In a current project of mine, I analyze a certain probability distribution to model a biological process. This probability distribution is not "mainstream", but it is relevant enough to have its own name (general projected normal), and it is acknowledged to be a very complicated distribution.

In this project, I had to derive analytic expressions for approximating the moments of this distribution (actually, of a slightly more general version of the distribution). I also spent some time converting the messy resulting expressions into nice matrix formulas that can be computed efficiently, and implementing them in code.

My question is that I am not sure whether the work I put into the statistical model should be buried in the supplementary of my computational biology paper, or whether I should try to publish it separately.

The biology paper will be focused on explaining how these formulas relate to the process in question, and relating their behavior. The formulas would be highlighted in the paper, but, for example, the name of the distribution (projected normal) would not make the title, and maybe not even the abstract. The derivations would be in the supplementary. Thus, I'm worried that if these results are a useful contribution to the statistics literature, they would be difficult to find.

On the other hand, although my derivations were very involved for me as a non-statistician, I'm not sure what is general interest for these formulas, and I am not sure whether the "difficulty" of these derivations would merit an independent statistics publication.

How can I tell whether publishing this analysis in two papers, one for the statistics derivations, one for the biology applications makes sense? Or whether I should publish everything in one paper, and risk having a useful statistics contribution go unnoticed.

  • 6
    Check paper and journal, "On maximum likelihood estimation of the general projected normal distribution", Journal of Statistical Computation and Simulation Volume 91, 2021 - Issue 16 Feb 13 at 16:44
  • 6
    You have to be careful and check for related work regarding your derivation, lest you become another biologist who discovers integration Feb 13 at 18:57
  • 2
    @user4052054 The post you linked to was from 2007, and mentioned that said article was at the time cited 75 times. Now it has 553 citations in google scholar. Feb 13 at 19:12
  • Here's hoping that most of those citations are disparaging, though.
    – alexis
    Feb 14 at 15:57

4 Answers 4


Yes, publish as two papers. It is more likely that someone in a field unrelated to yours would succeed in locating the model in a statistical journal than in a journal related to the applied domain of your work.

The issue I think will be to choose an appropriate statistical journal for the statistical model. A few journals only seem to publish work that breaks entirely new ground, but most journals in most fields publish what one might call incremental results. Phrasing matters in that way should not be taken as a denigration. Most work is incremental rather than earth shattering ... and it is frequently very useful to find a paper that fills in a gap in one's own knowledge.


This should be a separate statistics paper

What you have described is a substantive contribution to statistical theory and so it would be reasonable to publish the mathematical/statistical analysis of this distribution as a stand-alone paper in a probability/statistics journal. This will allow you to develop the theoretical analysis in detail as the main focus of the paper and put this work within the relevant theoretical context with an appropriate literature review of the statistical area. Your biological analysis in your existing paper can then reference the statistical material from your other paper using standard referencing.

Separating these things into two papers has the advantage that it allows each paper to focus on the substantive thing of interest without being side-tracked with long and complicated matters that are ancillary to the main subject matter. What you have described here sounds like a contribution to statistics that would be useful and would ---in principle--- be publishable.


Actually, publishers control what is published, not authors. You can submit one or two papers and decisions will be made.

However, from your description, submitting two papers seems reasonable, assuming that you can justify the significance and applicability of the model on some problem. Then you need to decide if the biology paper, using the already published paper makes a significant advance in that field.

I think I would try that strategy (two papers) first and if the model paper is rejected, combine them into one and try again. But, you may have to wait for a decision on the model paper before it makes any sense to submit the biology paper. All of that can take a while. That may work for you or not. I suspect that the journals for the two paper strategy would be different.

Any paper you submit will be judged on its merits.

  • 3
    Thanks. Maybe I wasn't clear in my question, but I am well aware of how publishing works (e.g. I don't get to decide if something gets published). My question is, given how time consuming it is to write and submit a paper, what are some tools I could use to evaluate whether it makes sense to write 2 different papers. Mostly, how can evaluate whether the statistics paper would be interesting for the statistics community
    – dherrera
    Feb 12 at 22:47

This is the constant struggle of working in interdisciplinary research.

Let me ask you another question - what is your ultimate career trajectory? Do you want to be known as a computational biologist who is very statistics savy? Or a statistician that tinkers in biology? Realistically, you can only pick one, and it'll help you determine what types of venues you can focus your efforts on.

It takes about 2 years to publish a statistics paper — so you could either spend time wooing two sets of reviewers (with conflicting agendas), or you can reinvest it in your next paper. Regarding visibility, I wouldn't worry too much. There are enough statistics-focused folks in biology that will jump to the supplemental methods. If your method has proven value and very large improvement over competing methods, you will get visibility eventually.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .