A chapter outline from a question treeপ্রশ্নের গাছ থেকে অধ্যায়ের রূপরেখা

The question tree

A chapter answers one question, and a question that size is never answered in one move. Break it into three to five sub-questions, each small enough that a reader could see, on its own, what evidence would settle it. "Why does uptake of index insurance remain low" is a main question nobody could answer in a single section; "what do farmers themselves say is stopping them" is a sub-question with a shape: interview or survey evidence, not a regression.

Each sub-question asks for a different kind of evidence, and naming that in advance is most of the outline's work: a question about how a term has been used needs a review of definitions, a question about what farmers say needs qualitative material, coded and quoted, and a question about willingness to pay needs an experiment or a survey with a price manipulation in it. Write the evidence type beside each sub-question before drafting; one with no evidence type beside it is a question the chapter cannot actually answer.

1. Write the main question as one sentence, the one the chapter's last paragraph has to answer.
2. Break it into three to five sub-questions, each answerable from one kind of evidence.
3. Name the evidence each sub-question needs: a definition, an interview, an experiment, a comparison.
4. Turn each sub-question into a section, and give the section a heading that states what you found, not what you looked at.

A heading is a claim, not a topic

"Farmer attitudes to insurance" is a topic: it says what a section is about and nothing about what it argues, so the reader must read the whole thing to see whether it agrees with the last section or contradicts it. "Farmers cite basis risk before they cite price" is a claim, and a reader of only the headings already has the argument, in miniature. A claim heading disciplines the writer too: a topic lets a section drift, since almost anything about attitudes belongs under it, while a claim is a promise, and every paragraph beneath it supports that promise or does not belong.

Order by argument, not by chronology

The tree's sub-questions do not have to appear in the order you thought of them, and the sections built from them should not appear in the order the sources were published. A chapter is an argument, not a shelf of research summarised as it happened to be read, and the ordering question is always the same one: what does the reader need to believe before the next section makes sense. A definitional confusion is usually worth clearing up first, since a reader who does not know "uptake" is measured three different ways will misread every effect size that follows; evidence about what people actually say tends to come before evidence about what an experiment measured, because the numbers are easier to read once the reader knows what they are meant to explain; and a comparison with what has worked elsewhere usually comes last, where the chapter turns from describing a problem to pointing at what might fix it.

Word budgets

A section with no target length grows to fill whatever time is left before the deadline, and it is rarely the section that most deserves the space. Set a budget for every section before drafting, in proportion to how much of the argument's weight it carries rather than how many sources you happen to have collected for it: a section built on one decisive experiment can be short, while one untangling three incompatible definitions used across forty papers usually needs the room.

The paragraph plan: claim, evidence, so what

A section is a sequence of paragraphs, and each paragraph is the same three-part move at a smaller scale: it opens with a claim, the one thing this paragraph is arguing; gives the evidence for that claim, a finding, a number or a quotation, cited so a reader can check it; and closes by saying what that evidence means for the chapter's question, the "so what" that stops the paragraph being a fact sitting on the page with nothing attached to it.

A paragraph that ends on the evidence and never says what it means has done two thirds of the work, and a reader left to draw the conclusion may draw a different one from the one intended.

Worked example: index insurance uptake

Notice the order: definitions first, since a reader who does not know "uptake" is measured three ways will misread what follows; farmer voice before the price experiments, because the numbers make more sense once the reader knows what is on farmers' minds; the comparison with other programmes last, where the chapter turns from diagnosis toward what might be done.

Where it goes wrong

• A main question broad enough that no sub-question could answer any part of it.
• A heading naming a topic, so two consecutive sections could swap places unnoticed.
• Sections ordered by publication year, so a strong recent finding sits buried mid-chapter instead of doing its work at the start or the end.
• No word budget, so the section with the most sources grows longest, whether or not it carries the most weight.
• A paragraph that states a claim and stops, its evidence left as a citation the reader must go and check.

• The main question is one sentence the chapter's last paragraph has to answer.
• Three to five sub-questions, each with its evidence type written beside it.
• Every heading states a claim, not a topic.
• The order follows what a reader needs to believe first, not the order sources were read.
• Every section has a word budget, set before drafting, summing to the chapter's target length.
• At least one paragraph was planned as claim, evidence, so what, before the section was drafted in full.

The questions room builds the tree, sub-question under main question, linked to what the library holds; the writing desk turns the outline into a draft with a running word count against the budget, so a section past its target is visible before the chapter is finished, not after.