Writing an effective thesis discussion can be challenging because it is so open ended. The purpose of the discussion is to (1) interpret your results, (2) discuss the significance of your results, (3) place your work in the context of previous work, (4) discuss the limitations of your study, and (5) suggest next steps to advance understanding and/or to improve real-world situations. The discussion section should expand upon ideas presented in the introduction, literature review, and results, and sometimes even the materials and methods. It should convince the reader that your work was worthwhile.
General considerations
Throughout a thesis or other scholarly manuscript, the results of completed studies should be described in the past tense (e.g. "36 of 43 patients showed improvement with treatment A, while only 25 of 44 patients showed improvement with treatment B"). In contrast, present tense is used to describe ideas that currently appear to be true based on available evidence (e.g. "treatment A is more effective than treatment B for most patients"). A good thesis discussion will typically progress from discussing specific results to discussing broader concepts that currently appear to be true.
You should include citations and modifying statements as needed (e.g. "Previous work by Lee et al. (2017) found no difference in the effectiveness of treatment A compared to treatment B. However, their study was limited to patients in advanced stages of the disease."). You will also need to include suggestions for further research, which can be described in present tense (e.g. "Additional studies are needed to determine. . .") or future tense (e.g. "Future studies will determine. . .").
I will now move into more detailed suggestions for each part of the discussion. Notice the occasional use of first-person pronouns (e.g. "I suggest. . ." or "We believe. . ."), which is now generally acceptable in many fields if used sparingly. Check the specific guidelines provided by your graduate program (and target journal). Also notice the use of subheadings to break the text into shorter sections on specific topics, which is recommended for any long manuscript.
Discussion part 1: Interpret your results
Your discussion should relate directly to the research question(s) you presented in the introduction, so a good way to start the discussion is to clearly restate your research question(s) (e.g. "The current study aimed to determine whether treatment A or treatment B was better for patients with condition X.") The discussion section follows the results section, so the next obvious step is to state what you can conclude from your results (e.g. "We found that a higher percentage of patients showed improvement with treatment A than with treatment B, and therefore conclude that treatment A is more effective than treatment B for most patients with condition X").
There are two general strategies for structuring the discussion of results. The first strategy is to discuss the results in the same order as they were presented in the results section. Alternatively, you can discuss your most interesting or surprising results first. This second strategy is most effective when the results are especially novel. As with any manuscript, the goal is to keep your reader engaged.
In some fields, the thesis discussion is expected to include comments on all of the results, even minor results that may not be statistically significant (e.g. "Treatments C, D, and E produced results that were statistically indistinguishable from placebo, leading to the conclusion that they were ineffective at the concentrations tested."). Note that you should not just state the results (e.g. "Treatments C, D, and E produced results that were statistically indistinguishable from placebo"), but should also state a conclusion (e.g. "they were ineffective at the concentrations tested").
In some other fields, brief comments on minor results are typically included in the results section, and the discussion section is used to elaborate on the most noteworthy results. In some cases, it may be appropriate to write a combined "results and discussion" section, where results are presented alongside initial conclusions, and broader implications are discussed toward the end. Check the requirements for your graduate program (and target journal).
Discussion part 2: Discuss the significance of your results
Clearly explain what is gained from your research. Sometimes a study is very significant (e.g. These results are among the first to show that treatment A is more effective than treatment B for condition X."). Other times—especially for undergraduate or master's theses based on a few months of data collection—the results primarily provide information about what doesn't work (e.g. "These results indicate that treatments C, D, and E are ineffective at physiologically relevant concentrations.)
Discuss some of the possibilities that your research opens up (e.g. "These results raise the possibility that early treatment could improve daily life for patients with mild cases of condition X, and might even delay disease progression.) but do not overstate (e.g. "These results prove that early treatment will improve daily life for patients with mild cases of condition X, and will delay disease progression."). Be optimistic but realistic, whether your results are highly significant or not (e.g. "These results suggest that the solubility of molecules like C, D, and E need to be improved before they can be effective.").
Discussing the significance of your results (part 2 of this essay) is a natural extension of discussing what your results mean (part 1 in this essay), and may not require a separate section.
Discussion part 3: Place your work in the context of previous work
Put your work into a broader context. Here you should revisit some of the previously published work that was described in your introduction and literature review, and clarify how your work expands upon or challenges that work. Most importantly, you should discuss how your work advances understanding in your research field.
For a thesis based on disappointing data, your ability to effectively discuss previous work and potential future work can support the argument that you have developed the necessary skills to move forward in research. These skills are further demonstrated by many other aspects of a well-researched and well-written thesis.
In some cases, your results may reveal unexpected connections. This may require you to discuss additional bodies of work in the discussion section, which can become a second literature review section. As you may be starting to see, there are many different topics you can include in your discussion section. You need to choose your topics carefully, and cover them in sufficient but not overwhelming detail. You want your discussion to cover all essential information, remain a manageable length, and retain the reader's interest. All of the different possibilities can make writing your discussion section very challenging.
If your results contradict previous work, you need to provide some possible explanations. As a rationale scholar, it is especially important to consider all reasonable options. It could be that (1) the earlier work was flawed, incomplete, or not exactly comparable to your work, or (2) your work is flawed, incomplete, or not exactly comparable to the earlier work. You might not be able to determine the exact cause of the discrepancy, but contradictions provide an obvious way to transition into the next section.
Discussion part 4: Discuss the limitations of your study
Every study has limitations, and here you should discuss the major limitations of your research. What can't the results tell us? In what situations are the findings not relevant or well supported? There may be limitations due to your sample population, experimental techniques, data analysis, etc. Discuss these limitations honestly, but then explain why your study is useful despite these limitations. You might point out that your study included multiple different populations, different experimental techniques, alternative methods of data analysis, etc. You might also emphasize that your conclusions are limited to a subset of a larger population, and that additional work needs to be done to determine whether the same conclusions apply to a larger population. This brings us to our next point.
Discussion part 5: Suggest next steps to advance understanding and/or to improve real-world situations
Once you've placed your work into a broader context, you can imagine the future possibilities. Here you might discuss questions that remain to be answered, or questions that have been newly raised by your work. There might be specific studies that would be an obvious extension of your work, and/or exploratory studies that would shed light on promising areas for novel research. Alternatively, you might suggest ways in which your research could be applied to real-world situations to improve outcomes.
Many manuscripts and presentations end with a discussion of possible next steps. I tend to like this type of ending for discussions of basic research, such as which genes are activated in which cells. However, other people recommend always ending with conclusions, where you restate the main findings of your research in a compact form. This might involve a separate "conclusions" section or as a final paragraph in the "discussion" section. Ending with conclusions makes a lot of sense for studies that might directly impact real-world situations (e.g. Treatment A it more effective than treatment B for condition X.).
In any event, you should follow departmental recommendations, and end with a compelling and memorable message for your readers.