By Claire Aitchison
I love a good conclusion. There’s nothing more satisfying than reading a good paper that finishes strongly, but what a letdown when there is a poor – or non-existent – conclusion!
We know that most of us read the abstract, scan the introduction and then move quickly to the discussion and conclusions sections when we read research papers (Feak & Swales, 2011 p. 40). Whether it is a thesis or journal article the conclusion is really important, so why is it that it is so often badly done? And how can we make sure it’s as great as it can be?
Firstly, I think there are some useful processes that can help ensure a successful conclusion. Especially because a PhD thesis is such a long time in the making, it is useful to begin building the conclusion over months and years – at least from the time data is being collected and analysed. I suggest these steps to students I work with.
Build a ‘Conclusions Bank’
- From mid-stage in your PhD make a new file called ‘The Conclusions Bank’ and throw into it inspirations and ‘big ideas’ as you construct your thesis. For example, this is the place you can dump insights that come to you during data analysis or when reading the literature, and it’s a good place to store chapter leftovers.
Don’t worry about organising this information until you have finished all your data chapters and you are ready to begin your conclusion. It is easy to lose sight of such thinking in the latter stages of the thesis writing when you are mentally and physically exhausted. It can be an absolute delight to find this treasure trove of ideas just as you think you’ve run out of energy and inspiration.
Within the Conclusions Bank make a separate section into which you copy and paste each of the conclusion sections from each of your chapters as you write them. Having these together means you can eyeball all the parts in one place, enabling you to better synthesise these parts and see the big picture required to make the ‘big claims for significance’. Remember that a key task of a conclusion is to identify what it is that makes the whole greater than the sum of the parts. It’s a big job for a totally blank page and an exhausted mind!
- At some point when you are toward the end of your writing, remove yourself from your work and freewrite to these questions:
- So, what have I found – and why does it matter?
- What do I know now, that I didn’t know before? (e.g., before I read the literature or before I collected and analysed the data)
- Who cares? / Who should care? (e.g., are these things of value for practitioners, for policy or theory, for improving how we collect or analyse data?)
- What do I know that no one else knows? (e.g., things that arise from my unique context or data sets)
The disappointing conclusion
As an editor or examiner, one of the most common failings I come across is a conclusion that looks and reads as if the author has run out of steam. That kind of conclusion is generally way too brief, sloppily written – and incredibly disappointing. Some examples include:
- A failure to overview the whole project, perhaps just focusing on one aspect (e.g., something the author has just explored in the section above, or their favourite aspect/part of the project).
- A collection of motherhood statements disconnected from the literature, ‘soap box’ announcements or imperatives for action that don’t necessarily flow from the evidence presented. For example, I recently reviewed a research paper where the author seemed to consider the final section as his/her chance for chest beating on issues not at all substantiated by the research presented: ‘Thus teachers should blah blah blah…’
- A lazy reiteration (even duplication) of statements from the abstract or the introduction or abstract.
- A bland re-summarising of the research and/or listing of findings.
- A failure to highlight the ‘take-home message’ – be that the key argument, key finding(s) or implications. This ‘high pass’ claim or observation is what makes a conclusion great.
So what should a conclusion do?
Remember that a conclusion may be read as a stand-alone item. As such it needs to inform the reader of what was done, how and why, what was found, and why it matters. It can be a challenge to reiterate all of this succinctly and without boring repetition, nevertheless, that’s the task of the conclusion.
Conclusions should do some or all of the following:
- remind the reader of the research problem and purpose and how they were addressed
- briefly summarise what has been covered in the paper
- make some kind of holistic assessment/judgement/ claim that pertains to the whole project (i.e., more than a descriptive summary)
- assess the value/relevance/ implications of the key findings in light of existing studies and literature
- ‘speak’ to the Introduction
- outline implications of the study (for theory, practice, further research)
- comment on the findings that failed to support or only partially support the hypothesis or research questions directing the study
- refer to the limitations of the studies that may affect the validity or the generalisability of results
- make recommendations for further research
- make claims for new knowledge/ contribution to knowledge.
(adapted from Belcher, 2009; Paltridge & Starfield, 2007; Swales & Feak, 1994)
How is a conclusion organised?
A conclusion is sometimes described as a mirror image of the introduction, in that it moves from the particular to the general. There is another sense in which the discussion and conclusion section is the reverse of the introduction: an introduction contains extended discussions on the previous existing research and literature on the topic, and relatively little on the current research. In the conclusion section the new research, positioned against existing knowledge, is the primary focus. In the concluding section, existing literature and previous research is used for confirmation, comparison or contradistinction (Swales, 2004 cited in Paltridge & Starfield, 2007, p. 147).
