A smooth and effective quality control process involves establishing guidelines at the beginning of the research and writing process, and then ensuring accuracy and consistency from start to finish. Here are eight steps towards good quality control in scientific writing.
1. Track your data
We’ve all heard time and time again to make sure your research is replicable. That’s true for both research methods and your results and review process. Data collected and decisions made along the way need to be tracked and maintained by someone who has eyes on the whole project. Typically, this is a project manager or a QC specialist. When reviewing sources for inclusion in a literature review, for example, every source reviewed should be logged. If it was excluded, note the reason. Whenever possible, assign ownership to collected data so that a senior author or QC specialist can contact the person later if any questions arise. Logged data is also a great resource to have if you are preparing for publication in peer-reviewed literature, so you can quickly and accurately respond to editor or reader inquiries about your research process.
2. Set the stage with writing conventions
Before you begin writing, create a writing conventions document. It is essential that all authors and editors be aware of these guidelines while they draft and review a document. Examples include formatting (number of spaces between sentences, page margins, capitalization and punctuation in titles and subtitles), abbreviations (versus/v/vs, U.S./US, e.g./ex.), terminology (gender vs. sex, white vs. Caucasian, youth vs. adolescent). When in doubt, lean on published articles or style guides in your target journal, or reports already published in the subject you are investigating. Consistency is key, especially when reporting study results or extracting data.
3. Determine your citation strategy
Choose a citation style for the bibliography, and don’t forget to select and enforce internal citation formatting. If using a citation engine, ensure that the internal citation format can be identified by the engine for easy formatting at the end. An inconsistent internal citation, even something as small as “{Jones, 2017}” vs. “(Jones 2017)” can mean hours of unexpected work for the person who revises the document at the end of the project.
4. Draft and pilot test your tables and figures
Before you present 40 rows of results or extract data from a multitude of published studies or clinical trials, draft the shells for your tables and figures, and take a representative sample of data points to populate them. Tables and figures are often neglected, but the eyes of many readers go straight to the imagery. Make sure it’s easily digestible and contains valuable information. We recommend having a methods expert and a subject-matter-expert review the table and figure shells to ensure they appropriately reflect the essential information in a written piece. Another perk: getting this section neat and tidy makes for smooth sailing when writing up your results.
5. Set up safeguards to catch mistakes when they happen
Many large reports or manuscripts have different authors for different sections. First and foremost, each section author must be responsible for reporting data accurately. Not only should they double-check data to ensure it was reported correctly, but they should also double-check references to other sources to ensure they have accurately interpreted the source information. When large blocks of data are abstracted by research associates or junior authors, a senior reviewer should review a sample to ensure accurate, consistent abstraction and interpretation.
6. Do an internal cross-check
An experienced QC specialist should review the document to make sure there are no inconsistencies. We call this an “internal” cross-check because the document is being checked against itself. This means ensuring that each number in the results is identical to its corresponding value in the data tables, that conclusion statements are truly reflective of the results, and that no overtired authors made an embarrassing error interpreting or reporting data. The internal cross-check is limited to reading the report itself, however, if an inconsistency is noted, the source content should be identified and the issue resolved. This may involve speaking to the section author, re-reading a source publication, or reviewing the raw data (which, if you followed #1, should be easy to access).
7. Author review
Every author should be offered the chance for a final review of the document. Sometimes looking at your writing or a co-author’s with a fresh set of eyes makes all the difference. Nominate someone to be the keeper of the document while it is being passed around to maintain version control.
8. Put a bow on it
Someone with experience in QC should conduct a final review of the document, be it five or 500 pages, to ensure the correct writing conventions were applied and there are no spelling or grammatical errors. Only after all inconsistencies are addressed and the authors have consented to report finalization can the QC director begin final formatting and reference management. If submitting to a publication, the document formatting must be in compliance with the publication’s guidelines.
Instituting a QC process before writing begins is the best way to avoid unexpected (and costly) delays. QC isn’t glamorous, but it is an essential part of the writing process. Someone with an understanding of the subject matter and experience in QC is key in creating a high-quality, scientifically relevant piece.