The discussion in a bachelor’s or master’s thesis is the section where you explain your own research findings and relate them to existing knowledge. It is not enough to repeat how many people chose a particular answer or reproduce conclusions from tables. You need to show what the findings mean, how they compare with other authors’ research and what might explain the similarities or differences observed. This is why students often find the discussion more challenging than describing the results themselves.
The results show what you found. The discussion explains what it means. This guide takes you from a single finding to a complete paragraph, comparison with the literature and cautious conclusions.
Dean R. Hess (2023, pp. 1771-1774) identifies explaining the meaning of the findings to the reader as the discussion’s central task. Rather than presenting the results again, it should answer the question: what follows from them?
Similarly, Thomas M. Annesley (2010, pp. 1671-1674) emphasises that the discussion should explain the meaning of the findings and show how the study contributes to understanding the issue. It should also relate the results to previous research and consider possible explanations and the study’s limitations.
Results and discussion serve different purposes
An example makes the distinction easiest to understand.
The three versions below concern the same fictional finding. They illustrate the difference between description and interpretation.
Description in the results chapter
What did the study find?
Among the teachers surveyed, 62% identified an excessive workload as one source of occupational stress.
This describes a result. It reports what the study found.
Repetition in place of discussion
A version that needs revision
62% of teachers identified an excessive workload as a source of stress. This means that most teachers identified an excessive workload.
The second sentence explains nothing.
Interpretation in the discussion
A version that explains the finding
The finding indicates that an excessive workload is a frequently reported problem among the teachers studied. Its interpretation should take account of the nature of their work, which involves preparing documentation, communicating with parents and additional organisational duties alongside teaching. The finding should then be compared with previous research on sources of occupational stress among teachers.
Writing guidance for scientific texts emphasises this principle: the discussion should interpret results rather than duplicate the results chapter.
If you first need to organise your data, read the step-by-step guide to analysing and presenting research results. Base the discussion on results already presented in the thesis.
How should you begin the discussion?
The opening need not be long. Briefly remind readers what the study examined, then identify its main finding.
Docherty and Smith (1999, pp. 1224-1225) propose beginning the discussion with the study’s main findings. Only then should you turn to strengths and weaknesses, comparison with other studies, the meaning of the results and questions requiring further research.
A fictional example of a discussion opening
The study aimed to establish the level of occupational stress among primary school teachers and identify its most frequently reported sources. The findings indicate that factors related to workload and the organisation of work were particularly important in the group studied. Differences in reported stress levels were also observed according to participants’ length of professional experience.
There is no need to repeat every percentage, mean and statistical value. Detailed data belong in the results chapter.
A simple structure for writing the discussion
For each important finding, you can follow five steps:
- Your own finding: identify the key finding.
- Interpretation: explain what it might mean.
- Other authors’ research: compare it with specific publications.
- Explaining agreement or differences: consider the group, measurement and other methodological features.
- Brief conclusion: state what the data allow you to conclude.
You can apply this separately to each main research question or hypothesis.
Busse and August (2021, pp. 909-913) describe the discussion as a section that begins with the main results, then moves to their meaning and comparison with previous literature before considering the study’s wider implications.
A master’s thesis need not follow one structure mechanically, however. Logical order and linking your own results with the literature matter more.
How do you discuss the first result?
Suppose a fictional study examined students’ physical activity and stress.
It found that people who exercised more frequently had lower stress scale scores.
A fictional example of discussing your own finding
One of the study’s main findings was an association between the frequency of physical activity and perceived stress. People who exercised more frequently had lower stress levels. This finding is consistent with the research assumption adopted.
Only then should you turn to actual studies by other authors.
A template to complete with real literature
A similar association was found in a study by [author, year], which showed [...]. Similar results were also reported by [authors, year]. This may suggest that the association observed in the present study is not an isolated finding and is supported by previous research.
Replace the brackets with publications that actually exist and genuinely found the stated result. Do not select an unrelated article merely because its title concerns a similar topic.
