A conclusion chapter that restates the discussion chapter wastes your reader's time, and one that promises more than the data showed undermines your credibility right at the finish line. The job here is synthesis and grounded recommendations. Nothing more, nothing less.
Of all the chapters in a dissertation, the conclusion is the one students most often either shortchange or overload. Shortchanging it means treating it as a two-page formality tacked onto the end of the discussion. A paragraph of "in summary" bullet points that adds nothing a reader hasn't already absorbed. Overloading it means the opposite mistake: introducing fresh interpretation, new citations, or a late attempt to argue a point that should have been made three chapters earlier. Both failures come from the same root cause, not having a clear mental model of what a conclusion chapter is actually *for*. It is not a second discussion chapter, and it is not a new introduction. It is the chapter where you step back from the specific findings and speak, briefly and directly, to the bigger picture: what the study accomplished, what it means beyond its own dataset, and what should happen next.
| Discussion Chapter | Conclusion Chapter |
|---|---|
| Detailed interpretation of each finding | Brief synthesis of the overall study |
| Connection to specific literature | Broader statement of contribution to the field |
| Limitations, in detail | Recommendations grounded in those limitations |
| Specific, actionable suggestions for future research |
Notice the pattern in that table: everything on the right side is *shorter and more general* than its counterpart on the left. That's the discipline of writing a conclusion. You've already earned the right to be detailed in the discussion chapter, so the conclusion can move faster and speak at a higher altitude. If you find yourself writing a paragraph in the conclusion that re-explains a specific statistical result or re-quotes a participant, that paragraph almost certainly belongs one chapter earlier, and its presence here is usually a sign the discussion chapter needs another pass rather than that the conclusion needs to be longer.
A weak recommendation reads as generic: "future research should explore this topic further" applies to almost any study and tells the reader nothing. A grounded recommendation comes directly from a specific limitation or unanswered question your own study surfaced.
It helps to think of every recommendation as answering a single implicit question: "so what should someone actually do differently because this study exists?" If a recommendation doesn't have a clear answer to that question, it isn't a recommendation yet. It's a wish. Committees read a lot of dissertations, and generic recommendations are one of the fastest ways to signal that a candidate hasn't fully processed the implications of their own work. The fix is almost always the same: go back to your limitations section, pick the two or three limitations that matter most, and write one recommendation that responds directly to each.
Consider a hypothetical qualitative study examining why newly licensed nurses leave rural hospital positions within their first two years. The study's sample was drawn from three hospitals in a single state, and interviews revealed that inadequate mentorship during the first six months was the most frequently cited factor in participants' decisions to leave. A weak conclusion would say: "Future research should examine nurse retention in more settings." That sentence is true, but it is true of almost every nursing workforce study ever written, and it gives the next researcher nothing to act on.
A grounded version instead reads something like: "Because this study's sample was limited to three hospitals within one state's rural health system, and because mentorship structure emerged as the dominant factor in participants' attrition decisions, future research should replicate this design across rural systems in at least two additional states, with particular attention to how formal mentorship program duration (rather than mere presence or absence of a program) relates to first-year retention." Notice what changed: the recommendation now names the specific limitation (geographic scope), the specific finding that drove it (mentorship structure), and a specific, testable next step (comparing program *duration*, not just existence). A hiring manager or nurse educator reading this conclusion also gets a distinct, practical recommendation, formalize and extend mentorship windows past the first ninety days, that stands apart from the research-agenda recommendation aimed at future scholars.
Most conclusion-chapter problems fall into a small, predictable set of patterns. Knowing them in advance makes them far easier to catch in your own draft.
Master's theses and doctoral dissertations both need a conclusion chapter, but committees hold them to different standards of scope and originality. A master's thesis conclusion is generally expected to summarize competently and suggest sensible next steps. Demonstrating that the student can apply an existing method and think clearly about its implications. A doctoral dissertation conclusion carries a heavier burden: it has to make an explicit, defensible claim about the study's contribution to the field's existing body of knowledge, not just to the specific problem it addressed.
In practice, this means a doctoral conclusion typically needs a distinct subsection, often titled "Contribution to the Field" or "Implications for Theory", that doesn't have a direct equivalent in most master's-level work. It also means doctoral recommendations for future research are expected to engage with the theoretical or conceptual framework established in chapter two, not just propose a logistical variation on the same study (more sites, larger sample). A doctoral committee wants to see that the candidate understands how their specific findings shift, confirm, or complicate the theoretical conversation they situated themselves in from the beginning, which is exactly why the purpose statement and conceptual framework from chapter one deserve a second look before this chapter is finalized.
Before sending the conclusion chapter to your advisor, it's worth running through a short self-check rather than relying on a single read-through, since by this point you've likely read the chapter so many times that genuine errors and gaps start to feel invisible. Ask yourself, honestly, whether the chapter passes each of the following tests. And if any answer is "not quite," that's the section to revise before anyone else sees it.
Read your purpose statement from chapter one again before drafting the conclusion. A strong conclusion explicitly answers whether and how the study fulfilled that original purpose. Closing the loop the introduction opened, rather than introducing new ideas at the very end.
Synthesis, grounded recommendations, and a clear answer to your original purpose.
Typically shorter than the discussion chapter. Often 5–10 pages depending on program requirements, since it's synthesizing rather than re-explaining everything in detail.
Generally no. Anything substantial enough to discuss should have appeared in the discussion chapter first. The conclusion synthesizes what's already there.
Most programs expect both, since they serve different audiences. We structure these as clearly separated subsections so each is easy for a reader to locate and apply.
That's still a legitimate, reportable outcome. The conclusion should state plainly what was and wasn't supported, and explain what that means, rather than trying to spin a null or contradictory result into something it isn't. Committees generally respect an honest, well-reasoned account of unexpected findings far more than a conclusion that overstates support for the original hypothesis.
A brief, high-level mention is fine if it directly sets up a recommendation, but a full re-listing of every limitation belongs in the discussion chapter. In the conclusion, limitations should appear only as the springboard for a specific recommendation, not as their own standalone section.