LAISHA
Open Laisha

CSIR-UGC NET Multivariate Analysis: practice questions

CSIR-UGC NET - Mathematical Sciences (statistics option) · India · National Testing Agency for CSIR

Original practice questions written to the published syllabus · not past papers
10worked here, free
39written to this exam in this subject
12syllabus units

About CSIR-UGC NET

CSIR-UGC NET - Mathematical Sciences (statistics option) is a PhD, fellowship and lectureship exam run by National Testing Agency for CSIR, for candidates in India. Computer-based, 3 hours, Parts A (general aptitude), B and C (subject); statisticians answer from the statistics unit.

Official site: csirnet.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 CSIR-UGC NET

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 39 CSIR-UGC NET questions in this subject: 0 easy, 18 medium, 21 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 CSIR-UGC NET and are not shown on any other exam's page.

Question 1 · medium

Which condition is needed for the sample dispersion matrix S of p variables to be non-singular (with probability one, under normality)?

  1. n > p
  2. n = p
  3. n > 2p
  4. n ≥ 2
Answer and working

A. n > p

S has rank at most n − 1, so n − 1 ≥ p, i.e. n > p, is necessary; under normality it is then non-singular with probability one. In practice n should be several times p.

Question 2 · medium

For a p-variate normal sample, which pair is sufficient and complete for (μ, Σ)?

  1. (x̄, S)
  2. (x̄, tr S)
  3. x̄ alone
  4. S alone
Answer and working

A. (x̄, S)

The likelihood depends on the data only through x̄ and the SSCP matrix, and the full-rank exponential-family structure makes (x̄, S) complete. tr S discards the covariances.

Question 3 · hard

W ~ Wₚ(m, Σ) with m ≥ p. How is |W|/|Σ| distributed?

  1. As F(p, m − p + 1)
  2. As a product of p independent χ²_m variables
  3. As a product of independent χ²_m, χ²_(m−1), ..., χ²_(m−p+1)
  4. As χ²_(mp)
Answer and working

C. As a product of independent χ²_m, χ²_(m−1), ..., χ²_(m−p+1)

Bartlett's decomposition writes W in terms of a triangular matrix with independent chi entries on the diagonal having m, m − 1, ..., m − p + 1 df, so the determinant is their product.

Question 4 · hard

When are Wilks' Λ, Pillai's trace, the Hotelling-Lawley trace and Roy's largest root equivalent tests?

  1. When W⁻¹B has only one non-zero eigenvalue
  2. When the variables are uncorrelated
  3. When the sample sizes are equal
  4. When N exceeds 30
Answer and working

A. When W⁻¹B has only one non-zero eigenvalue

With a single non-zero eigenvalue (g = 2 or p = 1) all four are monotone functions of it and give the same test. Otherwise they weight the eigenvalues differently.

Question 5 · hard

A maximum-likelihood factor analysis of 6 variables with 2 factors is tested for adequacy. How many degrees of freedom does the test have?

  1. 4
  2. 9
  3. 12
  4. 21
Answer and working

A. 4

df = [(p − m)² − (p + m)]/2 = (16 − 8)/2 = 4: the number of distinct covariances, 21, minus the free parameters in L and Ψ after allowing for rotation. 9 subtracts only the loadings and forgets Ψ and the rotation constraint.

Question 6 · hard

A factor analysis reports an estimated uniqueness of −0.05 for one variable. What is this called and what does it suggest?

  1. A singular S, signalling n < p
  2. An oblique solution, signalling correlated factors
  3. A Heywood case
  4. A varimax solution, signalling a successful rotation
Answer and working

C. A Heywood case

Variances cannot be negative; a zero or negative uniqueness (Heywood case) indicates an improper solution, from over-factoring, small samples or near-collinear variables.

Question 7 · medium

With g groups and p variables, at most how many canonical discriminant functions are there?

  1. min(p, g − 1)
  2. p
  3. p(g − 1)
  4. g
Answer and working

A. min(p, g − 1)

The discriminant functions are eigenvectors of W⁻¹B, whose rank is at most min(p, g − 1).

Question 8 · hard

For the Fisher discriminant score y = a′x with a = S_p⁻¹(x̄₁ − x̄₂), what is the difference between the two group means of y?

  1. 1, after standardisation
  2. The trace of S_p
  3. D, the square root of D²
  4. D²
Answer and working

D. D²

a′(x̄₁ − x̄₂) = (x̄₁ − x̄₂)′S_p⁻¹(x̄₁ − x̄₂) = D², which is also the within-group variance a′S_pa of the score.

Question 9 · hard

Which approach gives a likelihood-based choice of the number of clusters and membership probabilities?

  1. Classical multidimensional scaling
  2. Normal-mixture model-based clustering
  3. Single-linkage clustering
  4. k-means with one random start
Answer and working

B. Normal-mixture model-based clustering

Fitting mixtures of multivariate normals by EM allows comparison of numbers of components by BIC and gives posterior membership probabilities.

Question 10 · hard

Which property makes D² preferable to Euclidean distance for genetic divergence based on correlated traits?

  1. It is always smaller than the Euclidean distance
  2. It is unaffected by environmental variation
  3. It does not need an estimate of the dispersion matrix
  4. It allows for correlations and units
Answer and working

D. It allows for correlations and units

D² uses the inverse pooled dispersion matrix, so units and correlations are accounted for. It needs S, and phenotypic D² does include environmental effects.

A study plan for Multivariate Analysis in CSIR-UGC 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 (46% of the 39 questions): Canonical correlation analysis (5, 13%); Wishart distribution (5, 13%); Discriminant analysis and classification (4, 10%); Factor analysis (4, 10%).

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

  2. Then the middle band: Multivariate normal distribution: properties, marginals and conditionals (4, 10%); Estimation of the mean vector and dispersion matrix (3, 8%); Multidimensional scaling and correspondence analysis (3, 8%); Multivariate analysis of variance (MANOVA) (3, 8%).

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

  3. Then the lighter units: Cluster analysis: hierarchical and k-means (2, 5%); Hotelling's T-squared and its applications (2, 5%); Mahalanobis D-squared and genetic divergence (2, 5%); Principal component analysis (2, 5%).

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

  4. Leave time for the hard ones: Hotelling's T-squared and its applications (2 hard of 2), Mahalanobis D-squared and genetic divergence (2 hard of 2), Principal component analysis (2 hard of 2).

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

  5. 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 — CSIR-UGC NET 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.