A Guide to Writing an Effective Results Section for Your Research Paper
You have done the hard work of outlining your research study’s problem, conducted a literature review, clearly defined your objectives and research questions, and laid out a transparent process of what was done to answer your research objective(s) and question(s). Now, it is finally time to share with the reader what you found. However, in doing so, a number of questions arise, such as:
- In what order should I describe the results?
- How much detail should I provide?
- What information should go in a table/figure versus in-text?
In this article, I address these questions one by one to help you write a results section that is structured, transparent, and well-detailed.
What is a Results Section?
A results section is the third component of a research paper that follows the introduction and methodology sections. This third component provides researchers with the opportunity to objectively report their findings. The key term here is objectively or as the Merriam-Webster definition states, “with a basis in observable facts rather than feelings or opinions”. Although philosophers have suggested that we can never separate our point of view from our subjective experiences (Morales, Bax, & Firestone, 2020), the aim of the results section is to provide a brief account of what you found from the data collected without making speculations about the meaning of your results or their relationship to prior literature (this latter aim is meant for the discussion section). I often see novice writers violate this latter stipulation by citing literature in their results section. In general, there should be no citations included in this section because your statistical findings serve as the basis for your claims around the findings.
How Should a Results Section be Organized?
Depending on the research objective(s) and methodological approach(es) taken, you could have an extensive set of analytical information to share with the reader. So, how should you organize this information? The great news is that you have already done the hard work of structuring your results section by outlining in a logical fashion your research objective(s) and question(s) in the introduction section. In some cases, you may need to provide descriptive or ancillary information (e.g., sample size and demographics), but for the most part, the results section should be organized into a separate subsection for each of your research questions. I have found this structure to be intuitive and it has helped me to remember that the point of this manuscript section is to present my observations in relation to each of my research questions. With that noted, the overall structure should include a brief orienting paragraph that describes the layout of your results section, the inclusion of ancillary analyses, and conclude with the main findings for each research question. This structure can be thought of as a funnel, with the most general information provided first, while building momentum to the most important and best findings.
What Results Should Be Presented In-text or in Tables/Figures?
For many, deciding on what information to include in-text versus in tables/figures can be a difficult decision. I used to struggle with this problem until I realized that the tables/figures are the basis for telling an intriguing account of findings as they depict the most important results. Given this perspective, I often will create my tables/figures prior to writing as they will serve as an outline of what I want to discuss.
However, one may question whether the choice of using a table versus a figure is trivial. I can say that it’s not. In general, figures provide your audience with an easy way to detect patterns quickly via an illustration as opposed to extensive data values included in a table. As the old adage says, “a picture is worth a thousand words”. Although figures are generally more effective, tables may be preferable when patterns are not important, such as when presenting a large number of correlations and when your audience needs to know exact statistical values (e.g., the presentation of beta-coefficients from a regression analysis).
Given that you will often create figures to illustrate your findings, let’s discuss a few best practice guidelines for creating graphical representations of your data. The first is to ensure that you are using the appropriate graph type based on your data. As an example, it is common for researchers to display continuous data using graphical representations that present descriptive results (e.g., histograms with error bars). Such representations ignore distributional findings. Where possible, it is recommended that individual data should be presented for continuous variables by using plots such as dot plots for small samples and ridge plots for larger samples. The latter graph type is particularly useful in determining whether there are distributional assumption violations (see figure below). Second, researchers should graphically depict estimates of variability. If variability within the sample is of interest, the standard deviation should be reported. However, if the interest lies in uncertainty of the sample mean to that of population, then the standard error of measurement is a better statistic to report. Regardless, providing depictions that present transparency of your findings is critical, and where possible, this can be best accomplished by displaying the raw data.
Upon creating your tables/figures, the next step is to describe the story observed in your findings for each research question investigated. In doing so, use your outcomes to support your account of what you found. I would recommend drafting your results by writing the text for each table/figure individually. You can then piece together your findings into a coherent recounting.
What Level of Detail Should be Provided in the Results Section?
As in any good story, the level of detail has to be just right. Provide too little information and the reader may feel that you did not address your research question. Provide too much information and the reader may lose interest as your presentation has become overly convoluted with detail. So, how do you address this issue? I would recommend telling a story with a minimal amount of statistical detail for each primary outcome investigated. Getting a feel for the right-level of detail will come with experience, however, to develop this intuition, I would recommend two strategies.
