A law dissertation does not run on hypotheses in the statistical sense a psychology or nursing dissertation uses. Most South African LLB and law mini-dissertation research is doctrinal — it asks what the law currently says, whether that is settled, and what it should say — and doctrinal research is structured around a research problem, a primary question, sub-questions and propositions rather than a testable statistical hypothesis. Seven steps to write all four, in the order a supervisor expects to see them.
Step 1: Understand what replaces a hypothesis in legal research
Where an empirical dissertation states a hypothesis to be tested against data, a doctrinal legal dissertation states a proposition to be tested against legal sources — statutes, case law, Bills, parliamentary records and academic commentary. A proposition is your working answer to the research question, stated before you write the analysis chapters, that the chapters then defend or qualify through legal reasoning rather than statistical inference. If your faculty’s proposal template uses the word “hypothesis,” most South African law departments accept a proposition in its place for doctrinal work; confirm with your supervisor rather than forcing a statistical framing onto research that is not empirical. If your study is genuinely empirical — a survey of legal practitioners, for instance — it needs ethics clearance and a different design entirely, covered in our explainer on what ethics clearance involves at a South African university.
Step 2: Identify which type of legal question you are asking
Three question types cover most South African law dissertations, and naming yours before you draft anything decides what your propositions will look like:
| Question type | What it asks | What a proposition looks like |
|---|---|---|
| Doctrinal / interpretive | What does a specific provision or line of cases currently mean, and is that meaning settled? | “Section X should be read to require Y, based on its wording, purpose and the courts’ treatment of analogous provisions.” |
| Law reform / normative | Does the current law achieve what it should, and if not, what change would? | “The current provision fails to achieve [stated purpose] because [reasoned gap], and [specific reform] would close it.” |
| Comparative | How does South African law’s treatment of an issue compare with a chosen jurisdiction’s, and what follows from the comparison? | “The comparator jurisdiction’s approach to [issue] addresses a gap in South African law that could be adapted without conflicting with [named constitutional or statutory constraint].” |
Output: one sentence naming your question type, before you draft the research problem.
Step 3: Anchor the problem to a primary source you can name today
A researchable law problem rests on a primary source you can point to precisely — a numbered section, a case citation, a Bill number — not a general area of law. “Access to justice” is a field, not a problem; “the effect of section 4(2) of a specific Act on unrepresented litigants in the magistrates’ court” is a problem. The same anchoring logic used to select a topic at LLM level applies at LLB and mini-dissertation level, and the four tests set out in our guide to researchable LLM dissertation topics — can you name the primary source, is the question genuinely open, can it be answered from sources you can reach, does it fit your word count — apply just as directly to an LLB research report, scaled to a shorter word count and usually one primary source rather than several.
Step 4: State the research problem in one paragraph
The research problem paragraph does three jobs: names the primary source, states the tension or gap precisely (a provision whose wording admits two readings; a judgment that leaves a question open; two statutes that appear to conflict), and states briefly why the tension matters — to litigants, to a specific industry, to a constitutional value. Keep it to one paragraph. A research problem that takes a page to state is usually a field description wearing a problem’s clothes.
Output: a single paragraph naming the source, the tension, and its significance.
Step 5: Draft the primary research question as one interrogative sentence
Turn the problem into a single question a reader could answer yes, no, or with a stated legal position — not a topic restated with a question mark at the end. Weak: “What is the law on expropriation without compensation?” Strong: “Does section 12(3) of the Expropriation Act satisfy the ‘just and equitable’ standard in section 25(3) of the Constitution?” The strong version names both instruments, states the specific legal standard being tested, and can be answered by the end of your analysis chapters.
Output: one interrogative sentence, both instruments or sources named, no vague adjectives (“effective,” “adequate”) left undefined.
Step 6: Break the primary question into three or four sub-questions
Sub-questions give you your chapter structure, and each should be answerable from a distinct body of source material. A typical doctrinal pattern: what does the provision’s text and legislative history show (statutory interpretation chapter); how have the courts applied or interpreted it (case law chapter); what does comparative or international practice suggest, if your scope includes it (comparative chapter); and what follows for the specific problem you named in Step 4 (analysis and recommendations chapter). Each sub-question should map to one chapter, so that by the time you have answered all of them, the primary question in Step 5 is answered too.
Step 7: State a proposition for the primary question and, where useful, for each sub-question
Write your proposition as your reasoned, provisional answer — not a guess, but a position you can currently defend from the sources you have already read, which your dissertation will then test rigorously rather than merely illustrate. A proposition that turns out, after full analysis, to need qualification or partial revision is a normal and often stronger outcome than one confirmed exactly as stated; examiners read an unrevised proposition with some suspicion, since real legal analysis usually complicates a starting position at least somewhat. State the proposition for your primary question, and, for a longer mini-dissertation, one proposition per sub-question as well.
