So you’ve finally got all of your data collected and analysed, and now you need to write it all down. But the thought of writing your results section is driving you crazy! How do you structure it? Which bits of data should you focus on? There’s a lot to think about.
The results section is a vital part of your paper. Writing it in a structure, like the one I’m about to show you, can simplify things for both you and your reader. Not only will it mean that your data is much easier to understand, but it will actually save you time while writing.
A lot of my clients say that finding the time to write their paper is a huge struggle. That’s why I decided to share this technique, because it’s my mission to help as many researchers as possible to finish their research paper.
Before we get into the detail of your results section, it’s worth thinking about the overall picture as well.
You’re probably already familiar with the IMRAD format which is your Introduction, Methods, Results And Discussion. Essentially, this format lays out the following questions: why have you done this, how have you done it, what did you find, and what does it mean.
Before you start writing your results section, think about how the results will fit the overall story that you are trying to tell. If you’ve got a protocol available for your study, you might have already put some thought into this. But if you’re doing a secondary analysis of the data you might not have a protocol, so it’s really helpful to think about this before you start analysing data or before you start writing.
If you can find a way to group the different results up into something that flows logically and refer back to your overall story, those section headings within your Results should start to flow quite naturally.
Consider the following questions:
Why are you doing the analysis in the first place?
What message are you trying to tell your reader with this paper?
And how do your results fit into that?
You shouldn’t spend a long time on this, just roughly sketch those ideas on a bit of A4 paper. It can really help you to focus on what’s relevant to put into your results section and what isn’t.
It’s important to note that I’m not telling you to be selective with what results you show. Obviously, you need to be transparent and show everything, especially if you’ve got a protocol. But when it comes to secondary analysis, you might have lots and lots of results that aren’t necessarily relevant to the message you are trying to get across.
If this is the case, you might need to think carefully about how to present them. The results should be displayed in a series of tables and supplementary tables. But the job of the text is to help the reader to interpret those tables.
This is where you need to be a bit selective in your presentation. Which parts of the results do you need to focus on more in the text? This should be the ones for your primary outcome and other key findings, including any safety analyses.
Break it down into subheadings
When I’m writing a results section, I always use subheadings. I encourage all of my clients to do the same. Subheadings act as signposting which is really helpful for your reader to understand what you’re telling them.
You can also write them down before you put any content in the subsection. This helps you plan ahead before you start writing. Depending on how you are getting on time-wise, you might be able to just write one of those sections at a time. This works well if you’re focussing on task-based goals for your day.
The first section you have will always be about your participants. You might call this section ‘Participants’, or you might see it called ‘Population’. From there, you’d move onto Primary Analysis, to Secondary Analysis and then possibly a Sensitivity Analysis.
Let’s dig deeper into each of these subsections.
In this subsection, the first thing you should explain is how people got into that study. If you are doing, say, a clinical trial, you would usually make a CONSORT diagram to show this. There are other flowcharts available for each type of study that you might be conducting.
At the top-line level, how many people did you start with (how many people were approached to do that study)? How many of them said yes and were screened? Of these, how many of the screenings were failures, and how many went onto the next stage. Keep going until you get right down to the participants who were included in the analysis.
I would normally show all of that detail within a flowchart. In the text, I would refer to that flowchart and just pull out the key numbers according to your particular study. The most important one, though, is how many people did you end up with in your final analysis. That needs to be both in the flow chart and the text itself. This makes it really easy for your readers to find.
Next, focus on the descriptive characteristics, i.e. Who are the people in your study? Table 1 of my results section is nearly always the descriptive characteristics of the population at baseline. For example: demographic characteristics, any kind of baseline medication, medical history, anything that’s relevant to what it is your study is looking at.
The reason that we want to look at this detail is to give us some ideas about whether our study population is representative of the wider population that we want our results to apply to. This helps us figure out how far the results can be generalised. For example, if your study has only looked at men, then you probably can’t generalise the results to women. But unless you tell people that you’ve only got men in your study, people won’t know that.
You’re not talking about the actual generalisability or representativeness (that comes later on in the Discussion). What you are doing is setting up that information here, so your Discussion can refer back to it.
This is the really important one. It’s the reason you have done the study in the first place. So we want it to be front and centre for your readers. Make it really easy for people to see and really clear that it is your main analysis.
Your primary analysis usually focuses on your primary outcome. If you haven’t got a primary outcome because you are doing an unplanned secondary analysis of data, then I would just ask you to just really think about why you are doing that secondary analysis. Go back to your original research question and your hypothesis, and see whether there could be a primary outcome. Perhaps you just haven’t quite thought of it in that way and maybe that’s where you can start to tease it out.
Obviously if you’ve got a protocol, you should be making it quite clear within that protocol what your primary analysis is. In these instances, the primary analysis is self-explanatory.
People tend to remember a result that they see in a figure better than one that they see in a table. So consider whether you could put your primary result in a figure because then it will stick with your reader for longer.
Your primary analysis text doesn’t actually have to be very long, You just have to make sure that all of the details and information is included to answer your primary research question.
Your secondary analysis might be, for example, looking at your primary outcome in a different population. It might be where you’ve started to look at your secondary outcomes. Depending on what it is that you’ve done as your secondary analysis, it might make sense to split the results out.
Quite often, I will use the section heading ‘Secondary analyses of primary outcome’ and talk about the results around that. Then I would then have another subsection after that titled ‘Secondary outcomes’. If you’ve got a lot of secondary outcomes, you might want to split them down even further and group them.
Since these results are less important than primary analysis, they tend to go into a table, rather than a figure. That’s a rule of thumb though, not a hard and fast rule.
Sensitivity analysis is where we repeat our main analysis with a different assumption to see whether our results change because of this assumption. It shows us how sensitive our results are to the assumptions that we’ve made.
Missing data is an example of one of the main assumptions that we change in sensitivity analysis. Outliers is another example.
I present the sensitivity analysis in one of two ways.
1. If it’s a re-analysis of all of the outcomes, I tend to put it in a section at the end because it refers to all of my results.
2. If the sensitivity analysis is only a re-analysis of the primary outcome then I tend to put it in with the primary analysis section.
If you do pop this into an earlier section, make sure it’s really clear upfront which paragraphs are the sensitivity analysis. Don’t let this get muddled up with your primary analysis.
The Golden Rule: Keep it simple
And there you have – a perfectly structured results section!
My overall message is to keep it simple. The whole idea of this is to make it really clear for you, and for the reader, what you’ve found. Don’t make your readers go digging around to find this out. You want those results to be jumping out and screaming at them about all the good work you’ve done, and why it’s important.
You’re trying to tell your readers a story in your paper, and each section is another layer that breaks it down for them. Once you’ve explained the why and presented your data, the results section is your opportunity to draw out patterns from your research. It’s the first layer of interpretation as you begin to set up your discussion section.
The discussion section that follows is where you should be explaining what the results mean. What conclusions can we draw from the patterns you found in your results? If your results section is written correctly, this will feel like the next natural step in the story for your readers. It certainly shouldn’t seem like your conclusions are coming out of the blue.
Looking for extra support with your research?
If you would like some support with writing up your research then check out the options that I offer around this.