PrepKar, by Compete India · Mains

How we mark

This is the whole rubric your answer is read against, published before you write rather than after. It is ours. The Union Public Service Commission publishes no marking rubric, so nobody has one — and anyone who tells you they are scoring your answer against the Commission's own scheme is telling you something that does not exist.

1. The four bands

Calibrate to this examination, not to a school scale. Every band below is a judgement about one dimension of one answer, and the bands are set where a successful candidate actually sits — which is a good deal lower than most people are told (section 4).

Competitive and Topper range are separated by well under one mark on a ten-mark question. That is the entire reason we publish bands instead of numbers.

The same question, in four bands

Each of these describes what an answer does — none of them is an answer, and none of them is anyone's copy. The stem is a real one, and it carries two demands:

Examine whether the constitutional office of the Lok Sabha Speaker has become vulnerable to partisan politics under the current anti-defection regime. What institutional changes are required to ensure the neutrality of the Lok Sabha Speaker in managing a polarised House?

Off target

The page is about the Speaker in general — when the office was created, what its powers are, a list of its duties. Nothing on it is about vulnerability to party politics, and the second demand is never attempted. It answers the subject, not the question.

Below competitive

The first demand gets three lines and the rest of the page is background that would fit any question about the Speaker. The second demand appears once, at the end, as “hence reforms are needed”. Both demands are named. Neither is carried.

Competitive

Both demands have their own space on the page. The first is argued from a named provision rather than asserted; the second gives three changes and says, for each, what it would fix. The last sentence takes a position instead of calling for balance.

Topper range

The same, and then the answer is built around the conflict the stem is about rather than around the topic. A named committee and a specific provision carry an argument instead of decorating one, the space is divided in proportion to what each demand is worth, and the conclusion is a proposition the page has earned.

This sentence is printed beside every band we ever give you, unchanged: Estimated. The Commission publishes no marking scheme, and after moderation it does not keep question-wise marks at all — so no one, including us, can tell you what this answer would have scored. These bands are PrepKar's own, and how we set them is on How we mark.

2. The six dimensions

They are in this order for a reason: most confidently judged from a photograph first, least last. Demand coverage and relevance decide the overall band; length, analysis and substantiation matter; presentation is a note rather than a score. We do not average them.

1. 1 Did it do what was asked

   Whether the answer did each separate thing the stem ordered, with real content and real space against each.

   Below competitive looks like
   A stem with two demands where one is answered thoroughly and the other is a closing sentence. It has structurally lost half its marks, however good the half is.

2. 2 The question, not the topic

   Whether every paragraph is traceable to a phrase in the stem, rather than background that would fit any question on the subject.

   Below competitive looks like
   The first 40% of the answer is background that would fit any question on the same subject.

3. 3 Length

   Whether the answer landed near its limit, and what the gap cost — unclaimed marks here, or time the rest of the paper needed.

   Below competitive looks like
   Well over the limit, which costs time and space that later questions in a three-hour paper need — or well under it, which leaves marks unclaimed.

4. 4 Analysis, and a view

   Whether the answer names the conflict the question is about, weighs it, and lands somewhere with a stated reason.

   Below competitive looks like
   Points listed on both sides and a closing line that “a balanced approach is needed”, which is not a view. The opposite failure is asserting one side as though there were no conflict at all.

5. 5 Named specifics

   Whether an argument is carried by a named article, committee, judgment, scheme or a figure with its year. Presence, not correctness.

   Below competitive looks like
   “Several committees have recommended” and “recent studies show”, where a named committee and a named study would have done the same work.

6. 6 Presentation

   Legibility, paragraphing, whether the headings help. Judged last and weighted least: a sentence that cannot be read scores nothing, and that is the whole of the reason.

   Below competitive looks like
   Nothing published says the Commission awards marks for handwriting, underlining or diagrams. The honest reason legibility matters is mechanical: a sentence that cannot be read scores nothing.

