Quick Answer
SSC CGL Tier 2 Paper 2 (Statistics) 2026 is a 100-question, 200-mark paper held over 2 hours, with +2 correct and −0.50 wrong. Only JSO applicants and Statistical Investigator Grade-II applicants take it. Qualifying marks: 30% UR (60/200), 25% OBC/EWS (50/200), 20% others (40/200). Scribes get 2 hours 40 minutes. Final merit comes from Tier 2 only: Paper 1 gives 390 marks; JSO and Statistical Investigator candidates add Paper 2 on top, for a total of 590. Tier 1 only shortlists. Tier 2 is tentatively December 2026; no date announced yet.
Key takeaway: Paper 2 is a specialist paper. If you did not apply for JSO or Statistical Investigator Grade-II, you will never see it. If you did, it is worth 200 of your 590 merit marks.
Source: SSC CGL 2026 notice (PDF), Para 13 and Para 16.
Latest Changes
The dated record of this cycle. The vacancy revision covers the whole CGL 2026 examination — the notice states no post-wise split, so this note quotes no JSO or Statistical Investigator vacancy numbers.
| Date | What happened | Status |
|---|---|---|
| 21 May 2026 | SSC uploaded the CGL 2026 notice, with the Paper 2 scheme and syllabus in Para 13 and the merit rules in Para 16 | Official |
| 24 Sep 2026 | SSC revised the tentative vacancy count for CGL 2026 from 12,256 to 10,731 | Official |
| 30 Sep – 30 Oct 2026 | Tier 1 held across multiple shifts; Tier 1 marks count only for shortlisting, not for final merit | Over |
| 4 Nov 2026 | SSC released the Tier 1 provisional answer key with response sheets | Official |
| 6 Nov 2026 | No Tier 1 result and no Tier 2 date notice has been published on ssc.gov.in | Awaited |
| Dec 2026 | Tier 2, tentatively — no date announced by SSC as of 7 November 2026 | Tentative only |
Syllabus: The Complete Table
The notice lists the syllabus topic by topic but gives no topic-wise marks. The weightage column is therefore this note's guidance — not the notice's. The shares add up to 100%.
| Topic | What it covers | Guidance weightage |
|---|---|---|
| Collection, classification and presentation of data | Primary and secondary data, classification, frequency tables, histograms, bar diagrams, pie charts | 5% |
| Measures of central tendency | Mean, median, mode, quartiles, deciles, percentiles | 10% |
| Measures of dispersion | Range, quartile deviation, mean deviation, standard deviation, relative dispersion | 10% |
| Moments, skewness and kurtosis | Raw and central moments, measures of skewness, kurtosis and what they say about a distribution's shape | 8% |
| Correlation and regression | Karl Pearson's coefficient, Spearman's rank correlation, regression lines, multiple and partial correlation for three variables | 12% |
| Probability | Conditional probability, compound probability, Bayes' theorem | 10% |
| Random variables and probability distributions | Binomial, Poisson, Normal and Exponential distributions, their means and variances | 12% |
| Sampling theory | Population and sample, sampling and non-sampling errors, standard error | 6% |
| Statistical inference | Estimation, Z test, t test, chi-square test, F test | 10% |
| Analysis of variance | One-way and two-way ANOVA, F-ratio | 6% |
| Time series | Components, trend fitting, moving averages, seasonal variation | 6% |
| Index numbers | Price and quantity indices, Laspeyres, Paasche and Fisher, base shifting | 5% |
Central tendency, dispersion, correlation and regression, probability and distributions are about 54% of the paper — the core to master first. Inference, ANOVA, time series and index numbers are the second layer: smaller, but cheap marks.
Who Must Take Paper 2
Paper 2 is tied to the post you applied for, not your degree. You take it only if you applied for Junior Statistical Officer (Ministry of Statistics and Programme Implementation) or Statistical Investigator Grade-II (Registrar General of India). Any other post means no Paper 2, and merit is decided by Paper 1 alone (390 marks).
The qualifying bar applies to Paper 2 itself: falling below it ends your run. The Computer Knowledge Test and Data Entry Speed Test are different — qualifying only. Paper 2 adds everything it scores. Ranking is separate: JSO and Statistical Investigator candidates out of 590, others out of 390.
