Marc D. Baldwin is the founder & CEO of Edit911 Editing Service. He is also Professor of English at Hillsborough Community College and a published author.
You can find more of his writing and editing advice on the Edit911 blog.
One of the most important things you will do as a scientist or researcher is publish your work. It isn’t just a matter of sharing information—an integral part of the scientific process—it’s also about furthering your career.
Publishing your work in a scientific journal is a requirement toward earning a graduate degree at some institutions. Beyond graduation, getting published is necessary for a career in academia and, increasingly, in industry as well.
I have proofread and reviewed hundreds of original manuscripts in my career as a research scientist and lecturer. I’ve noticed over the years that most mistakes can be placed into a few simple categories. In this article, I will discuss the Top 3 writing errors I encounter when reviewing manuscripts submitted for publication to scientific journals.
1) Failure to provide sufficient context for the reader
The Introduction section of a manuscript sets the stage for what follows. This section should provide a brief history of the subject (save the comprehensive histories for Literature Reviews and other such articles), with most of the attention focused on current research. The latter should be presented in such a way as to clearly identify the gaps in knowledge or conflicting results that your experiment is designed to address. As my post-doctoral supervisor used to say, “State the Problem”.
The Problem could be a specific hypothesis with conflicting study results, a gap in knowledge that has not yet been addressed (often because the appropriate methodology has not been developed), or there could be competing hypotheses with not enough evidence to rule out one or the other.
If you are attempting to address a specific hypothesis, be sure to state it clearly in the Introduction (and sprinkle it throughout the rest of the paper for emphasis). Ideally, you knew the Problem before you designed your experiment, but even if you didn’t you need to figure it out now. This presents your work as relevant, important, and interesting to the reader.
Once your reader clearly understands the nature of the Problem, it is time to sweep in like a superhero and explain how your experiment will attempt to solve the Problem. Keep this brief. Either you’ve invented a new method that can shed light on previously unexamined topics, or you’ve designed your study in such a way as to provide more robust data. You want to present your work as the solution to the Problem; that is what makes it compelling to the audience. Not only will this generate excitement and interest around your paper, but such information becomes imperative when it’s time to apply for grants. Finally, having a clear understanding of the Problem and how you are attempting to solve it provides the reader with a context that makes it easier to follow the rest of your paper.
2) Poor presentation of data
Many scientific studies require that data be manipulated in order to be useful, whether it’s as simple as calculating means and standard deviations, normalizing data via transforms or other mathematical formulae, or more complex procedures such as regression analyses or curve-fitting. Where at all possible, you should provide access to your raw data so that readers can determine whether your calculations, data transforms, and other manipulations were appropriate and accurate. If there is a great deal of raw data, it’s okay to provide it in an Appendix or as an online supplement (most journals provide this option). You can also present your raw data in a table along with the processed data, but be careful that you don’t end up drowning the key points in a sea of numbers. As a general rule, the less raw data you have, the more accessible it should be to the reader. Only those deeply interested in your results (usually others working in the same field whose own research is complementary to yours) will feel the need to wade through reams of raw data, so don’t let it become a distraction for everybody else. Transparency is an important part of the scientific process, but it needn’t compromise readability.
When describing your results in the text, stick to the major points and refer to tables and graphs where appropriate. Do your readers a favor and don’t use the text to walk them through every line and column in a table. Instead, describe the trend or point out one or two important data points, then use the Table as a means of providing additional information.
Poor presentation of results often reflects a lack of understanding of the purpose of tables and graphs. Their purpose is to visually present data that is too complex to describe in text (or too tedious to read). Whether to use a table or graph depends largely on the message you are trying to convey.
If the quantitative aspects of your results are of primary importance, a table allows the reader to easily locate the exact value of a measurement without having to wade through descriptive text. Tables can also help the reader to visualize simple relationships among treatment groups or variables, such as an increasing or decreasing trend, or readily identify those results that are statistically significant. However, where the relationship between numbers is a significant part of the story, a graph is the better choice.
While it is possible to determine the value of a specific point on a graph, it usually requires visually extrapolating to the axis and then determining the value based on the scale and range of values along that axis. Thus, graphs are not the best choice when the specific value of your measurements is an important part of the story (if only one or two values are important you can point them out in the text). Where graphs excel is in visually representing relationships among values. Bar graphs, for example, are excellent for displaying significant differences in results among treatment groups (if the bars look too similar in height, a table indicating statistical significance might be a better choice). Line graphs are perfect for demonstrating the nature of the relationship between dependent and independent variables. Whether it is a linear relationship that implies correlation or a more complex mathematical relationship that points to the underlying biological mechanism, graphs are an excellent choice when the nature of the relationship is an important part of the story you are trying to tell.