3. Why this rubric is ours, and what it is built from

The Commission publishes no rubric and no marking scheme. What it does publish about assessment is three sentences in the examination notification and the existence of a Head Examiner moderation process. So the six dimensions above are derived from those three sentences, and everything beyond them is our judgement — which is why it is on this page under our own name rather than presented as anybody else's standard.

The three things the Commission does say are that an answer must be

• relevant, meaningful and succinct — which is where demand coverage, relevance and length come from;
• that the General Studies papers test the ability to analyze, and take a view on conflicting socio-economic goals, objectives and demands — which is analysis and a view, almost word for word;
• that the Essay is credited for effective and exact expression — which is the only published support for anything like presentation, and it is why presentation is weighted least.

Those quotations are from the Civil Services Examination notification, Section III, published at upsc.gov.in.

4. Why we print a range, not a mark

Because the number does not exist to be told. The Commission's own frequently-asked-questions document is explicit that “once the evaluation process is complete, neither ‘raw marks’ nor ‘question-wise marks’ subsist” — after moderation there is no question-wise mark, for anyone, ever. Nobody can tell you what your answer would have scored, and anyone who prints a number to two decimal places is printing a guess.

The second reason is calibration, and it is the one that will sting. Three numbers, all of them published by the Commission itself, all of them written totals across the seven papers that count for merit — 1750 marks:

• The floor. The General-category written cut-off for the last cycle was 739 out of 1750 — 42.2%. (cut-off, PDF)
• The middle, and this is what Competitive is anchored to. Across all 958 candidates the Commission actually recommended, the median written total was 775 out of 1750 — 44.3%. (marks of recommended candidates, PDF)
• The ceiling. The highest written total in that whole list — rank one — was 867 out of 1750 — 49.5%. Nobody cleared half.

So the entire useful distance between an answer that just misses and the best written total in the country is about seven marks in a hundred — under one mark on a ten-mark question. No source publishes a General-Studies-only figure for that population, so these are all-papers totals and we do not pretend otherwise.

If you have been told a good answer is 7 out of 10, that scale is not this examination's. A band is the honest resolution: it says which side of the cut-off your answer sits on and what would move it, and it does not pretend to a precision that no evidence supports.

So where does the range at the top of your copy come from? The six bands are the reading; the range is arithmetic on top of them. Each band is worth points — nothing for off target, one for below competitive, two for competitive, three for topper range — and the six are added up and placed on the scale above, so six competitive bands come out at about 4 to 5 marks in 10, and six topper-range bands at about 5.5 to 6.5. It is a range and not a single figure because the distance between the top two bands is under a mark, and one number would be claiming to see a difference nobody can see. And if 4 to 5 in 10 reads low, read the three landmarks again: the cut-off is about 42 marks in every hundred, and nobody on that list cleared half.

5. What we do not do

• We do not compare your answer to anyone else's. No copy attributed to a high scorer, no ranked example, no “delta” against a previous attempt. The provenance and the consent for those copies are unstated wherever they appear, and we are not adding to that.
• We do not predict your rank or your score, and we do not claim a percentage for how accurately we read or mark anything.
• We do not claim the Commission's authority. There is no official rubric, no official marking scheme and no published meaning for a directive word. Where the marking says “here” or “in this marking”, that is deliberate and literal.
• We do not tell you a fact on your page is wrong — we list it as worth re-checking. A confident correction that turns out to be ours is the one mistake that actively costs you something.
• We do not mark diagrams well yet. A real answer can carry a lot of hand-drawn diagram; we describe each one and check whether it is labelled and referred to, and we say so on your feedback rather than quietly counting it as prose. We do not put a fraction on it, because nobody has measured one we would stand behind.
• We do not publish your answer, use it to train anything, or show it to another student. Your photographs are yours.

The transcript we read off your page is shown to you beside the marking, with a button on every line that says I wrote something different here. It is there because we will misread handwriting sometimes, and a misreading you can see and correct is the only kind worth having.