The conclusions are the shortest part of the work and the most read after the abstract. Many people reach the end of a paper and read only the last paragraph, so what you put there is what stays. And yet they tend to be written in a hurry, repeating the discussion or adding claims that have not been shown anywhere.
Their function is very specific: to answer the objectives you set, with what you found, in a few lines. They are not the place to discuss, to cite or to introduce new ideas. They are where the reader checks whether the work did what it promised.
This guide explains how they differ from the discussion and the abstract, how long they should be, how to structure them and the commonest mistakes, with one paper example and one thesis example.
It is easy to confuse three things that speak of the same material. The abstract tells the whole work in a few lines, from question to conclusions. The discussion interprets the results, compares them with the literature and covers limitations. The conclusions, by contrast, only answer: each objective, with what the data showed.
One way to remember: the discussion answers “what do these results mean?” and the conclusions answer “what do we know now that we did not know?”. If what you write in the conclusions needs a reference or an explanation, it is discussion. If it repeats a figure, it is results.
In an article, four to six sentences, often a single paragraph. Many journals do not ask for a separate section and the close of the discussion acts as the conclusion; others do ask for one, and then it is brief. In a thesis, a short chapter that usually answers each specific objective separately, with a sentence or two per objective and a general conclusion at the end.
The length rule is that they fit on one screen and can be understood without having read the rest. If someone reads only the conclusions, they should be able to tell what was done and what was found.
For an article this sequence works, and fits in a paragraph:
“A brief cognitive restructuring programme reduced anxiety about oral exams in humanities students relative to a waitlist, with a medium effect. The result is consistent with what has been found in health sciences, although it comes from a convenience sample at two universities and a short follow-up. The data justify testing the programme with a longitudinal design and in other degrees before recommending its general use.”
Three sentences, no figure, no reference and nothing that is not in the results. It answers, delimits and closes.
Objective 1: “The programme reduced anxiety about oral exams relative to the waitlist.” Objective 2: “The effect was not significantly maintained at three months, although the estimate pointed in the same direction.” Objective 3: “No clear differences were observed by academic year, an exploratory result that needs confirmation.” General conclusion: “The results support the programme’s short-term efficacy and leave its persistence open.”
Each conclusion corresponds to an objective, says what was found (including what was not) and marks what is exploratory. If an objective has no conclusion, or a conclusion has no objective, there is a problem worth resolving before you hand in.
The first, the isolated reading: ask someone to read only your conclusions and tell you what was done and what was found. The second, the underline test: each claim has to be traceable to a result in the previous section; if not, it goes or moves to the discussion. The third, the objectives test: list your objectives and check that each has its conclusion, in the same order and with the same language.
If you pass them, your conclusions answer what you promised and claim nothing the results do not support. The check takes a few minutes.
Paste your conclusions, ideally with the last stretch of the discussion, and you will see the first objection a reviewer would raise, with the exact sentence from your text and what it would take to fix it. It looks at what this guide describes: whether the conclusions answer the objectives, whether they claim something the data do not support, whether the verb is stronger than the evidence, whether they introduce ideas that never appeared before and whether they hide what did not work out.
It returns one objection, the one that weighs most. It needs at least 350 characters and what you paste is not stored. Because it only sees the text you paste, if deciding something would require the objectives or the results it says so separately.
The discussion interprets the results and compares them with the literature; the conclusions only answer the objectives with what the data showed, briefly and without adding anything new. If you need a reference or an explanation, it is discussion.
In an article, four to six sentences, often a paragraph. In a thesis, a short chapter with a sentence or two per objective and a general conclusion. They should fit on one screen and be understood without reading the rest.
It should not be necessary: the conclusions answer your objectives with your results. If you need a citation to support a claim, it probably belongs in the discussion.
Better not, or very few. The figures are in the results and are interpreted in the discussion; the conclusions say in words what they showed, with a verb that matches the strength of the evidence.
Say so in the matching conclusion: that the expected effect was not observed or that results were inconclusive is a legitimate conclusion and carries information. Leaving it out is bias.
No. Many fold the conclusion into the last paragraph of the discussion and others ask for it separately with a word limit. Check the author instructions and the journal’s recent articles.