LAISHA
Open Laisha

CUET-PG Bayesian Inference: practice questions

CUET-PG - Statistics · India · National Testing Agency

Original practice questions written to the published syllabus · not past papers
10worked here, free
59written to this exam in this subject
11syllabus units

About CUET-PG

CUET-PG - Statistics is a postgraduate admission exam run by National Testing Agency, for candidates in India. Computer-based MCQ test used by central and participating universities for MSc Statistics admission.

Official site: exams.nta.ac.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.

Bayesian Inference in CUET-PG

Inference through prior and posterior: conjugate analysis, prior choice, credible intervals, Bayes factors and decision theory, hierarchical models, and computation by Markov-chain Monte Carlo, with agricultural uses in variety trials, genomic prediction and small-area estimation.

Our 59 CUET-PG questions in this subject: 19 easy, 31 medium, 9 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 CUET-PG and are not shown on any other exam's page.

Question 1 · hard

When may an improper prior be used?

  1. Only for testing with Bayes factors
  2. Never; every prior must integrate to 1
  3. Whenever the likelihood is bounded
  4. When the resulting posterior is proper
Answer and working

D. When the resulting posterior is proper

Improper priors are acceptable for estimation if ∫ f(x | θ)π(θ)dθ < ∞. They are not acceptable under the alternative in Bayes factors, where the arbitrary constant does not cancel.

Question 2 · easy

A beta(6, 14) prior for a disease incidence summarises earlier surveys. What is its prior effective sample size?

  1. 14
  2. 6
  3. 20
  4. 8
Answer and working

C. 20

A beta(a, b) prior is worth about a + b = 20 earlier observations (6 'successes' and 14 'failures'), which tells readers how strongly it will pull a new sample of a given size.

Question 3 · easy

Under squared-error loss, which posterior summary is the Bayes estimate?

  1. The posterior mode
  2. The MLE
  3. The posterior mean
  4. The posterior median
Answer and working

C. The posterior mean

Squared error is minimised by the mean of the posterior; absolute error by the median; 0-1 loss by the mode. The MLE is not a Bayes estimate in general.

Question 4 · medium

When do Bayesian credible intervals and frequentist confidence intervals nearly coincide numerically?

  1. In small samples with informative priors
  2. Never, because they answer different questions
  3. In large samples with diffuse priors
  4. Only when the posterior is bimodal
Answer and working

C. In large samples with diffuse priors

Under Bernstein-von Mises conditions the posterior is close to N(θ̂, 1/[nI]), so the credible interval matches the Wald interval; informative priors or small samples make them differ.

Question 5 · medium

Two trials both report a mean of 6.2 t/ha, one from 3 plots and one from 30. In a hierarchical model with overall mean 5.0, which statement is correct?

  1. The 30-plot trial is shrunk more towards 5.0
  2. Both are shrunk by the same amount
  3. Neither is shrunk, since the means are equal
  4. The 3-plot trial is shrunk more towards 5.0
Answer and working

D. The 3-plot trial is shrunk more towards 5.0

Shrinkage depends on each estimate's precision: the noisier 3-plot mean has the larger σᵢ² and hence the larger Bᵢ.

Question 6 · easy

What does each step of a Gibbs sampler do?

  1. Draws all parameters independently from their priors
  2. Maximises the posterior with respect to one parameter at a time
  3. Proposes all parameters jointly and accepts with probability 0.234
  4. Draws each parameter from its full conditional given the others
Answer and working

D. Draws each parameter from its full conditional given the others

Gibbs sampling cycles through full conditionals π(θⱼ | θ₋ⱼ, x). Coordinate-wise maximisation is an optimisation method, not a sampler.

Question 7 · medium

A model has log pointwise predictive density lppd = −120 and p_WAIC = 6. What is WAIC on the deviance scale?

  1. 252
  2. 240
  3. 126
  4. 228
Answer and working

A. 252

WAIC = −2(lppd − p_WAIC) = −2(−126) = 252. Omitting the penalty gives 240; adding it with the wrong sign gives 228.

Question 8 · easy

In comparing models by DIC or WAIC, which model is preferred?

  1. The one with more parameters
  2. The one with the smaller value
  3. The one with the larger value
  4. The one with p_D closest to zero
Answer and working

B. The one with the smaller value

Both estimate expected out-of-sample deviance, so smaller is better. The penalty already accounts for complexity.

Question 9 · medium

A Poisson model for aphid counts produces replicated data sets with far fewer zeros than observed. What does this posterior predictive check indicate?

  1. The prior on the Poisson mean is too narrow and should be widened
  2. The model fits well, because the mean count is reproduced
  3. The MCMC chain has not converged and should be run longer
  4. The model misses excess zeros
Answer and working

D. The model misses excess zeros

A test statistic (proportion of zeros) far outside its replicated distribution shows a systematic misfit of the model's form; a wider prior or longer run does not fix that.

Question 10 · medium

Among many fields, 40 had exactly one pest colony and 20 had exactly two. Using Robbins' estimator E(θ | x) = (x + 1)f(x + 1)/f(x), what is the empirical Bayes estimate of the colony rate for a field with one colony?

  1. 0.5
  2. 2.0
  3. 1.0
  4. 1.5
Answer and working

C. 1.0

(1 + 1) × 20/40 = 1.0, using observed frequencies in place of the marginal probabilities. No prior form is assumed.

A study plan for Bayesian Inference in CUET-PG

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 (46% of the 59 questions): Convergence diagnostics (7, 12%); Posterior predictive checks and model comparison (DIC, WAIC) (7, 12%); Posterior summaries and credible intervals (7, 12%); Bayes' theorem for parameters: prior, likelihood and posterior (6, 10%).

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

  2. Then the middle band: Markov chain Monte Carlo: Gibbs sampling and Metropolis-Hastings (6, 10%); Non-informative and weakly informative priors (6, 10%); Bayesian hypothesis testing and Bayes factors (5, 8%); Conjugate families (4, 7%).

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

  3. Then the lighter units: Empirical Bayes (4, 7%); Hierarchical (multilevel) models (4, 7%); Decision theory: loss functions and Bayes risk (3, 5%).

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

  4. Leave time for the hard ones: Markov chain Monte Carlo: Gibbs sampling and Metropolis-Hastings (2 hard of 6), Non-informative and weakly informative priors (2 hard of 6), Posterior predictive checks and model comparison (DIC, WAIC) (2 hard of 7).

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

  5. Bank the easy marks: Non-informative and weakly informative priors (3 easy of 6); Conjugate families (2 easy of 4).

    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 — CUET-PG Bayesian Inference

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.