Output: a stated, defensible proposition for the primary question, ready to be tested rather than merely asserted across your analysis chapters.
A worked example, from problem to proposition
Using the labour law example from Step 5: Problem — sectoral numerical targets set under recent employment equity legislation may function as a quota rather than an affirmative-action measure, a distinction the Constitutional Court has previously drawn but not applied to this specific mechanism. Primary question — are ministerially determined sectoral numerical targets a “measure” designed to protect or advance persons disadvantaged by unfair discrimination under section 9(2) of the Constitution, or do they impose a quota contrary to section 15(3) of the Employment Equity Act? Sub-questions — how has the Constitutional Court defined a section 9(2) measure in its leading affirmative-action judgments; what distinguishes a target from a quota in the Act’s own text and in labour law commentary; where do the current sectoral targets fall against that distinction? Proposition — the targets are more likely to satisfy the section 9(2) test than to constitute a quota, because [reasoned basis from the case law], though the rigidity of specific sub-sector figures raises a genuine, testable risk on the quota side of the line.
Common pitfalls supervisors flag at this stage
Four patterns account for most of the back-and-forth between a South African law student and a supervisor at the problem-and-question stage. First, a question that is really two questions joined by “and” — each half needs its own sub-question structure, and forcing them into one primary question usually means neither is answered properly. Second, a proposition that simply restates the question in the affirmative (“the provision does satisfy the standard”) without a reasoned basis attached, which is not a proposition but a guess dressed as one. Third, sub-questions that overlap rather than partition the ground, so that two chapters end up covering the same material from slightly different angles instead of each doing distinct work. Fourth, a comparative question with no stated reason for the chosen comparator jurisdiction — supervisors expect a sentence explaining why that jurisdiction, not merely that its law happens to be accessible in English.
None of these four is fatal if caught early. Bring your draft problem, question, sub-questions and proposition to your first supervision meeting rather than a fully written chapter built on an unreviewed foundation; the earlier a misaligned question is caught, the less rewriting it costs later.
Turning your question and propositions into a proposal this week
A research problem, question, sub-questions and propositions are worth nothing until they sit inside a proposal your supervisor has signed, and the proposal itself follows a structure covered in our guide to writing a research proposal for a South African university. Tesify takes the problem, question and propositions you have drafted using the steps above and builds the proposal around them — problem statement, research questions, a doctrinal method section and a reference list in your faculty’s Harvard or footnote variant — while the legal reasoning and the position you take stay entirely yours. There is a free plan.
Draft your law dissertation proposal with Tesify
If you are still choosing which statute, Bill or judgment to anchor your topic to, forty worked examples across seven fields of South African law are set out in our guide to researchable LLM dissertation topics, and the levels of degree your research problem needs to match are compared in the difference between an honours report, a master’s dissertation and a doctoral thesis.
Frequently asked questions
Can an LLB research report use a hypothesis instead of a proposition?
Most South African law faculties accept either term for doctrinal work, since the underlying logic is the same: a stated, defensible working answer tested through analysis rather than statistics. Use whichever word your faculty’s proposal template asks for, and explain in one sentence what you mean by it if the template is ambiguous.
How many sub-questions should a law dissertation have?
Three to four is typical for an LLB or mini-dissertation, matching a three- to four-chapter analysis structure; a research LLM or doctoral thesis can carry more because it covers more ground. More than five sub-questions usually signals a problem too broad for the word count.
Can my primary research question change after I start writing?
Yes, within reason, and it is common once deeper reading reveals the real tension is narrower or different from your first framing. Flag the change to your supervisor and update your proposal’s research question formally rather than letting the dissertation drift from what was approved.
Is a “what should the law be” question acceptable, or must it be purely descriptive?
Normative, law-reform questions are entirely acceptable and common, provided the reform proposal is reasoned from the doctrinal analysis rather than asserted as a policy preference. A normative dissertation still needs a rigorous doctrinal chapter establishing what the law currently is before it argues for change.
Do I need a proposition for every sub-question, or just the primary question?
A proposition for the primary question is essential; propositions for each sub-question strengthen a longer mini-dissertation but are sometimes optional for a shorter LLB research report. Check your faculty’s proposal template, since practice varies by department.
What if my analysis fully confirms my original proposition with no qualification at all?
It happens, but re-read your analysis chapters critically before submitting, since examiners are often more persuaded by a proposition that survives close scrutiny with some genuine complication than by one that is confirmed too neatly. If the confirmation genuinely holds, state clearly what alternative readings you tested and rejected, and why.