Twenty common serious mistakes when writing a paper
These mistakes are very commonly made when authors write their scientific paper. They increase the risk of rejection, especially when many of these errors are present. The following table will list these mistakes, explain why they are problematic, and how they can be avoided.
In addition to the hints given here, I should point out that many of these mistakes can be avoided by checking the relevant study reporting guideline for the reporting structure and elements that your paper will need. I have also created a simplified guideline for reporting human observational studies that includes relevant STROBE and CONSORT guidelines.
Part of paper | The mistake | Comment | How to avoid this mistake |
| ABSTRACT | 1. Abstract lacks one or more key elements
| The Abstract is the most read part of your paper. It is essential that it contains ALL of the key elements and messages of your paper. | Check that ALL elements required are included. CONSORT for Abstracts gives good guidelines for not just randomized controlled trials but also observational human studies in general |
| INTRODUCTION, DISCUSSION, REVIEWS | 2. Studies cited in these literature-based papers/paper parts are not fully described (a) Type of study is unclear (e.g. RCT, cross-sectional, case report, in vitro, animal study, in silico study) | This makes it difficult for the reader to determine the quality of the evidence for the claim being made. Is it reliable (RCT, large prospective cohort study) or weak (case report, in vitro study?) | Check that the study type and its important elements are fully described |
(b) Type of subject unclear:
| This makes it difficult to see how comparable similar studies are | Check that the study type and its important elements are fully described | |
| 3. References are incorrect | This mistake is unnecessary and it gives the impression of sloppiness. That could make the reviewer distrust everything about the paper. | Check that all references are appropriate and correctly cited | |
| 4. References are not the ORIGINAL source of the data supporting the claim/statement in your paper | This can lead to unreflected dogmas in the field that have not actually been tested experimentally. These unchallenged dogmas can be quite destructive to the progression of science | Try as much as possible to make sure that the cited paper provides original data supporting the statement | |
| 5. Accidental plagiarism | This is because of copy and paste and then not reformulating. Copy-paste is a common technique for extracting information from the literature but if not rewritten, it is often clearly detectable to readers and may trigger journal plagiarism software | Put the copied-pasted text in a color and then, when revising the text, make sure that only one in three/four words remains that color | |
| METHODS | 6. No/poor ethics section | Having a well-written and detailed ethics section gives the reader confidence that you understand the importance of ethics in biomedical science | All studies: "This study was approved by [the appropriate] ethics committee/institutional review board" Human studies: "This study adhered to the tenets of the Declaration of Helsinki and its revision"; "This study adhered to Good Clinical Practice guidelines"; "All patients provided written/oral informed consent to have their data included/ to participate in the study" Animal studies: "This study was conducted according to international [e.g. Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC International)], national, and/or institutional guidelines for humane animal treatment and complies with [relevant] legislation" |
| 7. Human study design not stated or incorrect (e.g. Cohort when it is cross-sectional) | Disclosing the study type helps the reader to quickly estimate the likely quality of your study findings. Not adding this information or using the wrong label may make the reviewer think that your understanding of study design could be limited | If unsure, consult the Equator Network to determine what type of study your study is | |
| 8. Not stated when the study in humans was conducted (e.g. January 2014-March 2015) | Was the study performed last year or two decades ago? This is particularly important for clinical studies because clinical practice can change very markedly over time | Make sure that you indicate when the study was performed in the Abstract and Methods | |
| 9. Experimental timelines unclear | e.g. when after the intervention did treatment start and when were samples taken? How long was follow-up? These experimental details should be described very clearly so that the reader knows how the experiment/study proceeded | Clearly describe these experimental details in the Abstract and Methods. Consider using a schematic depiction of the experimental timeline if multiple intervention timepoints and/or sampling timepoints are used | |
| 10. Detailing patient numbers in the Methods section e.g. “1022 patients were enrolled and 19 were excluded” | The Method section in human studies should only describe HOW the patients were selected | Only describe patient numbers in the Results section. Animal/well/cell numbers can be indicated in the Methods if the same number are used for all experiments. If varying animal/well/cell numbers are used, indicate this in the figure legends | |
| 11. Primary and secondary outcomes are unclear | The primary outcome MUST be clearly indicated because the power calculation determining the optimal study sample size should be based on being able to detect a a significant different regarding the primary outcome. This is often a confusing area in the Methods | Make sure to clearly distinguish between primary and secondary outcome measures in the Abstract and Methods | |
| 12. No power calculation | Does the study have enough power to detect a difference in primary outcome? Otherwise, the study is essentially useless | Consult a statistician to determine the optimal sample size. If necessary, a post-hoc analysis can be performed (but a priori power size calculations are better) | |
| 13. Statistics section incomplete/unused stats methods mentioned | This makes the reviewer think you do not understand your statistics | Make sure the Statistics section clearly explains ALL statistics you used and which analyses were performed | |
| METHODS & RESULTS | 14. Use of uninformative experimental group names | e.g. Group 1, 2, and 3. The reader has to work hard to remember which is the control, which is intervention A, which is intervention B etc | Generally, group names are NOT needed. If absolutely necessary, use short, memorable, and completely distinguishable groups names |
| 15. Use of complicated experimental group names | e.g. a study in rats where the sciatic nerve is denervated surgically and then half of the rats start exercise training (ET) at 2 or 6 weeks. Groups are called Den2w, Den6w, Den2wET, Den6wET. It can be very confusing for the reader | Generally, group names are NOT needed. If absolutely necessary, use short, memorable, and completely distinguishable groups names | |
| RESULTS | 16. Describing experiments that have not been mentioned in the Methods (or vice versa) | Many readers will quickly skim through the Methods before moving onto the Results. If an experimental method is suddenly mentioned in the Results but the method was not detailed in the Methods, this can confuse the reader | Make sure that all methods used to get the results are actually described in the Methods (and vice versa) |
| 17. Describing data that do not relate directly to the study objective. Often people do this because these data are not sufficient to write a whole paper about | e.g. cohort study examining whether 800 patients with rheumatoid arthritis on MTX mount good antibody responses to flu vaccine A. A subgroup of 20 patients is examined for T-cell responses to flu vaccine B antigen. The subgroup analysis will distract the reader from your main findings, disrupt the smooth flow of concepts, and cause confusion | Make sure that ALL data relate directly to the study question | |
| 18. Unclear how many patients/animals/wells per experiment, and how many times the experiment was performed | This is Science101 basic information that should always be included because it shows how reliable the findings are | Always indicate how many patients/animals/cells/wells were used for each experiment in the Results (human studies) or figure legends (other studies) | |
| 19. Inconsistencies between the Results section (and/or Abstract) and the data shown in the tables and figures | e.g. Results: “Of the 93 patients who received the study drug, seven (7.5%) had developed new-onset hypertension at 3 years.” Table 1:
| Always check that all data cited in the Abstract and Results match completely with the data in the figures and tables | |
| DISCUSSION | 20. No Study Limitations section | No study is perfect - all have flaws e.g.
| It is essential that you are completely open about the limitations of your study. It makes you look very trustworthy and shows that you are really interested in answering the scientific question |
