Part B at a Glance: Pattern, Qualifying Bar and Scoring Arithmetic
This chapter sets out the rules that decide whether a strong performance on the trade-relevant paper turns into a qualifying score. It covers the CBT-2 structure, the 35 per cent Part B bar, the arithmetic of one-third negative marking, the attempt counts that clear the bar, and a timing routine for 75 questions in 60 minutes. Later chapters assume the figures worked here, so each calculation is shown in full.
KarmSakha preparation topics covered
- CBT-2 structure: Part A and Part B, question counts and durations
- The 35 per cent qualifying bar for Part B and why Part B sits outside merit
- One-third negative marking and the expected value of a guess
- Attempt counts that clear the bar, with a worked table
- Choosing the one trade paper that matches an ITI or degree background
- A 60-minute timing routine for 75 questions and a two-pass method
- The notice clauses to check before sitting CBT-2
- What this edition covers and what it leaves out
CBT-2 structure: Part A and Part B side by side
The second stage computer based test, CBT-2, has two parts. Under the Detailed Centralised Employment Notice CEN No. 01/2026, Part-A has 100 questions and runs for 90 minutes, and Part-B has 75 questions and runs for 60 minutes. The notice gives a total of 175 questions in 2 hours and 30 minutes. Part A covers Mathematics, General Intelligence and Reasoning, and Basic Science and Engineering. Part B is the trade-relevant paper: the candidate sits one trade paper, chosen against the qualification groups printed in the notice. This chapter concentrates on Part B, but Part A matters because its marks are the only marks used for shortlisting.
Two rules in the notice keep the parts apart. Part A has its own category minimums for eligibility: 40 per cent for UR and EWS, 30 per cent for OBC (NCL) and SC, and 25 per cent for ST. Those minimums apply to Part A alone. Part B carries a single qualifying figure of 35 per cent for every candidate, irrespective of category or community. The notice calls Part B a qualifying test in nature, and it does not use Part B marks in the merit calculation. Part A is therefore the score that ranks a candidate, and Part B is the hurdle that must be cleared before Part A counts at all.
| Item | Part A | Part B |
|---|---|---|
| Questions | 100 | 75 |
| Duration | 90 minutes | 60 minutes |
| Subjects | Mathematics; General Intelligence and Reasoning; Basic Science and Engineering | One trade paper chosen by the candidate |
| Marks | 100 (as reported by secondary sources; not stated in the notice text) | 75 (as reported by secondary sources; not stated in the notice text) |
| Role in selection | Counted for shortlisting once Part B is qualified | Qualifying only, 35% required, not used in merit |
| Negative marking | One-third mark per wrong answer (stated for CBT-2 as a whole) | One-third mark per wrong answer (stated for CBT-2 as a whole) |
The 35 per cent Part B bar: 26.25 marks of 75
The notice states the Part B qualifying figure as 35 per cent, with no separate marks figure. The threshold in marks is found by applying that percentage to the Part B maximum. Using the maximum of 75 marks reported by secondary sources, the threshold is 0.35 × 75 = 26.25. Negative marking applies to CBT-2 (para 12.2(f)(iii)), so the marks counted against the bar are normally the net score after deduction. The notice does not spell out this step, so treat the net score as the figure compared with 26.25. A net score of 26.25 qualifies, while a net score of 26 does not, and that 0.25 gap can decide the outcome.
Net scores can be fractional. Each wrong answer removes one-third of a mark, so a candidate can finish with 26.33 or 25.67. The qualifying check is a comparison of decimal values: any net score at or above 26.25 passes. A candidate who rounds a mental score of 26.33 down to 26 has misread the result. The practical habit is to keep a running net score during the paper, using the formula in the next section, rather than a raw count of correct answers.
Part B also sits outside merit. The notice says only Part-A marks are counted for shortlisting for the further stages, provided the candidate secures the qualifying marks in Part B. A very high Part A score cannot compensate for a Part B fail. The two parts are not traded against each other, so Part B is the single gate that decides whether the rest of the candidate's performance is used at all. Consider a candidate with 95 marks in Part A and a Part B net of 26. Part B is checked first. At 26 the candidate is 0.25 short of the bar, so the 95 marks play no part in shortlisting.
One-third negative marking and the value of a blind guess
The notice states that there is negative marking at one-third of a mark for each wrong answer in CBT-2 (para 12.2(f)(iii)), and important note (xvii) states it for both CBT-1 and CBT-2. The rule is given for CBT-2 as a whole and is not split into Part A and Part B. A wrong answer costs one-third of a mark, and an unattempted item costs nothing. The net score is therefore S = C − W/3, where C is the number of correct answers and W is the number of wrong answers. Whether guessing helps depends on the chance that a guess is correct, and that chance can be estimated from the options that remain.
Take a blind guess among four options. The probability of a correct answer is 1/4 and of a wrong answer is 3/4. The expected mark per item is E = (1/4)(+1) + (3/4)(−1/3) = 0.25 − 0.25 = 0. Random guessing across a full four-option paper therefore has zero expected gain. It neither helps nor harms on average, and only an answer chosen with better than 25 per cent confidence raises the expected score. This is why elimination is the main skill for unknown items: each option removed raises the chance that a guess is correct.
For k equally likely options, the expected mark per item is E = (4 − k)/(3k). With k = 4 this gives 0. With k = 3 it gives 1/9, or about +0.11 per item. With k = 2 it gives 1/3. An item with one option removed is therefore worth attempting, while a fully blind item is worth leaving alone unless the candidate has a real reason to prefer one option. The arithmetic is the same whatever the trade, because the penalty is fixed by the notice.
Attempt counts that clear the bar
Because the bar is a fixed mark total, the number of correct answers needed depends on how many items the candidate attempts. For N attempts with c correct and N − c wrong, the net score is c − (N − c)/3. Setting this at least equal to 26.25 and solving for c gives c ≥ (78.75 + N)/4. The result is a minimum count of correct answers, and it is rounded up to the next whole number because a count of correct answers is always whole.
The table below applies the formula to five attempt counts. With all 75 items attempted, 39 correct is the minimum: 39 − 36/3 = 27, while 38 correct gives 38 − 37/3 = 25.67. The gap shows why skipping items is costly even though skipping carries no penalty. Skipping does not reduce the net score directly, but it lowers the number of correct answers that can be collected. The efficient approach is to answer every item the candidate can answer with confidence, then pursue the eliminated items in the second pass.
The accuracy column matters for planning. At 30 attempts the candidate needs 28 correct, which is 93.3 per cent accuracy, a level that few candidates sustain across a paper of unfamiliar items. At 60 attempts the requirement falls to 35 correct, or 58.3 per cent. These figures show how much accuracy each extra attempt must carry. A plan built around the 50-attempt row, which needs 33 correct at 66.0 per cent, is a realistic target for a prepared candidate. A plan built around 30 attempts asks for near-perfect accuracy.
| Items attempted | Minimum correct | Net score at minimum | Accuracy among attempted |
|---|---|---|---|
| 75 | 39 | 27.00 | 52.0% |
| 60 | 35 | 26.67 | 58.3% |
| 50 | 33 | 27.33 | 66.0% |
| 40 | 30 | 26.67 | 75.0% |
| 30 | 28 | 27.33 | 93.3% |
Choosing the trade paper: electrical and electronics groups
Part B is one paper, and the candidate sits one trade from the list the notice gives against the candidate's qualification. The notice groups trades into four qualification rows. The first two cover this book. Electrical engineering and its combinations are linked to Electrician, Instrument Mechanic, Wireman, Armature and Coil Winder, and Refrigeration and Air-conditioning Mechanic. Electronics engineering and its combinations are linked to Electronics Mechanic and Mechanic (Radio and TV). This edition does not teach the Refrigeration and Air-conditioning Mechanic paper, so a candidate taking that trade needs a separate source.
The notice gives a direct rule for ITI and trade-apprenticeship holders: such candidates appear in the section with questions from their own trade (para 12.2(i)(D)). Degree and diploma candidates choose one trade from the list against their engineering discipline (para 12.2(i)(E)). The choice should therefore match the trade of the ITI certificate, or the discipline of the degree or diploma. Where a qualification could fit more than one row, the notice's wording decides, not a coaching site's regrouping. The notice's own table, not a summary, is the reference.
Trade syllabi come from the Directorate General of Training (DGT), and the notice directs candidates to dgt.gov.in for the syllabus of each technical trade. Part B questions are drawn from the DGT syllabus for the chosen trade. This book teaches the electrical and electronics theory that those trades share, so it complements the DGT text rather than replacing it. A candidate should read the DGT syllabus for the chosen trade alongside the matching chapters. A practical way to use the two sources is to note, for each DGT trade unit, which chapter here covers its underlying theory, and to practise the matching items in that chapter first.
Time budget: 48 seconds per question and a two-pass routine
Part B allows 60 minutes for 75 questions. That is 3,600 seconds divided by 75, or 48 seconds per question on average. The average is a budget, not a fixed allowance. Some items can be recognised in ten seconds, while others need a calculation that takes two minutes. The routine therefore treats the paper as two passes, so that quick marks are banked before difficult items consume the clock. The 40-second first-pass limit used later in this chapter is a planning figure taken from that average, not a rule set by the notice.
In the first pass, spend no more than about 40 seconds on each item. Answer what is known at once, eliminate options on uncertain items, and mark each item still open for a return visit. Plan the first pass to finish after 50 minutes, which leaves ten minutes for the second pass. The arithmetic is 50 × 60 = 3,000 seconds, and 3,000 divided by 75 gives 40 seconds per item. A candidate who is behind schedule at the 25th item should shorten the time spent on the remaining items rather than stall on one of them.
In the second pass, return to the marked items in order of their chance of success. Items with two options left come first, since each has an expected value of +1/3. Items with three options left come next, at +1/9. Items with no elimination are attempted only if time remains and the candidate has a reason to prefer one option. In the last minute, check that the answer sheet matches the intended answers, because an answer entered against the wrong item is scored like any other wrong answer.
