LAISHA
Open Laisha

ASRB NET Statistical Inference II: Testing of Hypotheses & Sequential Analysis: practice questions

ASRB NET - Agricultural Statistics (and Computer Application / Bioinformatics) · India · Agricultural Scientists Recruitment Board (ICAR)

Original practice questions written to the published syllabus · not past papers
10worked here, free
41written to this exam in this subject
11syllabus units

About ASRB NET

ASRB NET - Agricultural Statistics (and Computer Application / Bioinformatics) is a professional certification run by Agricultural Scientists Recruitment Board (ICAR), for candidates in India. NET: one objective paper of about 150 questions, 2 hours, negative marking, online, Master's level (pattern as reported for other disciplines).

Official site: asrb.org.in.

We have not been able to confirm these details against the exam body's own specification, so fees, dates and pass marks are left out on purpose. Check the current format, eligibility, fees and dates on the official site.

Statistical Inference II: Testing of Hypotheses & Sequential Analysis in ASRB NET

Choosing and interpreting tests for trial and survey data, multiple comparisons, sequential sampling for pests, and replication from power and CV.

The logic and optimality of tests: errors and power, the Neyman-Pearson lemma, uniformly most powerful and unbiased tests, likelihood-ratio tests and their large-sample forms, and Wald's sequential probability ratio test.

Our 41 ASRB NET questions in this subject: 11 easy, 20 medium, 10 hard, across 11 units.

10 practice questions, worked

Choose an answer, then open the working. Each explanation says why the right answer is right and why the others are not.

These 10 were written to ASRB NET and are not shown on any other exam's page.

Question 1 · hard

Plot yields are N(μ, 16). From 16 plots, H₀: μ = 50 is rejected in favour of H₁: μ = 52 when x̄ > 51.645. What is the power of the test?

  1. 0.639
  2. 0.361
  3. 0.050
  4. 0.535
Answer and working

A. 0.639

SE = 4/√16 = 1. Power = P(X̄ > 51.645 | μ = 52) = P(Z > −0.355) = 0.639. 0.050 is the size; 0.361 is β; using σ = 4 instead of the standard error gives 0.535.

Question 2 · medium

Pump lifetimes are exponential with rate λ. For H₀: λ = λ₀ against H₁: λ = λ₁ with λ₁ > λ₀ (shorter lives under H₁), the most powerful test rejects H₀ when:

  1. max Xᵢ is large
  2. ΣXᵢ is small
  3. min Xᵢ is large
  4. ΣXᵢ is large
Answer and working

B. ΣXᵢ is small

Λ = (λ₁/λ₀)ⁿ exp[−(λ₁ − λ₀)ΣXᵢ] decreases in ΣXᵢ when λ₁ > λ₀, so large Λ means small total lifetime. A higher failure rate shows up as short lives.

Question 3 · hard

Two nested generalised linear models for pest counts differ by 3 parameters. Under the smaller model, what is the approximate distribution of the difference in deviance?

  1. F with (3, n) degrees of freedom exactly
  2. Chi-square with 1 degree of freedom
  3. Chi-square with 3 df
  4. Standard normal, after dividing by 3
Answer and working

C. Chi-square with 3 df

The deviance difference is −2 log Λ for the nested comparison and, by Wilks' theorem, is approximately χ² with df equal to the number of extra parameters. With over-dispersion an F-test with estimated dispersion is used instead.

Question 4 · medium

A one-sided level-α test of H₀: μ ≤ μ₀ against μ > μ₀ corresponds to which confidence statement?

  1. A two-sided 1 − α interval for μ: reject H₀ when μ₀ is outside it
  2. A lower 1 − α bound for μ: reject H₀ when μ₀ is below it
  3. An upper 1 − α confidence bound for μ: reject H₀ when μ₀ is below it
  4. A two-sided 1 − 2α interval for μ: reject H₀ when μ₀ is inside it
Answer and working

B. A lower 1 − α bound for μ: reject H₀ when μ₀ is below it

Rejecting for large x̄ corresponds to the lower bound x̄ − z_α σ/√n; H₀ is rejected when μ₀ falls below that bound.

Question 5 · easy

Using Wald's approximations, what are the SPRT boundaries A and B for α = 0.05 and β = 0.10?

  1. A = 9.5, B = 0.0556
  2. A = 9, B = 0.0526
  3. A = 18, B = 0.105
  4. A = 20, B = 0.100
Answer and working

C. A = 18, B = 0.105

A ≈ (1 − β)/α = 0.90/0.05 = 18; B ≈ β/(1 − α) = 0.10/0.95 = 0.105. Swapping α and β gives A = 9.5 and B = 0.0556; A = 9 and B = 0.0526 put the wrong error probability in each denominator.

Question 6 · medium

For an SPRT designed with error probabilities α and β, what are the approximate values of the OC function L(θ) = P(accept H₀ | θ) at θ₀ and θ₁?

  1. L(θ₀) ≈ β and L(θ₁) ≈ 1 − α
  2. L(θ₀) ≈ 1 − α and L(θ₁) ≈ β
  3. L(θ₀) ≈ α and L(θ₁) ≈ 1 − β
  4. L(θ₀) ≈ 1 − β and L(θ₁) ≈ α
Answer and working

B. L(θ₀) ≈ 1 − α and L(θ₁) ≈ β

Accepting H₀ when it is true has probability 1 − α; accepting it when H₁ is true is a Type II error, probability β.

Question 7 · hard

A sequential plan for pest scouting tests p₀ = 0.10 against p₁ = 0.30 infested plants. In the (plants inspected n, infested d) plane, both SPRT boundary lines have the same slope. What is it?

