Write in this order
Tables and figures first. They are the result. Build them before prose, and the paper’s structure follows from what they show. Writing first and illustrating afterwards produces text that argues past its own data.
Methods second. Factual, and written against the reporting checklist for the design. Drafting it early also surfaces anything that will not be reportable, while something can still be done.
Results third. What was found, in the order the methods described, without interpretation.
Discussion fourth. What it means, what the limitations are, and what follows.
Introduction fifth. Easier once the paper exists, because its job is to set up what the paper actually delivered.
Abstract and title last. Written from a finished paper. Most abstracts that misrepresent a paper were written before it was finished.
Where papers go wrong
The discussion over-claims. The single most common reviewer criticism. The verb in the conclusion outruns the design that produced it.
The abstract disagrees with the paper. Numbers in the abstract not matching the tables, or a conclusion the results do not reach. Checkable in minutes, and routinely not checked.
Limitations are generic. "Small sample size" costs nothing to write. The useful version names the specific threat and its likely direction of effect.
The introduction reviews the literature. It should establish the gap and state the question, not survey the field.
Before submission
Reconcile every number across abstract, text, tables and figures. Confirm every reference supports the sentence citing it. Complete the reporting checklist. Read the discussion once more asking only whether the design supports each claim.
In short
What to take from this.
- Write figures first, abstract last.
- Reconcile every number across abstract, text and tables.
- Name the specific limitation and its likely direction of effect.
Sources
- ICMJE Recommendations, on manuscript preparation
- EQUATOR Network, design-specific reporting checklists
Related