The first is to read research in your field, especially in the journal outlets where you would like to publish. Identify researchers whose work you enjoy reading because of their writing strategy. Observe how they describe their results, with a particular focus placed on level of detail and their utilization of tables/figures (I will discuss this latter strategy in a following section). These researchers have done the long, hard work of crafting an effective writing style. Why not use their styles as a model for your own writing?
A second strategy is to use trusted advisors/colleagues as readers. Provide them with a manuscript draft and ask them whether your results are sufficiently described based as questions such as:
- “Do my results follow my methodological description of how I planned to address my research questions or is new information provided based on analyses not described in the method section?” (there may be a need for post-hoc analyses, but where this is the case, a strong rationale should be provided)
- “Are the results organized in a manner that directly addresses each one of my research questions?”
- “Where can I better explain the evidence needed to answer the question at hand?”
- “Is it clear how the analytic information given connects to the research question being addressed?”
- “Are you able to get a clear sense that I answered my research questions based on the summary of findings given?”
- “Are there instances where the information provided reflects a subjective interpretation of results?”
These are example questions and are not meant to be exhaustive in nature, however, they may help to guide your conversation with an internal reviewer of your work.
On a related note is the discussion of how to present non-significant results. Given that published research largely reflects findings that are novel or possess a large effect size (Prager et al., 2019), there may be a tendency whether intentionally or unintentionally for researchers to avoid including nonsignificant and/or negligible findings. This tendency has been referred to as “p-hacking” or “selective reporting” (i.e., presenting only significant results) and it has a number of consequences. First, as Rosenthal (1979) noted, we may not know how many studies have been conducted and never reported (this is referred to as the “file-drawer problem”). If this is the case, there is a possibility that researchers continue to replicate the same nonsignificant findings unknowingly, leading to wasted time and resources. Second, if published research only reflects positive (i.e., good, affirmative, or constructive) results, our knowledge may be skewed. Consequently, we have an inflated sense of treatment effects, as an example, which could have serious implications for practice (imagine medical treatments).
How can we combat “selective reporting”? From a pragmatic perspective, it is my opinion that we can only make improvements to this issue if researchers are incentivized to do so by reviewers and journal editors. For instance, journals could clearly outline guidelines for results reporting that include the necessity to include all p-values up to three decimal places (if testing for statistical significance) and effect sizes regardless of magnitude or direction. An additional encouragement could come from requiring the publication of raw data to open repositories, where possible. Doing so could increase researcher accountability as there is the potential for others to identify flawed results reporting. Even without journals requiring these practices, they are good habits to adopt as researchers and are beneficial to improving our knowledge base and that is the point of research (see Head et al., 2015). So, embrace the fact that a non-significant finding can be a significant result.
Summary
The results section of your manuscript provides you with an opportunity to tell the audience what you have learned. As always, I recommend that you take a storytelling approach to writing this section. Begin by creating your tables/figures, which will highlight the most important findings that you would like to share with readers. Upon doing so, describe objectively what is highlighted in your tables/figures. Avoid making speculations about the meaning of your results or their relationship to prior literature. Furthermore, I would recommend steering clear of overly describing statistical outcomes. This information should be used sparingly as a means to support your observations. Next, put the results together. In terms of organization, I would recommend starting out with an orienting paragraph outlining the flow and structure of this section to readers, followed by ancillary analytic information, and the main findings for each research question under investigation. Feel free to include your detailed findings or access to your raw data as a supplemental file. Most journals are happy to include such information. Overall, the results section provides you with an opportunity to share with your audience potentially new and exciting findings. Take your time in crafting this section, because it will make or break your paper.
References
Head, M. L., Holman, L., Lanfear, R., Kahn, A. T., & Jennions, M. D. (2015). The extent and consequences of p-hacking in science. PLoS biology, 13(3), e1002106. https://doi.org/10.1371/journal.pbio.1002106
Morales, J., Bax, A., & Firestone, C. (2020). Sustained representation of perspectival shape. Proceedings of the National Academy of Sciences, 117(26), 14873–14882.
Rosenthal, R. (1979). The file drawer problem and tolerance for null results. Psychological Bulletin, 86(3), 638–641. https://doi.org/10.1037/0033-2909.86.3.638
Let’s stay connected. You can get more of my updates on professional development for researchers as well as contact me via Linkedin. Learn more about my current work at https://www.josephriosphd.com/
