked with. More importantly, I worried the work would at some point be retracted and consequently tarnish my reputation in the future. These concerns arose due to the fact that I had many questions about the data the study was based on and had limited access to the raw data.

The above-mentioned predicament worries me as I am at the beginning of my scientific career and I suspect that many of these clinicians are only temporarily vested in research to amp their publication records for the purpose of promotion. For them, these publications are one-offs and would have less devastating consequences on their careers if retracted. For me, its very different as I am strictly a researcher. It worries me that working on projects where I do not have full control/access to the raw data, and by working with clinicians with little scientific training in human subject research will result in catastrophic outcomes for me.

My questions to the forum are:

1) am I being too picky about who I collaborate with at this stage of my career?
2) should I relax my standards?
3) what are the consequences of being included as an author in poorly conducted research early on in one's research career for an aspiring academic?

Answer 1

I don't think that you're too picky or that you should relax your standards.

If you are responsible for creating a statistical model to test the data, then you should know everything there is to know about the process that generated the data that you are modeling. If you do not know everything and if you don't have access to all the data and any material relevant to how the data was generated (e.g. protocols), then you cannot in good conscience model the data. The reason for this is that you don't know where to look for problems that might invalidate your model. You might, therefore, overlook something. And if the wrong statistical model is used, then the sampling distribution of the parameters and your test statistics will be wrong. They will be wrong because the premises of your argument will be false.

Ideally, as the person conducting the statistical analysis, you will be involved in collecting the data. This means that you should be involved in designing the experiment. Fisher once said the following about involving a statistician after the fact:

"To consult the statistician after an experiment is finished is often merely to ask him to conduct a post mortem examination. He can perhaps say what the experiment died of."

In the case of an observational study this might not be as bad as you could collect other data after you're involved. But it is not ideal.

What I would recommend you to do is to use the argument that you cannot possibly model the data if you don't have access to all the materials and know how the data were generated. You can use this argument to justify yourself to your co-workers. And you can use this argument to justify yourself towards yourself for not working on certain projects. You're the expert here and you need to tell yourself and others what you need to be able to do your work. And if what you need isn't there, then you can't do your work. Period. This way you avoid re-thinking, worrying, and arguing about what you should work on and what not.

In addition, I would recommend you to document everything for each paper to make it clear what your responsibilities were and what the responsibilities of other people were. In particular, in those cases where you did not have access to the raw data.

As for the consequences of being part of a poorly conducted research. It will tarnish your reputation. By how much depends on what the norm is in your field and in your peer group. Your reputation is how you are judged by the standards of other people. As I am not in your field or know you (or what research you were involved in or would be), I cannot say how much of an impact this will have. But it will definitely be bad for your reputation. It might be worth it though. Although this depends on what your goals in life are.

Answer 2

Our institution, and many others Faculties of Medicine, have statisticians fully employed. Please make sure if your institution has one. These professionals are never coauthored since it is their job to analyze data. Same as most of the time you don't coauthor technician (for ex. TEM, AFM and other machine operators ).

However, all of them need to follow work ethics, if there is a huge error in data it is their duty to report. We had a problem when one professor pressured this stuff to process data according to his view ( confirmational bias ). I assume if your institution has one and teams you collaborate don't use it, either something fishy is going on, or the person responsible for the statistic is too busy. Which can happen depending on the size of the institution.

If you are the coauthor, then you are one among equal. That being said, one who submits manuscript sometime need to prove and confirm that all authors agreed upon this version. So you should not be picky, but rather "pushy" of your concerns.

2) should I relax my standards?

It is up to your academic and personal integrity. I know some authors that don't want even considering submission into low impact factors. Be aware if you are marked as the hard cooperating person, later on, you may find difficulties to establish colaboration, but that shouldn't matter to you if you have integrity and credibility!

what are the consequences of being included as an author in poorly conducted research early on in one's research career for an aspiring academic?

There are services ( similar to proofreading and editing of the manuscript ) that are offered in order to fit the data of your research in a way that will confirm your theory. I saw an article on Elsevier or Springlink, that was retracted because using this services. In my opinion, consequences are only if you submitted published and later retracted journal for academic promotion.

Consequences in sense of publishers, you might get blacklisted. There is a website that tracks all retracted articles and exposed professors. As I read, there is some Japanese author, cancer research, that got 8 times retracted. Nothing really significant happened at his home institution. Link to that website is on this site, and it was reference in question about postdoc.

Answer 3

In addition, I would recommend to document everything you did for each paper to make clear what your responsibilities were and what the responsibilities of other people were. In particular, in those cases where you did not have access to the raw data.