  1. 0.251 infested plants per plant
  2. 0.200 infested plants per plant
  3. 1.099 infested plants per plant
  4. 0.186 infested plants per plant
Answer and working

D. 0.186 infested plants per plant

The boundaries are d = h + sn with s = log[(1 − p₀)/(1 − p₁)]/{log(p₁/p₀) + log[(1 − p₀)/(1 − p₁)]} = 0.2513/1.3499 = 0.186. The midpoint 0.200 is a common wrong guess; 1.099 = log 3 is only the step per infested plant.

Question 8 · medium

Observations are N(μ, 1); an SPRT tests μ₀ = 0 against μ₁ = 1 with log A = 2.890 and log B = −2.251. After 4 observations, Σxᵢ = 5. What is the decision?

  1. Continue sampling, since Σzᵢ = Σxᵢ − n = 1 lies between the boundaries
  2. Continue sampling, since Σxᵢ/n = 1.25 lies between the boundaries
  3. Accept H₀, since Σzᵢ = Σxᵢ − n = 1 is below 2.890
  4. Reject H₀: Σzᵢ = Σxᵢ − n/2 = 3 > 2.890
Answer and working

D. Reject H₀: Σzᵢ = Σxᵢ − n/2 = 3 > 2.890

zᵢ = log[f₁(xᵢ)/f₀(xᵢ)] = xᵢ − 1/2, so Σzᵢ = 5 − 2 = 3 > log A = 2.890: reject H₀. Subtracting n instead of n/2 uses the wrong log-likelihood ratio.

Question 9 · medium

Which statement about the SPRT's sample size N is correct?

  1. N is fixed in advance from α and β, exactly like a fixed-sample test of the same size
  2. N never exceeds the fixed-sample size for the same α and β
  3. N is always smaller under H₁ than under H₀
  4. N is finite with probability one but unbounded; truncation imposes a maximum
Answer and working

D. N is finite with probability one but unbounded; truncation imposes a maximum

The SPRT terminates with probability one, but individual runs can be long; truncation caps N at some cost in error rates. Its E(N), not its maximum, is smaller than the fixed n.

Question 10 · medium

The standard deviation of yield differences between paired strips is 0.3 t/ha, while that of unpaired plots is 0.6 t/ha. For the same effect, α and power, how does the required sample size of the paired design compare with that of one group in the unpaired design?

  1. It is the same, because the effect size and error rates do not change with the design
  2. It is double, because each pair uses two plots instead of one
  3. It is one-half, because pairing halves the number of plots per group
  4. One-eighth: the factor 2 for two groups drops out and the variance falls to a quarter
Answer and working

D. One-eighth: the factor 2 for two groups drops out and the variance falls to a quarter

Unpaired, n per group = 2(z + z)²(0.36)/δ²; paired, the number of pairs = (z + z)²(0.09)/δ². The ratio is 0.09/(2 × 0.36) = 1/8. Reducing σ through design is often cheaper than adding replicates.

A study plan for Statistical Inference II: Testing of Hypotheses & Sequential Analysis in ASRB NET

This plan follows where our questions actually fall, so the time goes where the marks are likely to be. It is our reading of the syllabus, not the exam body's weighting.

  1. First, the heavy units (61% of the 41 questions): Sequential probability ratio test: OC and ASN functions (9, 22%); Multiple testing: family-wise error rate and false discovery rate (6, 15%); Likelihood ratio tests and Wilks' theorem (5, 12%); Power and sample-size calculation for common tests (5, 12%).

    Read the manual chapters for these first and do every worked question you can find on them.

  2. Then the middle band: Tests based on Z, t, chi-square and F distributions (4, 10%); Hypotheses, the two types of error, size and power (3, 7%); Monotone likelihood ratio and uniformly most powerful tests (2, 5%); Neyman-Pearson lemma and most powerful tests (2, 5%).

    Aim for one sitting per unit, then mixed questions across them.

  3. Then the lighter units: Relation between tests and confidence intervals (2, 5%); Unbiased tests and UMPU tests for exponential families (2, 5%); Wald and score (Rao) tests (1, 2%).

    Learn the definitions, classifications and standard numbers; these units are more often recalled than worked.

  4. Leave time for the hard ones: Wald and score (Rao) tests (1 hard of 1), Likelihood ratio tests and Wilks' theorem (3 hard of 5), Unbiased tests and UMPU tests for exponential families (1 hard of 2).

    Come back to these after a first pass, untimed, and read every explanation, including for the options you rejected.

  5. Bank the easy marks: Multiple testing: family-wise error rate and false discovery rate (4 easy of 6); Monotone likelihood ratio and uniformly most powerful tests (1 easy of 2).

    Do these early in revision so they are fast on the day.

  6. Finish timed. Sit the subject as a timed set without notes, mark it, and send each wrong answer back to its unit above.

Last updated · How Laisha writes and checks its pages · Report a mistake

Ask Laisha about this

Opens Laisha's chat with your question ready to send. Sign in or start free.

Unlock every question — ASRB NET Statistical Inference II: Testing of Hypotheses & Sequential Analysis

US$10 a month · one price worldwide · cancel any time

  • Every practice question in this subject, with full working
  • Timed exam papers and progress by unit
  • Ask Laisha to explain any answer — up to 1,000 questions a month
Free or Smarter
FreeSmarter
Grow guides, pest pages, articlesRead allRead all
Ask Laisha5 a dayUp to 1,000 a month
Exam practiceThe sample on each pageEvery question, timed papers
Crop calendar and pest alerts—For your own fields
Calculators and tools—All
Uploads and reports—100 a month, real files
10 days free for every new account, no card.