Writing up Research

Discussion

This page covers:

The purpose of the discussion section and what it includes

Common problem

Organization

An example

Some advice

PURPOSE AND CONTENT OF THE DISCUSSION SECTION

The discussion section is for comment on and explanation of the results.

It includes:

	Explanation of results: the writer comments on whether or not the results were expected, and presents explanations for the results, particularly for those that are unexpected or unsatisfactory.
	References to previous research: comparison of the results with those reported in the literature, or use of the literature to support a claim, hypothesis or deduction.
	Deduction: a claim for how the results can be applied more generally (a conclusion based on reasoning from the results, e.g. we fed fish a new feed, all the fish gained weight, therefore the new feed causes fish to gain weight).
	Hypothesis: a more general claim or possible conclusion arising from the results (which will be proved or disproved in later research).

COMMON PROBLEM

The discussion does not discuss - simply supplies more detail about the results obtained.

Solution: remember that the discussion should explain the results.

ORGANIZATION

There are two basic ways of organizing the results and discussion:

1. Presenting all the results, then giving a discussion (perhaps in a different section)

2. Presenting part of the results then giving a discussion, presenting another part then giving a discussion, etc.

The method of organization you use will depend on the quantity and type of results you obtain from your research. You should look for a method of presentation that makes the information and ideas you are presenting as clear as possible to the reader.

EXAMPLE

Below is part of the discussion section from "Strategies of failure diagnosis in computer-controlled manufacturing systems: empirical analysis and implications for the design of adaptive decision support systems." Part of the results section was included on the Results page of this online course. The square brackets indicate the information that has been left out.

Task: read through the discussion section and try to find the purpose of each sentence.  Click on the highlighted phrases for suggested answers. Here are some ideas to help you: explanation, purpose, supporting reference, theory, description, summary, comparison.

Strategies of failure diagnosis in computer-controlled manufacturing systems: empirical analysis and implications for the design of adaptive decision support systems

Konradt, U. International Journal of Human-Computer Studies (1995) 43, 503-521

SUMMARY (adapted from the abstract) Objective of the study: to investigate strategies in failure diagnosis at cutting-machine-tools with a verbal knowledge acquisition technique. Method: semi-structured interviews with mechanical and electrical maintenance technicians; protocol analysis was performed on the data. Analysis of strategies was performed according to technicians’ experience, familiarity with the problem and the problem complexity. The technicians were categorized by level of experience.

4. Discussion

[Introductory comments]

4.1 Strategies in Real-life Diagnosis

In fault diagnosis in advanced manufacturing systems, four typical strategies are found:
- Restriction of diagnosis to components which are known and susceptible to failures (“Historical information”);
- Performing tests that result in least efforts (“Least effort”);
- Reconstruction of the conditions that lead to the failure (“Reconstruction”);
- Perception of symptoms, i.e. loose connections, odors, sounds, play (“Sensory checks”).

These strategies appeared in about 60% of the total observed strategies. The primary strategy was “Historical Information”. This corresponds to the results of Hoc (1989). In information theory, strategies such as “Information uncertainty”, which eliminates the greatest number of failure causes, or “Split half”, which results in a binary splitting of the problem space, are economical ways to shorten the problem space. We found that in real-life failure diagnosis, even maintenance experts with more than 20 years experience seldom used these strategies. One reason may be that the use of this strategy requires information about conditional probabilities and a fully described problem space that cannot be supposed for troubleshooting in complex manufacturing systems.

[Discussion of literature on novel failures and strategies used.]

SOME ADVICE

	If you are putting your discussion into a discussion section separate from the results, you may want to provide a summary of the results to remind your reader of your main findings.
	Put your results in context (e.g. by comparing them with previous research, or with existing theory) in order to explain them.
	Give reasons to account for differences between your research and previous research or existing theory, or to explain unexpected results.
	Although there may be some repetition of information in the results and discussion sections, it should be kept to a minimum.
	Remember too that the focus should be different: while you are simply presenting results and making them meaningful to your reader in the results section, in the discussion section you are explaining them.

Back to the Writing up Research homepage

The Basics of Good Writing

How to reference

References (for works cited in Writing up Research)

Useful writing links