Suppose one wants to generate a number of publications from a single research idea or project, because other things being equal, more publications is better. What is a reasonable way to do so; one that does not dilute or compromise the quality of ones papers?
There is one trend that I see around me, which is in some cases to split a piece of work in two pieces: a high-impact letter, and follow-up full paper with all the details.
There are many reasons why this might be a good idea. Increase in number of publication is one of them, but there are others:
If you work in a rapidly developing and highly competitive field (i.e. you risk being scooped if you wait 3 more months to publish your idea), this allows you a fast publication of the idea and first results (proof of concept, if you will). Also, letters tend to be reviewed more rapidly, which also decreases publication time. Then, you will publish all the details, influence of method parameters, etc.
If your work would be interesting to a wider community, it allows you to deliver two different messages (or at least, the same message at two different levels) to two communities. This increases the overall impact of your research.
Of course it depends how narrow or wide you define what a "project" is.
I'm chemometrician working with vibrational spectra of biological samples. So there are:
- Applications (e.g. a particular biological/medical question)
We write papers about medical diagnostics or biochemical characterization of samples or biochemical changes that occur with some disease, ...
Even within this "application" topic, there may be distinct subtopics. E.g. basic research about a disease is different from developing a diagnostic method.
- As chemometrician, I develop data analysis methods (often triggered by the application).
We write methods/theory papers.
- We also develop instrumentation to measure our samples.
We write papers about that as well
These separations are sensible to me: A reader who wants to learn about a particular disease may not want to dive into chemometric theory development but instead wants to see biochemical findings. Another reader may be interested in the chemometric details but not in the particular disease. Readers looking into instrumentation details may not care about the disease or the particular statistical model applied to our data.
This will probably vary from field to field, but I've seen instances of publishing different analyses in different papers. One rich data set can yield many different analyses which may be largely unconnected with each other.
By way of example, a single longitudinal study of depression may collect data on a number of fronts; information about the participants (gender, age, location, etc), their depressive episodes, family history, genomics, neuroimaging, etc. Each of these can result in a different set of analyses, many of which would be of interest to completely different fields. A neuroscientist interested in activity patterns in the brain would not necessarily be interested in a study examining instances of suicide in high-SES vs. low-SES populations.
By thinking carefully about study design before collecting data, you can position yourself to examine questions in many different fields, leading to numerous publications.
As a final point, I'll just briefly mention that cross-disciplinary collaboration is a wonderful thing, and is very relevant to this discussion.
If you have a specific research question/problem, and devised a technique, method, algorithm or system, to address it, then I think it makes more sense to showcase that method in a way that seems natural, in a single self-contained publication, to the extent that it is possible within space limits, rather than artificially breaking it up across multiple publications solely to increase the number of publications. This will, first, annoy readers, who don't care about your publication rate. It will likely also make editors/reviewers unhappy. Of course, if the method naturally breaks up into more than one paper, than that is fine.
To generate multiple publications around a single method, I think a good way to go would be to write additional papers that supplement and possibly enhance a single major publication. For example, suppose you were to devise a new technique to analyse some data set(s). Then additional publications could be, for example, an extension of that technique to analyze additional, different, data sets. Also, you could have a separate publication which just describes the details of the software implementation. This would not fit well into a (say) statistical research paper, but could be a perfectly viable publication on its own.