Wednesday, September 10, 2014

From The Haiku of Writing a Paper: Introduction

Some people have said that I can write papers quickly. I'm not sure if that's true or not, but I have learned to write them faster than before. And part of that is having a standard structure to guide the writing process.

A long time back, I tried to crystallize what I had learned about putting together scientific papers. Mostly from making a lot of mistakes. I did it mostly just to get all the different ideas organized for myself.

The general approach was to reduce the structure of the paper down to a skeleton outline. Not quite a haiku, but close.

Apparently, the full document has been passed around a bit--I'm always surprised when people tell me they were using it.

As an example, introductions can be tricky to write. In most of my papers I try to follow the 3+1 model. With this model, you funnel from big ideas to specific points to be tested.
First Paragraph: Big question. This is the general broad reference to your work. For example, it might be that atmospheric CO2 concentrations are rising, or nitrogen is an important driver of ecosystem dynamics. 

Second Paragraph: Proximal question. Within the broader framework, the proximal question should be stated that you are directly addressing. For example, although atmospheric CO2 concentrations are rising, the controls over soil C storage are poorly known. Or although nitrogen controls ecosystem dynamic, there are important questions regarding the role of denitrification in controlling N availability. Note that a proximal question can often be framed as the big question—it’s all a matter of perspective and how you want to tell the story.

Third Paragraph:  Scope of research with hypotheses. In order to better understand the role of soil C storage in responses of ecosystem to elevated CO2, we tested whether elevated CO2 increased the C stored in the soil within soil aggregates.

The "Plus One" Paragraph: Competing hypotheses. The best introductions and research designs test between competing hypotheses. Often when there is a single hypothesis that is rejected, the authors can derive alternative explanations that don’t require the theory to be rejected. Therefore, might as well start with competing hypotheses since there are always competing hypotheses. Framing the hypothesis in the null form is not necessary when using competing hypotheses. For example, in testing the role of N in decomposition, an experiment could test whether stoichiometry predicted responses of decomposition to greater N availability. Or we can test between stoichiometry or N mining in predicting the responses of decomposition to greater N availability. No on experiment is generally able to reject a theory, so you can test between two theories and whether data supports one theory or another.

Note, I call the last paragraph the +1 paragraph, because it isn't always there in a paper--it depends on design.

Still, using this model, you should be able to write your introduction in 3 or 4 sentences. Once you have that you can expand each sentence to a paragraph. These can be expanded out more to maybe 2 paragraphs, but if you deviate too far from this, the introduction is likely running long and will be confusing to readers.

There are tricks to writing other sections efficiently and a lot of details to look after as you get through the sections.Still, once you settle in on your framework for papers, it makes writing the paper a lot more fun, since you can concentrate on the message rather than the structure.

No comments:

Post a Comment