The passages in square brackets are placeholders, not completed citations. Do not leave them in the thesis you submit.
What exactly do you compare with the literature?
‘The results are similar to other authors’ findings’ is not enough. Explain in what respect they are similar.
A fictional comparison with space for a real source
The present study found that reported occupational stress decreased as professional experience increased. A similar association was described by [authors], who observed lower stress levels among employees with longer professional experience. In both studies, experience may therefore have been associated with better coping with work demands.
If the studies differ in method or participant group, you can state this immediately:
How do you acknowledge limited comparability?
Although the association was in a similar direction, direct comparison requires caution. The study by [authors] involved a different occupational group and used a different instrument to measure stress.
This is more useful than simply stating that the results agree.
Höfler and colleagues (2018) note that a discussion must go beyond a simple comparison of ‘consistent finding - inconsistent finding’. It should also consider differences in sampling, measurement and other methodological features that may have affected the result.
A short checklist for comparing two studies
- Group and sampling: who participated, and how were they recruited?
- Measurement: did the authors measure the same phenomenon with the same or a comparable instrument?
- Conditions: do the time, place and context allow a meaningful comparison?
- Finding: are you comparing the direction or strength of an association, a response percentage or another specific result?
For example, reported job satisfaction and self-rated productivity are different outcomes, even if both articles concern hybrid working. Note the distinction before writing the paragraph. Include relevant studies whose findings differ from yours as well.
If you lack suitable publications for the discussion, you can commission the service “Literature search and selection for your thesis” for this chapter. When agreeing the scope, specify your own results and the issues to compare so the literature selected matches the discussion’s content.
What should you write when your results agree with other studies?
This is the simplest situation, but agreement alone is not enough.
Template: agreement on a specific finding
The finding is similar to results reported by [authors]. Both studies found [...]. Their agreement may suggest that the phenomenon also occurs in the group examined in the present thesis.
Template: agreement with the direction of previous findings
The findings of the present study are supported by the existing literature. [Authors] found [...], while [other authors] reported [...]. The results are therefore consistent with the predominant direction of findings in previous studies.
You can then try to explain why the results are similar.
Do not force a complex explanation if the study cannot establish the mechanism.
Agreement between findings does not itself prove that an interpretation is correct. Compare study limitations too, since similar findings may reflect similar sampling or measurement methods.
What if your results differ from the literature?
A difference does not mean that the study was conducted poorly.
This is precisely when the discussion can be particularly valuable.
Template: a difference between your own study and the literature
The finding differs from the results of a study by [authors], which found no significant association between the variables examined. This discrepancy may reflect differences in the characteristics of the groups studied, among other factors. The present study included only [...], whereas the group examined by [authors] included [...]. The studies also differed in how [...] was measured. Their findings should therefore not be treated as directly comparable.
Possible reasons for differences include the respondent group, sample size, instrument, study procedure, timing or criteria used to divide participants into groups.
You must not invent a cause and present it as a fact.
When proposing an explanation that the study cannot establish, use cautious wording:
- ‘one possible explanation is...’
- ‘the difference may be associated with...’
- ‘it may be that...’
- ‘the finding may partly reflect...’
‘The result is this way because...’ would be too categorical if the study did not establish the proposed cause.
If you need a complete chapter or a broader scope of work, see the service ‘Bachelor’s and master’s thesis writing’. We will agree the scope based on your text, results and university requirements.
What should you do with the hypotheses?
If the thesis sets out hypotheses, discuss the findings from testing them.
‘Hypothesis H1 was confirmed’ does not explain the result. Remind readers what the hypothesis stated and identify the finding consistent with it.
Fictional example: findings consistent with the hypothesis
The first hypothesis assumed that people who exercised regularly would have lower levels of perceived stress. The findings were consistent with this assumption. Lower scores on the stress scale used were recorded more often among those engaging in physical activity. This finding can be considered in relation to previous research.
Then turn to the literature.
