Two independent simple random samples are taken without replacement. Sample \(1\) has size \(90\) from a population of \(700\), and sample \(2\) has size \(120\) from a population of \(2000\). The population proportions are \(p_1=0.50\) and \(p_2=0.40\).
Check the \(10\%\) rule and the expected success and failure counts for both samples. Is the usual two-sample standard deviation formula justified? Does a failed \(10\%\) rule prove that the sampling distribution is not approximately normal?
Hints
- Check the fraction sampled from each population separately.
- Work out expected yes and no counts using each population proportion and sample size.
- Consider separately what sampling without replacement does to spread and what determines an approximately bell-shaped distribution.
Solution
1. Both samples are random and independent of each other. The expected success and failure counts are \(45\), \(45\), \(48\), and \(72\), all at least \(10\).
2. Sample \(1\) exceeds \(10\%\) of its population because \(90>0.10(700)=70\). Sample \(2\) meets the rule because \(120\le0.10(2000)=200\).
3. Because the \(10\%\) rule fails for sample \(1\), the usual approximation treating draws within that sample as independent is not justified. The usual standard deviation formula may need an adjustment for sampling without replacement.
4. This failure does not by itself rule out an approximately normal shape. The large-count condition still holds.
Answer
Sample \(1\) fails the \(10\%\) rule; sample \(2\) passes. All four expected counts meet the large-count condition. The usual standard deviation formula is not justified without adjustment, but the failed \(10\%\) rule does not by itself rule out an approximately normal shape.