PLOS
PLOS

In many fields, a statistical analysis forms the heart of both the methods and results sections of a manuscript. A well-planned, clearly described and faithfully executed statistical analysis can improve the chances of acceptance and ensure the long term reproducibility of your study.

Facts are stubborn things, but statistics are pliable. - Mark Twain

Strategies for ensuring a strong and reliable statistical approach

While the specific approach may vary widely across disciplines, all statistical methods employed in research should be:

Study Design

Appropriate for the study design

How to do it? Plan your approach in advance. Define your analytical methodology before your study begins, including line of inquiry, what you plan to do, what data you will collect, and how you will analyze it. Adhere to your original plan throughout the execution of your study.

Rigorously reported, with enough detail for others to reproduce the results

How to do it? Imagine yourself replicating the study years in the future. What information would you need to reproduce your own work? Don’t forget to capture any adjustments!
Reporting
Accountability

Free of manipulation, selective reporting, or other forms of “spin”

How to do it? Hold yourself accountable. Don’t adjust hypotheses after the fact or search for patterns that can be presented as statistically significant after results are in, in order to make the outcomes appear more interesting. Using statistics to misrepresent results can prevent publication, undermine a published study, or lead to retraction.

TIP

Preregister your study design with a journal
Preregistering is a great way to head off any intentional or unintentional flexibility in analysis. By declaring your analytical approach in advance and undergoing peer review you’ll strengthen your study, increase the credibility of your results, and minimize publication bias.

What’s next? Stay tuned for our next issue, all about writing your conclusion.

Like the Writing Toolbox?
Share it with your friends.