Finally, make sure your Results section describes your data and only your data. This is not the place for discussing the implications of your results, making assumptions, or drawing conclusions. Such information belongs in the Discussion section.
3) Failure to discuss the downsides
Some authors mistakenly believe that if they point out the weaknesses of their study it will reflect poorly on their work. Generally speaking, the opposite is true. Nobody expects perfect results that clearly prove or disprove a hypothesis; science rarely works that way. Hypotheses gain acceptance as the weight of evidence builds and this process requires that studies be critically examined. Pointing out the limitations of your study simply makes this process a bit easier for your colleagues. Such transparency is expected and a means of limiting bias as much as possible.
The perfect experiment is often compromised by the realities of life in the lab. You may not have access to the ideal solvent or piece of analytical equipment. Your experimental drug may be exorbitantly expensive, limiting the size of your study population. You may not have the time or resources to perform follow-up measurements for the optimal length of time. Stating the limitations of your experimental design is simply an admission of the realities faced by all of your colleagues and should not reflect poorly on your abilities as a researcher.
When discussing the meaning of your results in the context of current research, don’t just cite those studies whose results support your own. Cite those authors whose studies yielded different results and suggest why that might be so. You aren’t expected to provide a definitive answer, but you are expected to at least come up with some educated guesses. Perhaps your experimental design differed from theirs; perhaps you used different controls or a different sample population. It doesn’t make you wrong, but it does make you thorough and is the sign of a professional.
Unfortunately, the skills that make a good scientist do not always translate to good writing. Clear, concise writing and effective communication are essential when attempting to convey the importance and significance of your work to the broader scientific community. It doesn’t matter how brilliant your experiment or how groundbreaking your results, if your colleagues cannot understand what you’ve done and how it contributes to the field, you risk having your paper rejected for publication. The old adage of “publish or perish” is truer than ever in today’s world of research. Ensure your work stands the best chance of acceptance.
what about qualitative studies in social sciences?
I think that similar principles apply.
I’d contend that the 3 points still apply: give context and evidence, as well as acknowledging counter-arguments.
Reblogged this on Faculty Research Group and commented:
Excellent insight for our own researchers here at the FRG.
Thanks so much, FRG! I’m glad you found the article helpful. Feel free to contact me if you wish. Keep up your excellent work!
I have yet to read a paper with an attachment or appendix (online or otherwise) with the “raw” data. Since all results MUST be positive these days in order to get published its hardly a wonder.
Yes, so true. It’s an option, of course, and something of a dilemma, as I note: ” As a general rule, the less raw data you have, the more accessible it should be to the reader. Only those deeply interested in your results (usually others working in the same field whose own research is complementary to yours) will feel the need to wade through reams of raw data, so don’t let it become a distraction for everybody else.”
[…] Some mistakes to avoid when writing for publication; […]
I often find it hard to follow the advice “make sure your Results section describes your data and only your data.” I agree in principle, but sometimes it doesn’t hurt to include a few references, such as “these results are similar to those found by x, y and z”.
Thanks for the great post, very helpful.
Marc is having some technical difficulties, so he asked me to add this:
You may be right. Certainly, there is no concrete, one-size-fits-all formula. If you deem it necessary and helpful to “include a few references,” then that’s your choice and the journal editor’s decision whether or not to excise them or ask for some slight revisions.
I’m glad you like my article!
Reblogged this on Learning Analytics and You! and commented:
This post will be a constant reminder to me. Therefore, it aught to be reblogged. Thank you, for this resource full article.
[…] how they introduce themselves on their blog). Learn here how to chair a conference session or learn here about 3 common writing mistakes. Twitter: @researchwhisper 8. PhD talk – Random thoughts on the […]
Uhh, other disciplines may be more sure of themselves, but in Psychology, we never talk about a hypothesis being “proved”. This is because we test null hypotheses, which can only be accepted or rejected. Even if the null hypothesis is rejected, this does not provide “proof” that the alternative hypothesis is correct, only that the null hypothesis, in this instance, cannot be supported. (I refer toyour comment in section 3), line 2.)
Reblogged this on Spain90's Blog.