Exam Pattern
| Item | Detail |
|---|---|
| Questions | 100 |
| Marks | 200 |
| Time | 2 hours (2 hours 40 minutes for scribe-eligible candidates) |
| Marks per correct answer | +2 |
| Marks per wrong answer | −0.50 |
| Qualifying marks | UR 30% (60/200); OBC/EWS 25% (50/200); all other categories 20% (40/200) |
| Who takes it | JSO and Statistical Investigator Grade-II candidates only |
| Merit | Paper 2 marks added to Paper 1 merit (390), total 590 for these posts |
| Mode | Computer-based test |
Two hours for 100 questions is 72 seconds per question — generous against Tier 1's 36. Attempt all 100 with working on rough paper; keep the last 10 minutes for re-checking flagged arithmetic.
Negative marking is much kinder than Paper 1's −1 with +3. A blind guess has an expected value of 0.25 × 2 − 0.75 × 0.50 = +0.125 marks, and a guess after eliminating one option is worth about +0.33 marks. Attempt a question if you can eliminate one option.
The Syllabus, Topic by Topic
Collection, classification and presentation
Primary data is collected first-hand; secondary data comes from published sources. Classification groups data into classes; presentation turns it into tables and diagrams.
Measures of central tendency
Mean, median, mode, quartiles, deciles, percentiles. For ungrouped data: mean = sum ÷ count, median = middle value, mode = most frequent. Relation: mode = 3 × median − 2 × mean.
Worked example. Mean, median, mode of 10, 12, 14, 14, 18, 20, 24. Sum = 112, so mean = 16. Middle value = 14 (median); 14 appears twice (mode).
Measures of dispersion
Range = highest − lowest. Quartile deviation = (Q3 − Q1) ÷ 2. Mean deviation = average of absolute deviations. SD = √variance. Coefficient of variation = SD ÷ mean × 100.
Worked example. SD of 2, 4, 4, 4, 5, 5, 7, 9. Mean = 5. Deviations: −3, −1, −1, −1, 0, 0, 2, 4; squares sum to 32. Variance = 4, so SD = 2. SSC divides by n (population form), not n − 1.
Moments, skewness and kurtosis
Moments are averages of powers of deviations from the mean: the first central moment is zero, the second is the variance. Skewness: a long right tail means positive skew (mean > median > mode). Kurtosis: leptokurtic, mesokurtic, platykurtic; normal: skewness 0, kurtosis β2 = 3.
Correlation and regression
Correlation measures the strength and direction of a linear relationship, from −1 to +1. Pearson's works on values; Spearman's on ranks. Regression fits a line (y on x predicts y, x on y predicts x); both cross at (x̄, ȳ). Multiple correlation predicts one variable from two others; partial correlation holds the third.
Worked example. For x = 1, 2, 3, 4, 5 and y = 2, 3, 5, 4, 6: means x̄ = 3, ȳ = 4. Deviation products sum to 9; squared deviations sum to 10. So r = 9 ÷ 10 = 0.9. Slope = 0.9, intercept = 1.3, so ŷ = 1.3 + 0.9x; at x = 6, ŷ = 6.7.
Probability
Conditional probability: P(A|B) = P(A and B) ÷ P(B). Compound probability: independent events multiply, mutually exclusive events add. Bayes' theorem reverses a conditional probability. The trap: a 99%-accurate test can be wrong most of the time when the condition is rare.
Worked example. Probability that two fair dice add up to 7. There are 36 outcomes; the pairs adding to 7 are (1,6), (2,5), (3,4), (4,3), (5,2), (6,1) — 6 outcomes. So the probability is 6 ÷ 36 = 1 ÷ 6, about 16.7%.
Random variables and distributions
Binomial counts successes in fixed trials: mean np, variance np(1 − p). Poisson counts rare events: mean and variance both λ. Normal is the bell curve (mean μ, SD σ), 68–95–99.7 rule. Exponential models waiting times (mean 1 ÷ λ).
Sampling theory
A sample statistic estimates the population parameter; the standard error measures how much that estimate would vary across samples. Standard error of the mean = σ ÷ √n, so quadrupling the sample halves it. Sampling errors shrink with bigger samples; non-sampling errors do not.
Statistical inference
Point estimation gives one best value; interval estimation gives a range with a confidence level. The toolkit: Z test (large samples, known SD), t test (small samples), chi-square test (goodness of fit, independence), F test (comparing variances). Compute the test statistic, compare it with the critical value (usually 5%), and reject the null hypothesis if it is larger. Learn which test fits first.
