How do I write the discussion section?

Some time ago, on the advice of my good friend and efficiency guru Jason Downs, I read The 80/20 principle: how to achieve more by doing less by Richard Koch.

To be fair, Jason did tell me, in the spirit of efficiency, that I didn’t really need to read the book. The main message, he said, was in the title – and he was right. Koch claims that 80% of value comes from 20% of the work effort. The trick to an efficient work life, Richard Koch contends, is to identify ‘high value’ work and just do that as much as possible. High value work for me is writing and talking to people; low value work is email. As with most self help books, I finished it and did nothing different.

Then, late last year, my friend Jonathon, one half of the fantastic Research Whisperer team, sent me a spreadsheet analysis of all their blog search terms. His analysis  showed lots of people were looking for how to make a simple Gantt Chart. Jonathan’s excellent analysis left me wondering: what do readers think is ‘high value work Thesis Whisperer work’ based on their search behaviour?

I immediately dropped everything to repeat the method on Thesis Whisperer, using 10 years of search data from well over 9 million visits. Here’s what I found out:

• Thesis Whisperer has great brand recognition: around 50% of people find their way here through typing variations of the name of the blog (far less people come here by typing in my actual name, Inger Mewburn).
• As I expected, the next most popular search type was writing problems, in various manifestations.
• Third most common search term was ‘How to look clever’, which is both funny and sad. I’m guessing this leads people to this old post here, which is a personal favourite

I drilled down a bit to try and find out: what exactly is troubling people about writing? I thought I would find concerns about productivity, feedback, literature reviews, style and voice, perhaps grammar, but it wasn’t: 75% of the thousands of writing related searches were questions and anxieties about the discussion section. I’ve only written about the discussion section twice in 10 years. That’s not 20% of the effort producing 80% of the reward – It’s more like 0.003% of the effort!

I’ve been teaching writing for over 15 years and reviewed lots of development programs at other universities. I see very few workshops that focus on the discussion section as a separate piece of writing. I guess we assume that supervisors are helping out, but my search data suggests maybe not. I get the anxiety, I really do. The discussion section of the thesis is the heart of the creative endeavour: it’s where you have to be MOST original. Even if you don’t have a section in your thesis called ‘discussion’ (I didn’t) there will still be places in your thesis where you must explicitly make new knowledge in relation to the data you have collected and your analysis.

I reviewed the two posts to see what else I might have to say.  In How do I start my discussion section I offer a description of what the discussion section should contain, how to decide if you need one or not and a grab bag of tactics to go about getting started – each one of which should probably have its own blog post. In The difficult discussion section I describe the section as the ‘problem child’ of the thesis. In that post, I try to walk people through a step by step process for making sense of the mess of writing, findings and analysis that you can end up with towards the end of your degree.

But what should actually go in the discussion section and how should you write it?

I’m indebted to my colleague Josta Heylingers for pointing me at the literature on functional linguistics. Josta uses this literature to teach people how to write discussion chapters at Auckland University of Technology. To do this, she uses the work of John Swales, and his method of ‘move step analysis’. I touched on this method in my own PhD, and of my PhD students is using move step analysis in her work, so I’ve had to become passingly familiar with the method.

Swales starts by assuming that texts are social things: every reader has been ‘trained’ on what to expect from different kinds of texts. Job applications ‘sound’ different from grant applications, which sound different than a journal article. So readers are familiar with the linguistic ‘moves’ to expect. These linguistic moves are sort of like dance steps that build together to make a socially recognisable text.

Swales’ move step analysis is a way of breaking down the text dance so you can understand which bits go where and how to put them together in an accomplished performance. Think of any dance craze you can name. When I was a teenager, in the 1980s (!) it was ‘The Nutbush‘; by the 1990s it was The Macarena. In case you weren’t there, or don’t remember, here’s a helpful video. It’s worth watching, because it’s delightful, and a good way to understand what move step analysis is:

In the video, Jean Eu broke down the Macerana series of steps with the hands, arms and hips, that are put together to form moves. The moves must appear in the right order or sequence to become a dance. If you start with your hands down to your hips instead of out in front of you, you ARE dancing, but it’s NOT the Macarena. People watching you, expecting to see the Macarena, will be confused. If you do the right moves, but not in time with the music, it’s still the Macarena, but anyone watching you will think that you are a bad dancer.

So it is with the Macarena as it is with writing the discussion section of your thesis: use the right steps, build them into a move and do the moves in the right sequence and you will write a ‘socially correct’ text. If you do a socially correct text, the knowledge you are putting forward for consideration can be easily assessed by the reader, because they are not distracted by the bad performance. The steps also give you a formula you can use to give your thoughts about your research findings a shape and form.

