Improving scientific research papers: Aston University researcher outlines top ten pitfalls and how to avoid them
Aston University
image:
Dr Dan Green
view moreCredit: Aston University
- Dr Dan Green is the lead author on an article outlining the top ten common statistical errors he sees when reviewing scientific papers
- These include implying causation from association, poorly formatted abstracts and inadequate statistical analysis
- The paper is aimed at all applied researchers and scientists doing quantitative work to help avoid journal rejection.
Aston University biostatistician Dr Dan Green has identified the top ten errors he commonly sees in research papers in his work as a statistical reviewer for scientific and medical journals.
Since 2021, he has reviewed more than 200 papers for journals such as BMJ Heart and Addiction, and has now written an article for BMJ Heart explaining the errors to help researchers avoid them. He hopes the article will save time for authors with revisions, avoid rejections, and save time for reviewers in repeatedly reporting the same mistakes and advance “a research world that is transparent, specific, and reproducible.”
Dr Green worked with Dr Rebecca Whittle at the University of Birmingham and Dr Diane Smith at Fuze Research to write the article.
Top of the list is the incorrect use of causal language, for example: “we have shown that x leads to y”, in a study design where this cannot be definitively determined, for example, believing that the ice in a soft drink is the cause of weight gain, or higher ice creams sales at the beach lead to more shark attacks. This leads to misleading abstracts and consequently misleading take-home messages, and an untrained reader, such as a journalist, can read too much into the findings. Dr Green and his coauthors recommend that authors review their conclusions carefully and ask a colleague not involved in the research for feedback.
The problem of poorly formatted abstracts also makes the list. Abstracts are the most-read part of any paper and when incorrectly written can lead to misinterpretation and undermine the impact of the work. Dr Green has found incorrect subheadings, findings in the methods section and unquantified measures in the results section. Journals all have clear formats to follow for abstracts, and failing to do so can raise doubts about the authors’ ability. Dr Green recommends that researchers check other examples of abstracts in the journal, and ensure that the ‘five Ws’ – what, who, where, when, why – are covered concisely.
Another common error that Dr Green sees is putting results in the methods section, which disrupts the clear structure of scientific reports. Methods should only include methods, such as sample collection, research centres and clinics involved, and inclusion and exclusion criteria. Dr Green says that the methods section should be written as if the study is yet to be carried out.
The statistical analysis section of the report is crucial in any paper, to identify patterns and trends and show whether results are real or likely due to chance, using hard maths instead of intuition; there should be sufficient detail included that another researcher in the same field could copy the analysis approach on the same data, and crucially get the same answers. Dr Green unfortunately commonly sees inaccurate or incomplete statistical analysis sections. His main pointer for success here is writing a bullet point list of everything described in the methods, and writing a bullet point list of everything reported in the results. The two lists should be identical and in the same order. He recommends using supplementary text to ensure that the section in the main paper is not too long.
Other pitfalls covered in the article include poorly detailed and badly formatted tables, poor reporting of missing data, the use of the wrong type of data analysis (using univariable significance for multivariable models) and absent, or insufficient flow diagrams to explain the design of the study and who of the initial participants in the study was included in the end.
The article also offers five ‘bonus pointers’ on other statistical and chart errors.
Dr Green said:
“Researchers are human, and while some oversights can occur naturally in the rush of wanting to get your paper submitted, obvious errors in the structure and presentation of your article don’t give the best first impression to an editor or reviewer. Take a little more time to check the details of your submission before clicking submit, check those ‘Instructions for Authors’ for the journal again, and if still unsure, get another opinion.
“We have produced this article with future submissions in mind, so you can quickly whizz through the items, and question whether you have made any typical, but ultimately crucial errors. A little more care now saves a lot more time later on and avoids those annoying re-review edits, or even searching for a new journal to submit to!”
Read the full paper in BMJ Heart at https://heart.bmj.com/content/heartjnl/early/2025/06/23/heartjnl-2025-325939.full.pdf
Journal
Heart
Method of Research
Observational study
Subject of Research
People
Article Title
Top 10 statistical pitfalls: a reviewer’s guide to avoiding common errors
Article Publication Date
23-Jun-2025