Every thesis is different and writers need to decide what suits their particular needs, writing style and methodological approach; however, being aware of common patterns and genres can help writers make judicious decisions to suit their own particular thesis. We know, for example, the structure of a Conclusion section in a thesis commonly follows these stages or moves:
- An introductory restatement of research problem, aims and/or research question
- A summary of findings and limitations
- Practical applications/implications
- Recommendations for further research
Given what we know about reader behaviour wherein the abstract, introduction and conclusion are often the only parts many readers bother with, it is essential that the conclusion concludes the paper in a succinct and punchy fashion. This is the last (but not only) chance to ensure the reader has clarity about what’s been done and the merits of these endeavours. Is it important that the conclusion answers the question: ‘So what?’ This is the hardest challenge for a conclusion-writer, so using strategies such as The Conclusions Bank and freewriting big ideas can be critical for building a conclusion that is great.
And finally, perhaps it is useful to remind ourselves of relevant aspects of the definition of a ‘conclusion’ – the conclusion is the end or final part; it is the result or outcome of an act or process, a judgment or decision reached after deliberation. No wonder it’s so hard!
Belcher, W. (2009). Writing your journal article in 12 weeks: a guide to academic publishing success. Thousand Oaks: SAGE.
Feak, C. B., & Swales, J. M. (2011). Creating contexts: writing introductions across genres (Vol. Volume 3 of the revised and expanded edition of English in Today’s Research World). United States of America: University of Michigan Press.
Thomson, P., & Kamler, B. (2013). Writing for peer reviewed journals: strategies for getting published. London: Routledge.
Paltridge, B., & Starfield, S. (2007). Thesis and dissertation writing in a second language: A handbook for supervisors. Oxon: Routledge.
Swales, J. M., & Feak, C. B. (1994). Academic writing for graduate students. Ann Arbor: University of Michigan Press.
This dissertation was the result of an investigation into the relative importance of construction as a curriculum organizer for the field of technology education. In particular, it concentrated on the relationship between construction technology and the principles of general education and technological literacy. The review of literature focused on the historic roles and meanings of this curriculum organizer and these principles as the discipline evolved from the industrial arts into technology education. Operational definitions were synthesized and the linkages between them was clearly identified. To address technology education's contribution to general education, or the full development of the human personality, the spheres of human/technology interaction model was developed. The model is based on the idea that people interact with technology and evaluate those interactions from three fundamental perspectives. Those perspectives were identified as the civic-life sphere, the personal-life sphere, and the work-life sphere. One hundred and forty-eight faculty members of technology teacher education programs in colleges and universities throughout the United States were surveyed. A 77% return rate was obtained. The survey included four major sections in addition to requesting limited information about the respondents and their programs. The four major sections asked the respondents to: 1) Evaluate potential goals for a K-12 technology education program. 2) Determine the relative importance of 10 study areas or curriculum organizers as they related to each of the three spheres of interaction. 3) Determine the percentage of the technology education curriculum that should be allocated to each of the three spheres of human/technology interaction. 4) Provide selected information about the way construction is offered and taught in technology teacher education programs. Medoid cluster analysis was used to evaluate the data derived from the goals of technology education portion of the survey. Using this information, three clusters were formed and initial respondent membership for each cluster was established. Subsequently, discriminant analysis was used to accomplish three goals: 1) Refine the initial assignment of respondents to the clusters. 2) Identify those variables that offered a significant level of discrimination between clusters. 3) Determine the accuracy of assignment to the clusters or groups. The canonical correlation 2, calculated by the discriminant analysis program, indicated that 66.3% of the variance was explained by the variables that were significant at a .05 level. After comparing the mean scores of the discriminating variables across the three clusters, one cluster was identified as favoring technological literacy, one favored industrial technology education, and one was ambivalent. T-tests were used to determine if any significant difference existed between clusters or groups. It was of particular interest to this research that no significant difference was found related to the relative importance of construction. All groups concluded that construction should comprise approximately 10% of the technology education curriculum. Finally, a schedule was established which allocated various percentages of the curriculum to each of the 10 study areas or curriculum organizers as they relate to the three spheres of human/technology interaction. This schedule was based on the relative importance assigned by the technological literacy cluster. The technological literacy cluster offered the most balanced allocation of the technology education curriculum across the three spheres of human/technology interaction.