Some assessments have sections where every question must be answered, and others where students get a choice — "answer two questions from this section", or "answer EITHER Question 6 OR Question 7" — or where a marking guide's grid doesn't split the sections the paper actually has. The Section rules dialog lets you set each section's scoring rule, group the questions students choose between, add or remove sections, and move questions between them, so the paper totals and each student's score come out right no matter how it's structured.
This article explains the scoring rules, how choices are grouped, how to restructure sections, and how all of it affects marks.
On a submission review page, each section has a header band (for example Section A: Reading) with a Section rules button at its right-hand end — a gear icon plus the label; on a narrow screen only the gear shows. Click it to open the dialog.
If the assessment has no sections at all yet — common on older assessments, or on a marking guide that prints one combined grid with no section split — the dialog opens with a single seeded All section instead of refusing to open. You're never blocked from getting in to set this up.
Every section in the dialog's Sections list has a short code (A, B, or A1 / A2 when two sections share a letter), a title box and a Scoring rule dropdown.
The default. Every question in the section counts toward the total. If a section has three 10-mark questions, the section is worth 30 marks and all three are scored for every student.
An Every-question-counts section is always out of the sum of its question marks. If the marking guide printed a different total for that section, the printed figure isn't used, so a paper whose questions add up to 40 is always out of 40.
Use this for standard sections where all questions are compulsory.
Use this for "answer X of Y" sections — for example, a section with three essay questions where the student only has to answer one, or four questions where they pick two.
Choosing it reveals a keep [N] of [M] field, where M is the number of options the student chooses between. For each student, Marking.ai marks every option they attempted in the section, keeps the N highest-scoring options, and excludes the rest from the total — so each student automatically gets their best result without you deciding by hand.
The two-section rule. A paper where students answer one question from each of two groups — say a Shakespeare question and a 19th-century-novel question — needs two Best-of-group sections, each keeping 1. Putting every question into one Best-of-group section keeping 1 is the common mistake: it counts only the single highest-scoring answer across both groups instead of one from each.
In a Best-of-group section, students choose between options, and an option is a whole question — including all of its numbered parts. So in "answer EITHER Q6 OR Q7", where Q7 has parts 7.1 to 7.9, the section has two options: Q6, and Q7 with all nine parts counted together.
Under each Best-of-group section, the dialog lists one line per option — for example Q6 · Essay · 1 question · 25 and Q7 · Contextual · 9 questions · 25 — showing its name, how many questions it holds, and what it's worth. You can rename an option on its line.
If a section's options aren't worth the same, or don't add up to the total the marking guide printed, an amber adds up to N pill appears on the option and a warning appears under the section. It's a warning, not a block — you can still save. The section stays out of the same total for every student either way.
New papers usually arrive with their options already grouped, because extraction reads the choices from the marking guide. On a new paper, check the grouping is right rather than building it from scratch.