So, what does the discussion section dance look like? Let’s start with the big picture. In ‘The textural organisation of the discussion sections of accounting research articles’, Amnuai says:

“The discussion section is where authors place their ideas about their research findings and consolidate, generalize, and interpret their research outcomes for the benefit of those in their field or for other communities”

Each discipline is different. It’s important to base your moves on what is socially acceptable for your community,  but here is a simple list of basic moves that every discussion section needs to have:

• Restate Results (don’t repeat them!)
• Comment on the results
• Evaluate the Results
• Make suggestions based on the results

The helpful Manchester Academic Phrase Bank ‘discussing the findings’ section, gives you some sentences that you can use as the ‘steps’ for each move. Try some of these sentence starters to get you going:

Restate Results:
“The current study found that …”
“The results of this study show/indicate that …”
“The results of this study did not show that …/did not show any significant increase in …”

Comment on the results:
“These results further support the idea of …
“These results confirm the association between …
“These findings are consistent with  …”
“These match/don’t match those observed in earlier studies…”
“These results are in line with those of previous studies…”
“These findings are in agreement with those obtained by …”

Evaluate the Results:
“There are several possible explanations for this result…”
“It seems possible that these results are due to …”
“The reason for this is not clear but it may have something to do with…”
“These data must be interpreted with caution because …”
“The present results are significant in at least two major respects.”

Suggestions
“There are still many unanswered questions about …”
“There is abundant room for further progress in determining.”
“Despite these promising results, questions remain.”

Don’t just confine yourself to these sentences though, go and visit the Manchester Academic Phrase Bank where there are hundreds of sentence starters for all parts of the thesis.

Another way to get started is to take a leaf out of Pat Thomson and Barbara Kamler’s book ‘Helping doctoral students write’ by ‘stripping down’ a paragraph from a writer you admire and using it, almost like a garden trellis, to ‘train’ your own text in the right direction. Here’s the first paragraph from a paper in my library by Wajcman  and Rose (2011) Constant Connectivity: Rethinking Interruptions at Work. This paragraph is clearly part of the ‘summarising the results’ move:

Kamler and Thomson suggest you strip out the content. I like using the ‘strike through’ setting on my word processor, like so:

The picture that emerges from the analysis above is one of work practices being reshaped as employees negotiate the constant connectivity intrinsic to contemporary knowledge work. We have established a link between mediated communication and short, fragmented work episodes. What is striking is that the predominant mode of communication during the workday is now tech- nologically mediated rather than face-to-face. However, each communication episode tends to be of a short duration – on average a period of five minutes or less.

You can now insert your own findings in to this cleared out structure. This technique works best if you treat the original framework roughly, so that you produce something almost entirely new. Here’s the reworked paragraph using some concepts I am playing with in my latest work (although I haven’t actually proved any of this, so don’t quote me!):

The picture that emerges from the analysis above is one of many missed opportunities, which seems  intrinsic to the post PhD job search. We have established a link between previous experience of specific work environments and success in job seeking. What is striking is that the strength of this connection is how little the graduates paid attention to the need to articulate their previous industry experience. Each Graduate  tends to be living in a state of what Lovitts calls ‘pluralistic ignorance’.

You might be wondering: is this plagiarism? No because you are not using the original knowledge or ideas, just the structure. You can write a whole discussion section like this if you like, but you would need to find a study with very similar findings. My hunch is that you could have to hijack paragraphs from different texts and stitch them together like a patchwork quilt.

Once you get your head around the idea that there are the ‘right’ and ‘wrong’ kind of textural moves in every kind of academic writing, you have an amazingly powerful writing device that you can deploy on any section of your thesis.

There is so much more to say about the discussion section – I’ll try to get to it again this year. Your question will help me decide what to focus on, so please feel free to leave them in the comments. I’m also interested in your feelings about the discussion section and how you’ve approached constructing them in the past – do you have any suggestions for others?

Related posts

How do I start my discussion section

The difficult discussion section

References

The textural organisation of the discussion sections of accounting research articles 

Analysis of moves, rhetorical patterns and linguistic features in New Scientist articles

Constant Connectivity: Rethinking Interruptions at Work.

Want more content like Thesiswhisperer?

Visit The Whisper Collective, a site we built to continuously curate all the best research education material on the web.

Love the ThesisWhisperer?

I cover all the expenses of operating the Thesis Whisperer out of my own pocket. If you’d like to support my work, please consider becoming a $1 a month Patreon or buy a copy of my (cheap!) ebooks: Tame your PhD or The Year of Living Covidly.

Jason Downs and I have our first ‘On the Reg’ podcast book – on TextExpander. This book shows you how to use TextExpander to reduce your everyday cognitive load.

You can find links to print versions – and my other books – on the Buy Books page