Results & Abstract:

Start writing your results and abstract first. Use your first draft as an opportunity to wrap your head around the findings. The results section should have a linear, sequential feel with each set of analyses building on the previous analyses. If the results section isn’t leading readers anywhere (e.g., random assortment of findings), it will be difficult during later steps to create a supporting narrative.

Many people struggle with not knowing which statistical results to include. Find quality papers with similar methodological approaches and participants (e.g., Journal of Psychopathology and Clinical Science, Clinical Psychological Science, and Biological Psychiatry: Cognitive Neuroscience and Neuroimaging); use those as inspiration. Ask yourself what you, as the reader, need to know to believe the results (e.g., expected group differences, manipulation checks, repeating similar analyses with different variables).

• Reporting Statistics in APA Style | Guidelines & Examples

• Number and Statistics Guide, APA Style 7th Edition

• Reporting Statistics APA Style

Results sections are necessarily technical. The burden is on you to "show" how the statistical results fit together. Strong results sections demonstrate how to make sense of the findings. Do the thinking for your readers with your writing. Remind readers why you are performing analyses with plain language statements throughout.

As you write, take an outside perspective. Poke holes in your own arguments. Anticipate questions reviewers might bring up as they digest what you are saying. Insert your follow-up analyses where you might expect these questions to naturally arise in readers’ minds.

Once the results section is good enough, try to capture the essence of your take-home message in the abstract. This is your elevator pitch. Communicate in a clear way and make your writing accessible to scientifically literate outsiders. In other words, write for someone who is generally familiar with your field, but may not have direct knowledge of the terminology and methodology you take for granted (e.g., psychologists writing for physicians/psychiatrists, nurses). You are selling readers on the idea that your paper and perspective will be worth their time.

Complex writing is not necessarily better. Early stage writers often struggle with communicating the technical aspects of the science accurately. This can lead to an overemphasis on scientific precision above all else, which can make sentences long, unwieldy, and bloated.  I still encounter this pain as a writer—all of the details and nuances feel incredibly important (at least at first).

Once you have a first draft, edit with an eye toward using straightforward language. Be flexible. Streamline and prune the ancillary details in your sentences (“Can I say the same thing in a simpler way?”). Although there are exceptions, aim for one idea per sentence.

Step 1) Write the results (customize the subsection headers to be specific to your study):

• • Participant characteristics.

  • Main analyses.

  • Follow-up analyses.

Step 2) Write the abstract:

• 2-3 sentences for background. What is the problem that needs solving with your data?

• 1-2 sentences about your plan to solve the problem using your methods/approach.

• 2-3 sentences about key findings.

• 1-2 sentences for broad implications (e.g., theory, clinical practice, public health, etc.).

Tables & Figures:

Next, create rough drafts of tables and figures to emphasize your key findings. At this stage, you are not obligated to make them perfect. Your tables and figures will likely evolve over time based on co-author feedback. So, don't get too attached to them.

Use tools such as Excel and PowerPoint to quickly draft your good enough tables and figures. Include at least one table for participant characteristics. Use Google Images to find examples of tables in APA format.

Step 3) Draft the tables:

• Use APA formatting and add comprehensive table captions.

Step 4) Draft the figures:

• • • Add comprehensive figure captions.

Step 5) Get co-author feedback:

• Assess if you are on the right track?

By now, your co-authors have signed off on your progress. You seem to be on to something; your results form a compelling story and your tables and figures add useful information. Consult with your co-authors about potential journals. Use tools such as the Journal/Author Name Estimator (JANE) to determine which journals are publishing similar articles. Review each journal’s ‘Guide for Authors’ document. Take note of the required sections, word count, and specifications for tables and figures.

Methods:

Methods sections need to be somewhat technical, but the language should be clear enough that someone with your data set could replicate the findings. Authors have some flexibility about which subsections to include. If you are reporting human subjects research, consider at least addressing your participant recruitment strategy, study measures, data preprocessing, and statistical approach.

Step 6) Write the methods:

1. Participants.

   • Study population, recruitment sources.

   • Inclusion/exclusion criteria.

   • IRB approval and compensation.