If a hypothesis was not supported, it still needs discussion:
Fictional example: a hypothesis not supported
The findings did not support the proposed association between professional experience and occupational stress. No statistically significant association between these variables was found in the group studied. This differs from the findings reported by [...]. One possible explanation may be a difference in the composition of the sample [...].
An unsupported hypothesis does not make the thesis a failure.
Hypotheses should match what was actually measured. To check how they connect with questions, variables and instruments, use the guide to research methodology in a thesis.
How do you avoid overreaching conclusions?
A non-significant result also needs cautious interpretation
If a test found no significant association, do not automatically write: ‘It was proved that no relationship exists.’
A lack of statistical significance does not itself prove the absence of a relationship. Sample size, measurement precision and the size of the effect under investigation may all matter.
We discuss the practical meaning of p-values, effect sizes and confidence intervals in the guide to statistical analysis in a master’s thesis. It also provides an example of a complete test result report.
Höfler and colleagues (2018) explain that interpretation should account for limitations in measurement, sampling and the assumptions adopted. They also warn against drawing overly broad conclusions from statistical results alone.
Safer wording: No statistically significant association between the variables examined was found in this study.
‘It was proved that there is no relationship whatsoever between the variables’ goes beyond that finding.
Do not say ‘effect’ when the study showed only an association
This is a common problem, particularly in survey research.
If a one-off correlational study found an association between physical activity and stress, you should not automatically write:
A version that needs revision
Physical activity causes a reduction in stress.
The study may only have shown that the two variables are associated.
It is safer to describe the association without attributing a cause:
Fictional example: two more cautious versions
More frequent physical activity was associated with lower stress levels.
An association between physical activity and stress levels was observed in the group studied.
Höfler and colleagues (2018) place particular emphasis on distinguishing an association from a causal relationship. Inferring causes requires much stronger grounds than establishing a statistical association alone.
Examples: a paragraph and a longer discussion passage
How do you build a discussion paragraph?
A good paragraph can move from your own finding through interpretation and the literature to a brief conclusion.
A fictional example paragraph
The study found that teachers with less professional experience reported high levels of occupational stress more often than those who had worked in the profession longer. This may suggest that experience is associated with finding some teaching demands easier to manage. The finding should be considered in relation to previous research on the relationship between professional experience and stress.
If other authors found a similar association, you can describe the findings as consistent. If their results differed, identify the discrepancy and consider possible reasons, such as sample composition or the instruments used.
The example illustrates the beginning of an argument. A complete paragraph still needs a specific source that you have actually read and a conclusion matching the outcome of the comparison.
What might a longer discussion passage look like?
An original illustrative example. The passage below is fictional and does not describe an actual study. The suggestions for comparison with the literature need to be completed with suitable publications.
The study aimed to identify the difficulties preschool teachers encountered when working with children with special educational needs. Participants most often reported difficulties organising work in a diverse group and adapting their activities to individual needs. Problems with direct interaction with the child were reported less often.
This may suggest that difficulties do not always originate in direct work with a child with special educational needs. Organisational conditions, group size and access to specialist support may also matter. Interpretation requires caution, however, because the study examined teachers’ reports rather than direct observations of their work.
The findings should then be compared with other authors’ research on teachers’ difficulties in inclusive education. Start with publications examining a similar occupational group. You can then consider studies on access to specialist assistance, teacher preparation and cooperation with parents. If previous findings are similar, specify where they agree. If they differ, consider whether this could reflect a different participant group, research conditions or instrument.
The finding on self-reported professional preparation needs separate discussion. If teachers rated their competence highly while also frequently reporting difficulties, these findings should not be treated as contradictory. Difficulties may also reflect organisational conditions rather than professional preparation alone. Establishing this would, however, require a study covering a wider range of variables.
In this way, the discussion does not retell the entire results chapter. It selects the main findings and interprets them.
You can organise the discussion around research questions
In a master’s thesis, successive research questions or hypotheses can provide a useful organising thread for the discussion.
If the thesis contains four hypotheses, you can discuss each in turn. This does not mean you must create four separate subsections.
