Question:hard

The following two statements are about the use of stratified sampling for finite population:
Assertion (I): Sampling error of an estimator can always be reduced by using stratified sampling following principle of stratification.
Reason (II): Stratified sampling under optimal allocation for a fixed sample size reduces the mean square error of the estimator.
Which of the following is the correct explanation?

Show Hint

Both statements are individually correct results of sampling theory, but check whether the specific "optimal allocation" claim in (II) is really what makes the general claim in (I) true.
Updated On: Jul 4, 2026
  • Both (I) and (II) are true but (II) is not the correct explanation for (I)
  • Both (I) and (II) are true but (II) is the correct explanation for (I)
  • Only (I) is true but (II) is false
  • Both (I) and (II) are false
Show Solution

The Correct Option is A

Solution and Explanation

Step 1: Test Assertion (I) using the standard variance decomposition. For any population, the total variance can be split as $\sigma^2 = \sigma_w^2 + \sigma_b^2$, where $\sigma_w^2$ is the (weighted) within-stratum variance and $\sigma_b^2$ is the between-stratum variance. The variance of the stratified mean (proportional allocation) depends only on $\sigma_w^2/n$, which is at most $\sigma^2/n$, the SRS variance. This proves stratification (properly done) never increases sampling error, confirming (I) is true.
Step 2: Test Reason (II) independently. Given a fixed $n = \sum n_h$, minimizing $Var(\bar y_{st}) = \sum \dfrac{N_h^2 \sigma_h^2}{n_h}$ subject to that constraint by Lagrange/Cauchy-Schwarz gives the Neyman solution $n_h \propto N_h \sigma_h$, and this minimized variance is mathematically less than or equal to the variance under proportional allocation. So (II) is also independently true.
Step 3: Judge the explanatory link. (I) speaks generally about "following the principle of stratification" (which includes even equal or proportional allocation, not necessarily optimal), while (II) is a special, stronger claim restricted to optimal allocation. Since (I) would remain true even without invoking optimal allocation, (II) cannot be regarded as the reason behind (I); the two are true but logically independent statements.
So both statements hold, yet the reason given is not what explains the assertion.
\[\boxed{\text{Both (I) and (II) true, (II) is not the correct explanation of (I)}}\]
Was this answer helpful?
0