LAISHA
Open Laisha

ASRB Agricultural Research Service Categorical Data Analysis & Generalized Linear Models: practice questions

ASRB Agricultural Research Service (ARS) - Agricultural Statistics scientist · India · Agricultural Scientists Recruitment Board (ICAR)

Original practice questions written to the published syllabus · not past papers
10worked here, free
44written to this exam in this subject
11syllabus units

About ASRB Agricultural Research Service

ASRB Agricultural Research Service (ARS) - Agricultural Statistics scientist is a government recruitment exam run by Agricultural Scientists Recruitment Board (ICAR), for candidates in India. Combined NET/ARS/SMS/STO recruitment: preliminary (objective), mains (descriptive, discipline paper) and interview for ARS.

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.

Categorical Data Analysis & Generalized Linear Models in ASRB Agricultural Research Service

Models for counts, proportions and categories: contingency tables, the exponential-family GLM with its link functions, logistic and Poisson regression, log-linear models, overdispersion, ordinal outcomes and generalised linear mixed models, applied to germination, mortality, pest counts and adoption data.

Our 44 ASRB Agricultural Research Service questions in this subject: 7 easy, 30 medium, 7 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 Agricultural Research Service and are not shown on any other exam's page.

Question 1 · medium

A variety has a higher germination rate than another in each of two seed lots, but a lower rate when the lots are pooled. What is this called?

  1. Overdispersion
  2. Separation
  3. Simpson's paradox
  4. Lindley's paradox
Answer and working

C. Simpson's paradox

Pooling over a confounding variable (seed lot) that is associated with both variety and germination can reverse the association; stratified analysis (Mantel-Haenszel) avoids it.

Question 2 · medium

A germination model with 4 treatments has residual deviance 31.2 on 16 df; the null model has 52.8 on 19 df. What is the likelihood-ratio statistic for treatments and its df (dispersion fixed at 1)?

  1. 31.2 on 16 df
  2. 7.2 on 3 df
  3. 52.8 on 19 df
  4. 21.6 on 3 df
Answer and working

D. 21.6 on 3 df

The deviance difference 52.8 − 31.2 = 21.6 on 19 − 16 = 3 df, compared with χ²₃. Dividing by 3 gives a mean deviance, used in F tests only with an estimated dispersion.

Question 3 · medium

Trap A catches 42 thrips in 6 days; trap B catches 50 in 10 days. With log(days) as offset, what is the estimated log rate ratio A versus B?

  1. 2.000
  2. 0.336
  3. −0.174
  4. −0.511
Answer and working

B. 0.336

Rates are 7.0 and 5.0 per day, ratio 1.4, log 1.4 = 0.336. log(42/50) = −0.174 ignores exposure and wrongly favours trap B.

Question 4 · medium

After fitting a Poisson model to weed counts, the Pearson X² is 96 on 24 residual df. What does this indicate?

  1. Underdispersion, with φ̂ ≈ 0.25
  2. Separation in the predictors
  3. A good fit, since X² exceeds its df
  4. Overdispersion, with φ̂ ≈ 4
Answer and working

D. Overdispersion, with φ̂ ≈ 4

X²/df = 4 is far above 1: the counts vary about four times more than Poisson allows, so standard errors from the Poisson fit are about half what they should be.

Question 5 · medium

Lesion counts per leaf are mostly 0 to 3. Why is a Poisson (or negative binomial) GLM better than ANOVA on log(count + 1)?

  1. It needs no arbitrary added constant
  2. It needs no assumption about the counts' distribution
  3. It always gives a higher R²
  4. It removes the need to consider overdispersion
Answer and working

A. It needs no arbitrary added constant

With small counts, log(y + 1) depends on the arbitrary constant and does not stabilise variance well; the count GLM uses the right distribution and link. Overdispersion still needs checking.

Question 6 · medium

Farm type, crop choice and irrigation source are cross-classified in a survey, with none treated as the response. Which analysis describes their associations?

  1. A one-way ANOVA on cell counts
  2. A two-sample t-test for each pair
  3. A logistic regression of crop choice
  4. A log-linear model
Answer and working

D. A log-linear model

When variables are on an equal footing, log-linear models describe the association structure symmetrically; logistic regression singles out one response.

Question 7 · easy

In a proportional-odds model for disease grades, what does 'proportional odds' mean?

  1. The odds of each category are proportional to its rank
  2. The category probabilities are all equal
  3. Each cut-point has its own set of effects
  4. One set of effects β applies to every cumulative split
Answer and working

D. One set of effects β applies to every cumulative split

log[P(Y ≤ j)/P(Y > j)] = θⱼ − x′β: only the intercepts θⱼ vary with j. Separate effects per cut-point would be a non-proportional model.

Question 8 · hard

What assumption of the multinomial logit model can fail when two alternatives are close substitutes, such as two similar improved varieties?

  1. Independence of irrelevant alternatives
  2. Equidispersion
  3. Proportional odds
  4. Conditional independence
Answer and working

A. Independence of irrelevant alternatives

IIA says the odds between two options do not depend on other available options; with close substitutes this is unrealistic, and nested or mixed logit models are used.

Question 9 · hard

What do generalized estimating equations (GEE) estimate, compared with a GLMM?

  1. Exactly the same effects as a GLMM for any link
  2. Random-effect variances by maximum likelihood
  3. Population-averaged effects with sandwich SEs
  4. Cluster-specific effects, with model-based standard errors
Answer and working

C. Population-averaged effects with sandwich SEs

GEE targets marginal means using a working correlation and sandwich SEs; logistic GLMMs give conditional effects, which differ in magnitude.

Question 10 · medium

A binomial GLMM reports a block variance estimate of exactly zero. How should this be treated?

  1. Hidden by dropping the blocks from the report
  2. A singular fit, reported as such
  3. Treated as proof that blocks were identical
  4. Fixed by adding a constant to the variance
Answer and working

B. A singular fit, reported as such

Boundary estimates are common with few levels; they mean the data cannot distinguish block variation from zero, not that it is absent.

A study plan for Categorical Data Analysis & Generalized Linear Models in ASRB Agricultural Research Service

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 (50% of the 44 questions): Logistic regression for binary and binomial data (6, 14%); Zero-inflated and hurdle models (6, 14%); Generalized linear mixed models (5, 11%); Ordinal and multinomial logistic models (5, 11%).

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

  2. Then the middle band: Poisson regression for counts (5, 11%); Overdispersion: quasi-likelihood and negative binomial models (4, 9%); Contingency tables: association, odds ratios and tests (3, 7%); Exponential family and the generalized linear model (3, 7%).

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

  3. Then the lighter units: Link functions, deviance and iteratively reweighted least squares (3, 7%); Model checking: residuals and goodness of fit (3, 7%); Log-linear models for multi-way tables (1, 2%).

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

  4. Leave time for the hard ones: Logistic regression for binary and binomial data (2 hard of 6), Link functions, deviance and iteratively reweighted least squares (1 hard of 3), Overdispersion: quasi-likelihood and negative binomial models (1 hard of 4).

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

  5. Bank the easy marks: Overdispersion: quasi-likelihood and negative binomial models (2 easy of 4); Poisson regression for counts (2 easy of 5).

    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 Agricultural Research Service Categorical Data Analysis & Generalized Linear Models

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.