For example, the first few paragraphs might discuss the level of the phenomenon studied. The next might address its relationship with gender, followed by professional experience. The final part can bring the findings together and explain their wider meaning.
Şanlı, Erdem and Tefik (2013) propose a similar approach to organising the middle of the discussion: each main finding can be discussed in a separate, coherent paragraph.
Study limitations also belong in the discussion
A good discussion should not present the study as though it had no weaknesses.
Limitations may include:
- a small group of participants;
- convenience sampling or studying only one institution;
- limited scope for generalising the findings;
- the self-report nature of the questionnaire;
- a researcher-developed instrument whose psychometric properties have not previously been examined extensively.
A fictional example of discussing limitations
Interpretation of the findings should take account of the study’s limitations. The group was recruited through non-probability sampling, so the findings cannot be generalised directly to all preschool teachers. The study also relied on respondents’ reports, meaning that their answers reflect subjective experiences and assessments.
You need not apologise for every limitation. Simply explain how it may affect interpretation.
Considering strengths and limitations is a recurring element of proposed discussion structures in the scientific writing literature (Docherty, Smith, 1999; Annesley, 2010; Ghasemi et al., 2019).
Identify strengths too, where they reflect how the study was actually conducted. Discussing limitations should help readers assess the findings, rather than merely fill a required paragraph.
Can you discuss the practical implications of the findings?
Yes, if these genuinely follow from the study.
A fictional example of a cautious practical suggestion
The findings suggest a need to pay closer attention to organisational working conditions for teachers working with children with special educational needs. Further examination of access to specialist support and opportunities for teachers to work with specialists appears particularly worthwhile.
Do not, however, move from a small survey to sweeping recommendations:
An example of an overreaching recommendation
The results prove that the entire Polish education system needs to change.
The scope of your conclusions must match the scope of the study.
Ghasemi and colleagues (2019) explain that a discussion may address the theoretical and practical implications of the findings, but interpretation should remain within the limits of the data obtained.
How do you close the discussion?
The final passage should bring together the main point rather than summarise the entire thesis.
A fictional closing example requiring literature consistent with the actual findings
The study identified the difficulties teachers most frequently reported when working with children with special educational needs. The findings indicate that these problems concern both direct work with children and organisational conditions and access to support. The results partly align with findings in the existing literature, but their interpretation must account for limitations arising from how participants were selected. Further research could include a larger, more diverse sample and combine questionnaire data with other methods of collecting material.
The closing passage should lead readers to the main conclusion and show the study’s significance, a possible application or a specific direction for further research. Cals and Kotz (2013) propose a similar principle for a clear ending.
What should you avoid in the discussion?
The most common mistake is describing every table and chart again. The discussion is not a second results chapter.
Do not introduce new results that have not been presented earlier. If a test, percentage or association first appears in the discussion, it should usually have appeared in the results. Hess, as well as Busse and August, clearly distinguish the functions of these sections: present the data in the results and interpret their meaning in the discussion (Hess, 2023; Busse, August, 2021).
Avoid attributing things to the results that the study did not establish. Correlation does not automatically establish causation. A lack of statistical significance does not prove that no relationship exists whatsoever. Respondents’ opinions do not always describe an objective state of affairs.
The discussion should therefore explore interpretations confidently while drawing conclusions cautiously.
Useful sentence openings for the discussion
You need not start every paragraph with ‘the results showed’. That quickly becomes monotonous. You can vary the wording:
- ‘One of the study’s main findings was...’
- ‘The finding indicates that...’
- ‘The observed association may be related to...’
- ‘Similar findings were reported by...’
- ‘Different results, however, were obtained by...’
- ‘The discrepancy may partly reflect...’
- ‘Interpretation of this finding should consider...’
- ‘This finding supports/does not support the assumption that...’
- ‘However, this study cannot establish that...’
- ‘Further research could include...’
You do not need to use them all. They simply help you move from a bare result to discussing it.
How do you check whether the discussion is well written?
Before finishing the chapter, try one check: if you remove all the tables, is it still clear what the findings mean and how they compare with previous research?
