LAISHA
Open Laisha

Indian Statistical Service Examination Sampling Theory I: Simple, Stratified, Systematic, Ratio & Regression Estimation: practice questions

Indian Statistical Service Examination (with Indian Economic Service) · India · Union Public Service Commission

Original practice questions written to the published syllabus · not past papers
10worked here, free
25written to this exam in this subject
9syllabus units

About Indian Statistical Service Examination

Indian Statistical Service Examination (with Indian Economic Service) is a government recruitment exam run by Union Public Service Commission, for candidates in India. Written examination of six papers and a viva voce. Reported scheme: General English 100, General Studies 100, Statistics I and II (objective, 200 each), Statistics III and IV (descriptive, 200 each); viva 200.

Official site: upsc.gov.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.

Sampling Theory I: Simple, Stratified, Systematic, Ratio & Regression Estimation in Indian Statistical Service Examination

Probability sampling from first principles: sampling versus complete enumeration, simple random sampling with and without replacement, stratification and allocation, systematic sampling, and the ratio and regression estimators that use auxiliary information such as last year's area.

Our 25 Indian Statistical Service Examination questions in this subject: 0 easy, 14 medium, 11 hard, across 9 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 Indian Statistical Service Examination and are not shown on any other exam's page.

Question 1 · hard

As the sample size of a large survey is increased greatly with the same field staff, what typically happens to the total error?

  1. Both sampling and non-sampling errors usually fall in proportion to the sample size
  2. Sampling error rises, because more units mean more random variation
  3. Sampling error falls, but non-sampling error may rise and can dominate
  4. Total error is unchanged, because errors cancel in large samples
Answer and working

C. Sampling error falls, but non-sampling error may rise and can dominate

Sampling error shrinks with the square root of n, but spreading the same staff over more units weakens supervision and checking, so measurement and processing errors can grow and become the larger part of the MSE. Survey design aims at total error for the cost.

Question 2 · medium

Two districts have the same per-holding variability; one has 40,000 holdings and the other 4,000,000. An SRS of 400 holdings is drawn in each. How do the standard errors of the mean compare?

  1. Almost equal: precision depends mainly on n, not on N
  2. The smaller district's SE is 10 times the larger district's
  3. The larger district's SE is 10 times the smaller district's
  4. The larger district's SE is 100 times the smaller district's
Answer and working

A. Almost equal: precision depends mainly on n, not on N

SE = sqrt((1 - f) S^2 / n); with f = 1% and 0.01% the fpc is about 0.99 and about 1, so the SEs differ by about half a per cent. The sampling fraction matters little when it is small.

Question 3 · hard

Two strata have N_h = 400 each, S_h = 3 and 6, and cost per unit c_h = 1 and 4. With n_h proportional to N_h S_h / sqrt(c_h), how is a total sample of 60 split?

  1. 40 and 20
  2. 20 and 40
  3. 48 and 12
  4. 30 and 30
Answer and working

D. 30 and 30

N_h S_h / sqrt(c_h) = 400 x 3 / 1 = 1200 and 400 x 6 / 2 = 1200, so the sample is split equally: 30 and 30. Neyman allocation ignoring costs gives 20 and 40; dividing by c_h instead of its square root gives 48 and 12.

Question 4 · hard

Strata have weights 0.6, 0.3, 0.1 and standard deviations 2, 4, 10. For a total sample of 68, ignoring finite population corrections, what is the variance of the stratified mean under Neyman allocation?

  1. 0.2529
  2. 0.0500
  3. 0.0571
  4. 0.1700
Answer and working

D. 0.1700

V_opt = (sum W_h S_h)^2 / n = (3.4)^2 / 68 = 11.56 / 68 = 0.1700. Proportional allocation would give (sum W_h S_h^2)/n = 17.2/68 = 0.2529; omitting the square gives 0.0500.

Question 5 · hard

Holdings are classified by area in classes 0-10, 10-20, ..., 70-80 ha with frequencies 16, 25, 36, 49, 64, 100, 9, 4. Using the Dalenius-Hodges cumulative square-root-of-frequency rule for three strata, where are the stratum boundaries?

