LAISHA
Open Laisha

CUET-PG Multivariate Analysis: practice questions

CUET-PG - Statistics · India · National Testing Agency

Original practice questions written to the published syllabus · not past papers
10worked here, free
60written to this exam in this subject
12syllabus 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.

Multivariate Analysis in CUET-PG

Analysing many correlated variables together: the multivariate normal and its estimation, Wishart and Hotelling's T-squared, MANOVA, and the reduction and classification tools agricultural scientists use daily (principal components, factor analysis, discriminant and cluster analysis, Mahalanobis D-squared).

Our 60 CUET-PG questions in this subject: 21 easy, 37 medium, 2 hard, across 12 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 · easy

Under multivariate normality, two sub-vectors X₁ and X₂ are independent if and only if:

  1. μ₁ = μ₂
  2. Σ₁₂ = 0
  3. |Σ| = 1
  4. Σ₁₁ = Σ₂₂
Answer and working

B. Σ₁₂ = 0

For the multivariate normal, zero cross-covariance is equivalent to independence. Equal means or equal dispersion blocks say nothing about dependence.

Question 2 · medium

Some multivariate records have missing values. Why is pairwise deletion risky when estimating Σ?

  1. Positive definiteness is not guaranteed
  2. It makes x̄ and S dependent
  3. It discards all records with any missing value
  4. It always overestimates every variance
Answer and working

A. Positive definiteness is not guaranteed

Each covariance is computed on a different subset, so the assembled matrix may have negative eigenvalues; EM or multiple imputation give coherent estimates. Discarding every incomplete record is listwise deletion.

Question 3 · easy

To which distribution does the Wishart Wₚ(m, Σ) reduce when p = 1?

  1. Normal
  2. σ²χ²_m
  3. F(m, 1)
  4. Student's t with m df
Answer and working

B. σ²χ²_m

For p = 1, W = Σzᵢ² with zᵢ ~ N(0, σ²), so W/σ² ~ χ²_m.

Question 4 · medium

From n observations of Nₚ(μ, Σ), what is the distribution of the corrected SSCP matrix A = (n − 1)S?

  1. Wₚ(n, Σ)
  2. χ²ₙ₋₁ in each element
  3. Wₚ(n − p, Σ)
  4. Wₚ(n − 1, Σ)
Answer and working

D. Wₚ(n − 1, Σ)

One degree of freedom is lost to estimating μ by x̄, as in the univariate case. Off-diagonal elements of A are not chi-square.

Question 5 · medium

Why are factor loadings not unique?

  1. LT gives the same LL′ for any orthogonal T
  2. The specific variances can be chosen freely
  3. The factors can have any variance
  4. The number of factors is arbitrary
Answer and working

A. LT gives the same LL′ for any orthogonal T

(LT)(LT)′ = LL′ for orthogonal T, so the model fits equally well after any rotation; rotations such as varimax are chosen for interpretability.

Question 6 · medium

What does varimax rotation do?

  1. It maximises the communalities of all variables
  2. It allows the factors to become correlated
  3. It removes the specific factors from the model
  4. Orthogonal rotation towards simple structure
Answer and working

D. Orthogonal rotation towards simple structure

Varimax maximises the variance of squared loadings within each factor, giving simple structure while keeping factors uncorrelated; oblique rotations (promax, oblimin) allow correlation. Rotation does not change communalities.

Question 7 · medium

When the two populations have clearly different covariance matrices, which classification rule is optimal under normality?

  1. Fisher's linear discriminant
  2. Quadratic discriminant analysis
  3. Principal component analysis
  4. k-means clustering
Answer and working

B. Quadratic discriminant analysis

Unequal Σᵢ make the log density ratio quadratic in x, giving quadratic boundaries (QDA), at the cost of estimating more parameters.

Question 8 · medium

Why is the apparent error rate of a discriminant rule optimistic?

  1. It counts only one type of misclassification
  2. It uses the pooled instead of separate covariance matrices
  3. It ignores the prior probabilities of the groups
  4. The same units are used to build and to test the rule
Answer and working

D. The same units are used to build and to test the rule

Units that shaped the rule are classified more easily than new ones; holdout (Lachenbruch) or cross-validation gives honest error estimates.

Question 9 · medium

How is the significance of D² between two groups tested?

  1. By comparing D² with the average intra-cluster distance
  2. By a paired t-test on the trait means
  3. Through the two-sample Hotelling T² = [n₁n₂/(n₁ + n₂)]D²
  4. By referring D² directly to χ² (1 df)
Answer and working

C. Through the two-sample Hotelling T² = [n₁n₂/(n₁ + n₂)]D²

Scaling D² by n₁n₂/(n₁ + n₂) gives T², which has an exact F transformation under normality.

Question 10 · medium

A correspondence analysis has total inertia 0.12 and first two principal inertias 0.07 and 0.03. What proportion of inertia does the two-dimensional map show?

  1. 10.0%
  2. 0.83%
  3. 58.3%
  4. 83.3%
Answer and working

D. 83.3%

(0.07 + 0.03)/0.12 = 83.3%.

A study plan for Multivariate Analysis 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 (43% of the 60 questions): Cluster analysis: hierarchical and k-means (7, 12%); Estimation of the mean vector and dispersion matrix (7, 12%); Factor analysis (6, 10%); Multidimensional scaling and correspondence analysis (6, 10%).

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

  2. Then the middle band: Canonical correlation analysis (5, 8%); Discriminant analysis and classification (5, 8%); Wishart distribution (5, 8%); Mahalanobis D-squared and genetic divergence (4, 7%).

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

  3. Then the lighter units: Multivariate analysis of variance (MANOVA) (4, 7%); Multivariate normal distribution: properties, marginals and conditionals (4, 7%); Principal component analysis (4, 7%); Hotelling's T-squared and its applications (3, 5%).

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

  4. Leave time for the hard ones: Multivariate analysis of variance (MANOVA) (1 hard of 4), Discriminant analysis and classification (1 hard of 5).

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

  5. Bank the easy marks: Hotelling's T-squared and its applications (2 easy of 3); Multivariate normal distribution: properties, marginals and conditionals (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 Multivariate 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.