Analysis of variance
ANOVA tests whether three or more group means differ, splitting total variation into between-group and within-group parts and comparing them with an F-ratio. A large F means the groups differ. One-way has one factor; two-way has two.
Time series
Data over time, split into trend, seasonal, cyclical and irregular variation. Trend is found by moving averages or a least-squares line; seasonal variation by a seasonal index.
Index numbers
Index numbers measure change relative to a base year, set to 100. ₹120 rising to ₹150 gives an index of 125. Laspeyres' uses base-year quantities, Paasche's current-year quantities, Fisher's is their geometric mean — the "ideal" index.
Common Mistakes
- Dividing by n − 1 instead of n. For a complete data set, SD divides by n — by n − 1 only for a "sample".
- Mixing up the regression lines. y-on-x predicts y (slope r × σy ÷ σx); x-on-y (slope r × σx ÷ σy).
- Reading correlation as causation. A correlation of 0.9 does not prove causation.
- Forgetting Paper 2's own qualifying bar. 150 in Paper 1 does not rescue a Paper 2 score below 60 (UR).
- Treating the mode as always unique. A data set can have two modes, or none.
- Using the wrong distribution. Binomial: fixed trials. Poisson: rare events over an interval.
- Spending too long on one calculation. At 72 seconds a question, a five-minute chi-square is a bad trade.
Memory Tricks and Formula Sheet
- Empirical relation: Mode = 3 × Median − 2 × Mean.
- Adding a constant shifts mean, median, mode by that constant; range, variance, SD unchanged. Multiplying by k multiplies mean and SD by k, variance by k².
- Variance shortcut: variance = (Σx² ÷ n) − mean².
- Correlation bounds: r lies between −1 and +1 — if you get 1.4, recheck.
- Regression lines cross at the means: both lines pass through (x̄, ȳ).
- 68–95–99.7: normal: 68% within μ ± σ, 95% within μ ± 2σ, 99.7% within μ ± 3σ.
- Binomial vs Poisson: Binomial: mean np, variance np(1 − p). Poisson: mean and variance both λ.
- Fisher is the geometric mean: Fisher's index = √(Laspeyres × Paasche).
- Bayes in one line: P(disease | positive test) = true positives ÷ all positives.
- Chi-square degrees of freedom: r × c table: (r − 1) × (c − 1); goodness of fit with k categories: k − 1.
8-Week Preparation Plan
A statistics graduate needs exam practice; a non-statistics graduate must learn the ideas first. Both tracks share one rule: Paper 1 first, Paper 2 second — Paper 1 is 390 of your 590 merit marks.
Non-statistics graduate (about 3 hours a day)
- Week 1–2: the core. Central tendency, dispersion, moments — work formulas by hand.
- Week 3–4: relationships and chance. Correlation, regression, probability, Bayes' theorem — examples, then 50 fresh questions per block.
- Week 5: distributions and sampling. Binomial, Poisson, Normal, Exponential, standard error — memorise each mean and variance, and the 68–95–99.7 rule.
- Week 6: inference and ANOVA. Z, t, chi-square, F tests, then one-way and two-way ANOVA — focus on which test fits which situation.
- Week 7: time series and index numbers. Short topics, quick marks. Finish the syllabus table, then revise the formula sheet.
- Week 8: mocks and revision. One full Paper 2 mock every two days, timed. Next day, re-solve only wrong questions.
Statistics graduate (about 1.5 hours a day)
- Week 1: diagnostic and core revision. Take a full Paper 2 mock cold, list lost-mark topics, then revise the core topics.
- Week 2–3: the weak blocks. Probability, distributions, inference — where graduates lose marks from careless test choices.
- Week 4: the small topics. Time series, index numbers, ANOVA, sampling theory — quick to revise, about 23%.
- Week 5–6: timed mocks. Two mocks a week, each followed by a full error review. Practise the 72-second pace.
- Week 7: formula sheet from memory. Write the sheet from memory, then check it against the formulas note.
- Week 8: light revision and rest. One mock early in the week, then revise your error list only. No new topics in the last week.
For Paper 1, see the Tier 2 preparation strategy 2026 and the Computer Knowledge Test syllabus.
Practice Questions
Q1. Find the mean of 5, 8, 11, 14, 17. Answer: Sum = 55, so mean = 55 ÷ 5 = 11.