- Have you stated what the study found?
- Have you explained what the main findings mean?
- Does the comparison with the literature specify where findings agree or differ?
- Have possible explanations for discrepancies been distinguished from facts?
- Have the study’s limitations been linked to interpretation?
- Do the conclusions stay within the scope of the data?
- Does every citation identify a publication that actually presents the findings discussed?
- Have draft placeholders been removed and bibliographic details completed?
There is no single compulsory discussion structure for every thesis. Docherty and Smith propose an organised structure, but later authors also stress that the arrangement should primarily reflect the research question and the nature of the findings (Docherty, Smith, 1999; Şanlı, Erdem, Tefik, 2013).
The key distinction: the results show what you found; the discussion explains what it means.
If you have collected your material and need the results analysed, discussed and brought together into conclusions, you can choose the service “Analysis and presentation of research results for your thesis”. If the scope should also cover developing the methodology or instrument and linking these with the results and discussion, see the service “Thesis research section: methodology, results and discussion”. We will agree the tasks needed based on your stage of work, outline, available sources and actual data.
Frequently asked questions about the discussion
Do I have to repeat all the numbers in the discussion?
No. Detailed data belong in the results section. In the discussion, cite only those findings that are necessary to explain the significance of the study, compare it with the literature, and formulate cautious conclusions.
Does an unconfirmed hypothesis mean a poorly executed thesis?
No. Such a result also requires discussion. Describe what was not found in the studied group, compare the result with the literature, and consider the study's limitations. Do not equate a lack of significance with proof of a complete absence of a relationship.
What should I write when my results differ from those of other studies?
Show what the difference is and compare the groups, research instruments and methods. Make clear that any proposed explanations are tentative. If you cannot establish the reason for the discrepancy, state this explicitly.
Is there one required structure for the discussion section?
No. The structure should suit the research question, the nature of the results and the university's requirements. It can be helpful to organise the discussion around successive questions or hypotheses. Whatever the structure, the task is to interpret the results, relate them to the literature and state the limits of the conclusions.
Bibliography
These publications concern the principles of writing and interpreting a scientific discussion. They are not sources for the fictional data on teachers, students or physical activity used in the examples.
- Annesley T. M. (2010). The discussion section: your closing argument. Clinical chemistry, 56(11), 1671-1674. DOI: 10.1373/clinchem.2010.155358.
- Busse, C., & August, E. (2021). How to Write and Publish a Research Paper for a Peer-Reviewed Journal. Journal of cancer education : the official journal of the American Association for Cancer Education, 36(5), 909-913. DOI: 10.1007/s13187-020-01751-z.
- Cals, J. W. L., & Kotz, D. (2013). Effective writing and publishing scientific papers, part VI: discussion. Journal of clinical epidemiology, 66(10), 1064. DOI: 10.1016/j.jclinepi.2013.04.017.
- Docherty, M., & Smith, R. (1999). The case for structuring the discussion of scientific papers. BMJ (Clinical research ed.), 318(7193), 1224-1225. DOI: 10.1136/bmj.318.7193.1224.
- Ghasemi, A., Bahadoran, Z., Mirmiran, P., Hosseinpanah, F., Shiva, N., & Zadeh-Vakili, A. (2019). The Principles of Biomedical Scientific Writing: Discussion. International journal of endocrinology and metabolism, 17(3), e95415. DOI: 10.5812/ijem.95415.
- Hess D. R. (2023). How to Write an Effective Discussion. Respiratory care, 68(12), 1771-1774. DOI: 10.4187/respcare.11435.
- Höfler, M., Venz, J., Trautmann, S., & Miller, R. (2018). Writing a discussion section: how to integrate substantive and statistical expertise. BMC Medical Research Methodology, 18, 34. DOI: 10.1186/s12874-018-0490-1.
- Şanlı, Ö., Erdem, S., & Tefik, T. (2013). How to write a discussion section? Turkish Journal of Urology, 39(Suppl. 1), 20-24. DOI: 10.5152/tud.2013.049.