2. [Clinical] Assessment and/or [Experimental] Procedure (split into two subsections, if necessary).

   • Instruments, procedures used to characterize your participants (e.g., MMPI-3, CAPS).

   • Response invalidity (e.g., VRIN, TRIN), behavioral performance (e.g., d-prime < 5 SD from mean) exclusions.

   • Procedures for determining diagnoses, definitions of study groups for categorical analyses, etc.

3. Data Preprocessing (especially for EEG/fMRI studies).

   • Technical parameters of the data collection.

   • Preprocessing steps.

   • Postprocessing data quantification (e.g., mean amplitude ERP window of 200 - 500 ms).

4. Analysis.

   • Step-by-step plan for the statistics, presented in the same order as they appear in the results section.

Introduction:

The introduction can be the most difficult section to write. Draft the first and last paragraphs and then pivot to the middle paragraphs. If you have not already done so, do a literature review. Often, there are ~2-5 articles that most closely match your particular analyses. Read those articles carefully. However, you will not have time to read every article with that same level of intensity.

Plan on skimming the rest. Develop a note-taking approach that works for you. Consider writing down 2-3 take-home points for each article. After you have skimmed ~20-30 papers, reread your notes. Try to group the articles together by theme. Each theme can then become a candidate paragraph for your introduction.

One effective approach is to use a “funnel” format: start broad and get narrow. Take readers on a brief journey, which leads directly to your project. Each paragraph in the introduction should get readers closer and closer to your particular study. Patience is key. Avoid rushing ahead toward too many study-specific details before you have provided sufficient context.

It should be obvious to readers how your study flows from past studies. Leave “breadcrumbs” and “signposts” along the way, especially in the first paragraph. You do not have to give away all the details, but provide sufficient hints early on about the problems you intend to solve. Readers should not have to guess what you are up to—they will get bored if they don't know where you are taking them. No surprises. Repeatedly make it easy for readers to appreciate why your study exists.

As you write, put yourself in your readers’ shoes. All of the “obvious” reasons for your study will not be obvious to your readership. You have likely been mired in the details of your project for a long time. Spell out your logic and your motivations explicitly. Take inventory of the assumptions that led you to your study. Then, place them into your introduction within the funnel format.

As a general guideline, each paragraph should not exceed one page, double-spaced (i.e., about 300 words). If your paragraph is longer than one page, your paragraph may have too many separate ideas or be too verbose—consider splitting your paragraph apart or simplifying your sentences. Conversely, paragraphs 3-4 sentences or shorter are likely too brief—consider expanding your ideas or merging these ideas into other paragraphs.

While you write, use a citation manager. Export the citation files directly from Google Scholar into software like Zotero. Do not spend time manually formatting in-text citations or your references section—Zotero makes this unnecessary.

Step 7) Write the introduction:

1. First paragraph

   • First two-thirds focusing on general background about the problem you are trying to solve.

   • Last 1-2 sentences for a brief summary of how you plan on solving the problem.

2. Last paragraph

   • First two-thirds briefly outlining details of your particular study. Be clear how your study will provide an answer to the problem you previously discussed. Your work should fill some niche.

   • Last 1-2 sentences for discussing the potential broader impacts of this work. State clearly why people should read further. Be explicit about why solving this problem will be valuable.

3. Middle paragraphs

   • Review of background literature, but more importantly show readers where the current holes exist or the extent of the problem you are trying to solve. Create the “funnel” that connects the first and last paragraphs.

Discussion, Finalizing Tables & Figures:

The discussion section should start by summarizing your findings in plain language. Try to avoid overly statistical statements. Instead, explain the observed relationships between your variables as one might share them with a colleague.

Each of the middle paragraphs should stand alone as ~3-5 separate discussion points. Unlike the introduction, you are not obligated to have each of these middle paragraphs build off of the preceding paragraphs (think “bookshelf” rather than “funnel,” see below). Often, the discussion section is a great place to put overflow ideas that did not fit neatly into the introduction. However, make sure every idea links back clearly to your particular study and your research questions.