Q2. Find the median of 3, 7, 9, 12, 15, 18, 21. Answer: 7 sorted values: median is the 4th, 12.
Q3. Find the mode of 4, 6, 6, 8, 8, 8, 10. Answer: 8 appears three times — more than any other — so the mode is 8.
Q4. For 20, 28, 34, 40, 46, 52, 60, find the range and quartile deviation. Answer: Range = 60 − 20 = 40. Q1 = 28, Q3 = 52. Quartile deviation = (52 − 28) ÷ 2 = 12.
Q5. Find the SD of 6, 8, 10, 12, 14. Answer: Mean = 10. Deviations: −4, −2, 0, 2, 4; squares sum to 40. Variance = 8, SD ≈ 2.83.
Q6. For x = 1, 2, 3, 4 and y = 10, 8, 6, 4, find r. Answer: As x rises by 1, y falls by exactly 2: a perfect falling line, so correlation = −1.
Q7. x̄ = 10, ȳ = 20, regression coefficient of y on x = 1.5. Find the predicted y when x = 12. Answer: Intercept = 20 − 15 = 5, so ŷ = 5 + 1.5x; at x = 12, ŷ = 23.
Q8. A card is drawn from a pack of 52. What is the probability it is a face card? Answer: 12 face cards out of 52: probability = 12 ÷ 52 = 3 ÷ 13, about 23.1%.
Q9. A fair coin is tossed 4 times. What is P(exactly 2 heads)? Answer: Binomial, n = 4, p = 0.5. Ways to choose 2 heads: 4C2 = 6; each sequence: 1 ÷ 16. Probability = 6 ÷ 16 = 0.375.
Q10. 1% of a population has a disease. A test detects it in 99% of cases, with a 5% false-positive rate. A person tests positive. What is the probability they have the disease? Answer: Out of 10,000: 100 have it, 9,900 do not. True positives: 99; false positives: 495; total: 594. Probability = 99 ÷ 594 = 1 ÷ 6, about 16.7%.
Frequently Asked Questions
Who has to take Paper 2?
Only JSO and Statistical Investigator Grade-II applicants. Candidates for all other CGL posts do not take Paper 2.
What is the Paper 2 pattern?
100 questions, 200 marks, 2 hours; +2 correct, −0.50 wrong. Scribes get 2 hours 40 minutes. Qualifying: 30% UR, 25% OBC/EWS, 20% others.
What are the qualifying marks?
60/200 UR, 50/200 OBC/EWS, 40/200 others. These are the minimums; the actual selection cut-off will be higher.
Is Paper 2 tough for non-maths students?
Demanding but learnable: the syllabus is fixed and formula-driven, with no surprise topics. This note's 8-week plan, worked by hand, covers the whole syllabus. Budget extra time for probability and statistical inference.
What is the Paper 2 syllabus?
Collection and presentation of data; central tendency; dispersion; moments, skewness, kurtosis; correlation, regression; probability; Binomial, Poisson, Normal, Exponential distributions; sampling theory; statistical inference (Z, t, chi-square, F); ANOVA; time series; index numbers. Full table in the syllabus section above.
When is SSC CGL Tier 2 2026?
Tentatively December 2026; no date announced as of 7 November 2026.
Do Tier 1 marks count in the final merit?
No. Tier 1 only shortlists for Tier 2. Final merit: Paper 1 (390) + Paper 2 (200) = 590.
Practice Next
- Pattern and syllabus: SSC CGL syllabus 2026, Tier 2 exam pattern 2026, SSC CGL exam pattern 2026, Tier 2 preparation strategy 2026
- Formulas and Computer test: Statistics formulas, Computer Knowledge Test syllabus
- Admit card and results: SSC CGL admit card 2026, SSC CGL answer key and result 2026
- Paper 2 mocks: Statistics Paper 2 mock 01, Statistics Paper 2 mock 02, Statistics Paper 2 mock 03, Statistics mock 01
- Paper 1 mock: Tier 2 Paper 1 full mock 18
Sources: SSC CGL 2026 notice (ssc.gov.in, 21 May 2026), Para 13 and Para 16, for the Paper 2 scheme, syllabus, qualifying marks and merit; vacancy revision and Tier 1 key per the house SSC CGL syllabus 2026 and Tier 2 exam pattern 2026; topic weightage is this note's guidance. As of 7 November 2026.