How to Check 200 Answer Sheets Faster Without Losing Fairness

Two hundred answer sheets is not an unusual pile. Five sections of 40 after a unit test, or one batch at a coaching centre after a weekly mock, and the stack is on your table. The checking itself is not hard. What makes it slow is making the same judgement 200 times. What makes it unfair is making it slightly differently on copy 180 than on copy 1.

This guide deals with both problems. Nothing in it needs software. The last section covers where AI first-checking fits, if you want to go further.

Why checking takes so long

For a sense of scale: CBSE’s 2025 instructions to board examiners ask for a full 8-hour working day and 20 answer books a day in main subjects (25 in other subjects). At that pace, 200 board copies would take one examiner ten working days. A school test is shorter, but the arithmetic is the same: one minute saved per copy is more than three hours saved across 200.

Most of that time goes into three places that have little to do with reading answers:

• Re-deciding. Without a written scheme, you decide again on every copy whether a half-right answer gets the half mark. Each decision is quick. Two hundred of them are not.
• Switching. Copy by copy, you jump from a 1-mark definition to a 5-mark long answer to a numerical and back, a dozen times per copy, reloading a different standard each time.
• Admin. Totalling, carrying marks to the cover page and the register, and fixing mistakes after students point them out. CBSE’s own list of common examiner errors is mostly this kind of thing: answers left unassessed, wrong totals, marks wrongly carried from the inside pages to the title page, and marks in words and figures that do not match.

Each method below attacks one of these.

Before the first copy: a scheme and a trial run

The biggest single time-saver is deciding what each answer is worth before you start. Write a marking scheme for every question worth more than a mark or two: the points you are looking for, the marks for each, and what earns a half mark. The points should add up exactly to the marks for the question. If you have not written one before, our guide on how to write a marking scheme has worked 2, 3 and 5-mark examples.

Then test the scheme on a handful of copies before you trust it.

1. 1

   Pick five to eight varied copies

   Take a few from the top, middle and bottom of the stack, not just the neat ones. Include at least one weak copy and one strong one.

2. 2

   Mark them against the scheme

   Note every answer the scheme does not cover: a correct point in different words, a common wrong answer, a half-right one.

3. 3

   Fix the scheme, then freeze it

   Add the alternatives you will accept and the near-misses you will not. After this, change it only for a real gap, and if you do, go back over the copies already marked for that question.

4. 4

   If colleagues share the paper, compare

   Each teacher marks the same few copies and you compare. CBSE builds the same check into board evaluation: the head examiner goes through the first five answer books each evaluator marks on day one before the rest are handed out.

Mark question by question, not copy by copy

Instead of finishing copy 1 and moving to copy 2, mark Question 1 on all 200 copies, then Question 2 on all 200, and so on. With paper booklets, a practical version is to go section by section: all of Section A across the stack, then all of Section B.

It tends to be faster because the scheme for one question stays in your head, and after twenty copies you have seen most of the ways students answer it. It can also be fairer. When one person marks a whole script, a strong or weak answer early on can colour how they read the same student’s later, unrelated answers. Researchers call this the halo effect. Ofqual’s 2014 review of item-level marking (each question marked separately) found that the evidence is limited, but that marking by item seems to be at least as reliable as marking whole scripts and may remove the halo effect. Examiners in one study it describes named speed as the main advantage.

The same review records the downside examiners reported: marking one question again and again is boring, and boredom can lead to slips of attention. So:

• Group two or three short questions into one pass rather than doing a single 1-mark question 200 times.
• Mark the long answers, which need the most judgement, when you are freshest.
• Shuffle the stack between questions. A review of marking research written for Ofqual by NFER in 2013 describes contrast effects: good work is judged more favourably when it follows weaker work, and poor work more severely when it follows strong work. If the pile is always in roll-number order, the same students always follow the same neighbours. A shuffle stops that bias landing on the same copies every time.
• Fold back or cover the name on the cover page if you can, so you are marking the answer and not the student you know.

Work in passes, one job per pass

Trying to mark, total, comment and fill the register in one go is how errors creep in. Split the work into passes, each with a single job.

A simple tally sheet makes the totalling pass far more reliable: one row per student, one column per question and a total column. Fill it from the margins, then add it up both ways. The sum of the student totals must equal the sum of the question totals. If it does not, there is a slip somewhere, and you find it before a student does. The sheet also shows at a glance which question the class found hardest.

Boards take this seriously too. When CBSE announced on-screen marking for Class XII answer books from the 2026 examinations, the first benefit on its list was the elimination of totalling errors.

Time-box the work

A pile of 200 feels endless because it has no shape. Give it one.

• Time ten. Time yourself on ten copies of one question and multiply by twenty. At 40 seconds a copy, that question costs a little over two hours for 200 copies. Now you can plan.
• Book blocks, not evenings. Plan blocks of 45 to 60 minutes with a short break between them, and decide which questions go in which block.
• Long answers first. They need the most judgement, so give them your best hour, not the last one of the night.
• Set a stop rule. If you notice you are reading the same line twice, stop. A rushed mark costs more time later than a break costs now.
• Clear the quick ones fast. Unattempted answers, one-word answers and multiple choice take seconds. Do them in their own quick pass so they never interrupt a judgement pass.

Sample for fairness: a moderation check

Moderation means checking a sample of your own marking, the way a head examiner checks an examiner’s. It is a small job next to the marking itself, and it catches drift you cannot feel while marking.

1. Pick the sample. About ten copies spread across the order you marked them in (some early, some middle, some late), plus every copy within a mark or two of a pass mark or grade boundary, plus the top two or three.
2. Re-mark blind. Cover your first marks and re-mark the long answers in the sample.
3. Compare. A half-mark difference here and there is normal. A pattern is not: if your late copies are consistently stricter than your early ones on one question, go back through that question on the whole stack.
4. Swap with a colleague when a paper is shared. Five copies each, marked against the same scheme, is enough to see whether you are applying it the same way.

Feedback shorthand students actually understand

Writing the same comment sixty times is a large hidden cost. Short codes fix that. The Education Endowment Foundation’s 2016 review of written marking reports that research finds no difference in effectiveness between coded and written-out feedback, provided students understand what the codes mean, and that codes are likely to save time. The same review is frank that the overall evidence on written marking is thin, so treat this as sensible practice rather than a guarantee.

The review also draws a useful line between careless mistakes and misunderstandings. A slip can simply be marked wrong. A misunderstanding is better met with a hint or a question that points the student back to the idea. Codes handle the first kind. Save your written comments for the second.

Keep a scrap list, as you mark, of the mistakes you see most often. Ten minutes of class feedback on those does more than the same comment written into forty copies.

Where AI first-checking fits

Everything above makes the first pass faster. AI first-checking takes the first pass off your desk. The discipline stays the same: a fixed scheme per question, a review by the teacher, and marks released only when the teacher is satisfied. Your job shifts from marking 200 copies to reviewing them, and the moderation step above applies to AI-checked copies exactly as it would to a colleague’s.

Before you adopt any tool, check what it does with copies it cannot read, and make sure marks never reach students before a teacher has looked at them. Our buyer’s guide to AI answer sheet evaluation lists twelve questions to ask, and our comparison with manual checking sets the two side by side.