  1. 20 and 40 ha
  2. 27 and 53 ha
  3. 30 and 50 ha
  4. 40 and 60 ha
Answer and working

C. 30 and 50 ha

Square roots of the frequencies are 4, 5, 6, 7, 8, 10, 3, 2; cumulated: 4, 9, 15, 22, 30, 40, 43, 45, total 45. One third is 15: the cumulative reaches 15 at the end of class 20-30 and 30 at the end of 40-50, so boundaries fall at 30 and 50 ha. Cutting the cumulative frequencies (not square roots) into thirds gives about 40 and 60 ha; 27 and 53 ha divide the range 0-80 into equal widths.

Question 6 · medium

According to Cochran's conclusions on stratifying by one good auxiliary variable, what happens as the number of strata is increased beyond about six?

  1. The variance begins to increase sharply
  2. The estimator becomes biased
  3. Extra gain in precision becomes small
  4. The design becomes self-weighting
Answer and working

C. Extra gain in precision becomes small

The reduction in variance comes mostly from the first few strata; further splitting on the same variable adds little, so resources are better spent on another stratifying variable or a larger sample. More strata do not create bias.

Question 7 · hard

What is the bias of the collapsed-strata variance estimator?

  1. Upward only when the paired strata have exactly equal means
  2. Upward, as it picks up variation between the paired strata
  3. None, because pairing is done at random among all the strata
  4. Downward, because pairing removes part of the within-stratum variation
Answer and working

B. Upward, as it picks up variation between the paired strata

The estimator picks up differences between the means of the strata that were paired, so it overestimates the variance; the bias is small when paired strata have similar means and vanishes only if those means are equal.

Question 8 · medium

For two variables with C_x = 0.3 and C_y = 0.4, above what correlation is the ratio estimator more precise than the simple sample mean (large samples)?

  1. 1.500
  2. 0.750
  3. 0.667
  4. 0.375
Answer and working

D. 0.375

The condition is rho > C_x / (2 C_y) = 0.3 / (2 x 0.4) = 0.375. Swapping the CVs gives 0.667; omitting the 2 gives 0.750.

Question 9 · medium

When are the ratio and regression estimators equally efficient in large samples?

  1. When the population regression line has a large positive intercept
  2. When the population regression line passes through the origin
  3. When the correlation between y and x is exactly zero in the population
  4. When the coefficient of variation of x equals one in the population
Answer and working

B. When the population regression line passes through the origin

The regression estimator's variance is S_y^2(1 - rho^2)/n; the ratio estimator reaches it only when beta = R = Y-bar/X-bar, i.e. the line passes through the origin. With an intercept, regression is better.

Question 10 · medium

With C_x = 0.2 and C_y = 0.5, for which correlations is the product estimator more precise than the simple mean (large samples)?

  1. rho < -1.25
  2. rho > 0.20
  3. rho < -0.20
  4. rho < -0.40
Answer and working

C. rho < -0.20

The product estimator beats the mean when rho < -C_x/(2 C_y) = -0.2/(2 x 0.5) = -0.20. rho > 0.20 is the ratio estimator's condition; swapping the CVs gives -1.25.

A study plan for Sampling Theory I: Simple, Stratified, Systematic, Ratio & Regression Estimation in Indian Statistical Service Examination

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 (52% of the 25 questions): Simple random sampling with and without replacement (5, 20%); Stratified sampling: estimation and allocation (proportional, Neyman, optimum) (5, 20%); Choice of strata, construction of strata boundaries and collapsing of strata (3, 12%).

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

  2. Then the middle band: Systematic sampling and its variants (3, 12%); Product and difference estimators (2, 8%); Ratio estimators: bias, MSE and comparison (2, 8%).

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

  3. Then the lighter units: Regression estimators (2, 8%); Sampling versus complete enumeration; sampling and non-sampling errors (2, 8%); Estimation of proportions and sample-size determination (1, 4%).

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

  4. Leave time for the hard ones: Choice of strata, construction of strata boundaries and collapsing of strata (2 hard of 3), Systematic sampling and its variants (2 hard of 3), Stratified sampling: estimation and allocation (proportional, Neyman, optimum) (3 hard of 5).

    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 — Indian Statistical Service Examination Sampling Theory I: Simple, Stratified, Systematic,

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.