Here are some questions to spark ideas for middle paragraphs: How do these results fit with other similar studies? Do these results change (or confirm) how we should think about this phenomenon? What do the results reveal to us about broader theory (especially if the results are more consistent with one theory versus another)? What have we learned about the underlying processes or associated features of this phenomenon? What are the (clinical) implications? What can these results teach us about people in the real world? Do these results give us clues for solving similar problems in related domains? Could the study framework or approach be useful in other contexts? Based on these results, what should other researchers do next? What are the strengths of this study and why should readers trust your conclusions? 

Step 8) Write the discussion:

1. First paragraph.

   • 1 sentence for briefly restating the study purpose.

   • Summarize the results (with straightforward, non-statistical language).

   • 1-2 sentences for explicitly showing how you addressed the study problem.

2. Middle paragraphs.

   • Each paragraph discusses separate implications or conclusions from your findings.

3. Second-to-last paragraph.

   • Limitations.

4. Final paragraph.

   • Briefly summarize your overall conclusions with an emphasis on the impact of your findings, future directions, etc.

Step 9) Finalize your tables and figures:

• Use publication-quality formatting with high resolution (at least 500 dpi).

• I prefer Arial font, which is commonly accepted in biomedical journals.

Closing Thoughts on AI:

Many early stage scientists are using AI tools to accelerate this writing process. From my experience, AI has indeed improved grammar and readability among trainees. However, every draft I have seen generated primarily from AI has required massive rewriting. Relying on AI to draft your scientific manuscript will only get you so far.

AI does a much better job filling blank space with generic text than thinking for you. In the age of "publish or perish," I still believe manuscripts should be about more than just getting your work accepted at prestigious journals. As scientists, we write up our results to gain understanding and communicate that new understanding to others. Research is about discovery. Thinking deeply about your results while you write is an important part of that discovery. An AI manuscript may sound like a science paper, but it will not contribute new knowledge gains. There is no substitute for the old-fashioned approach.

That being said, writing is hard work. The process of creating quality scientific prose and original thought can be emotionally taxing. I understand the appeal of AI. Many people want to avoid the discomfort of wrestling with their ideas. It is helpful to remember that polished writing is rarely generated in one sitting. Writing multiple drafts is normal; reworking sentences for clarity is normal; uncertainty about how to move your ideas from your head to the page is normal. You are not a bad writer simply for experiencing discomfort during the writing process.

As a scientific writer, your marching orders are to 1) think deeply, and 2) communicate those thoughts effectively. Writing is the byproduct of your thinking—the process of bottling up your thoughts so others can share the same mental experiences as you. Therefore, clear writing flows from clear thinking. Your work as a scientist is not complete just because you have results. If you shortcut the writing process with AI, it will be unlikely you will have anything interesting to say.

Writing (and thinking) well is rewarding, but it takes patience and persistence. AI is much better at improving the readability of your writing after you've generated original ideas to share. Don't skip the earlier thinking steps.

Supplemental:

Example papers I'm proud of:

• Marquardt, C.A., Hitz, A.C., Hill, J.E. et al. Trait absorption predicts enhanced face emotion intensity discrimination among military recruits. Motivation & Emotion (2023). https://doi.org/10.1007/s11031-023-10014-5

• Marquardt, C. A., Chu, C., Hill, J. E., Venables, N. C., Kuzenski, L., Davenport, N. D., ... & Urošević, S. (2022). Evaluating resilience in response to COVID-19 pandemic stressors among veteran mental health outpatients. Journal of Psychopathology and Clinical Science. https://doi.org/10.1037/abn0000789

Common errors with scientific writing:

• Corneille et al., 2023 | Point of View: Beware ‘persuasive communication devices’

• Ecarnot et al., 2015 | Writing a scientific article: A step-by-step guide for beginners

• Wenzel et al., 2009 | A Step by Step Guide to Writing a Scientific Manuscript

General advice about writing clearly:

• IEEE Pro Comm | Writing Clearly and Concisely

• Butte College | How To Write Clearly: Using Precise and Concise Language - TIP Sheet 

• University of Arizona Writing Center | Writing Clearly & Concisely

• Stanford Engineering | Writing clearly and concisely means choosing your words deliberately

Tools for publication-quality figures:

• Daniel's XL Toolbox

• GIMP and/or Inkscape