9.3 Two-Sample Test for Proportions

Two-sample proportion tests compare the proportions of two independent groups to determine if there's a significant difference between them. These tests are useful for analyzing binary outcomes in various scenarios, like comparing defective product rates from two factories.

To conduct a two-sample proportion test, you'll need to state hypotheses, calculate the pooled sample proportion, and determine the test statistic. The results help you decide if there's a meaningful difference between the two population proportions, guiding decision-making in business and research contexts.

Two-Sample Test for Proportions

Scenarios for two-sample proportion tests

• Compares proportions of two independent populations or groups
  • Determines if there is a significant difference between the proportions (defective products from two factories)
• Response variable is categorical with two levels
  • Binary outcomes (success/failure, yes/no)
• Samples randomly selected and independent of each other
  • Ensures unbiased representation of the populations
• Sample sizes large enough to assume normal distribution of sample proportions
  • Allows for the use of z-distribution in hypothesis testing

Assumptions of two-sample proportion tests

• Independence within and between samples
  • Random selection from respective populations (simple random sampling)
  • Selection of one sample does not influence the other (no interaction between groups)
• Large sample sizes for normal distribution of sample proportions
  • Rule of thumb: $n_1p_1 \geq 5$, $n_1(1-p_1) \geq 5$, $n_2p_2 \geq 5$, $n_2(1-p_2) \geq 5$
    • $n_1$, $n_2$ = sample sizes; $p_1$, $p_2$ = sample proportions
• Population sizes at least 10 times larger than sample sizes
  • Ensures samples are representative of the populations

Conducting two-sample proportion tests

1. State null and alternative hypotheses

   • $H_0: p_1 = p_2$ (population proportions are equal)
   • $H_a: p_1 \neq p_2$ (two-tailed), $p_1 < p_2$ (left-tailed), $p_1 > p_2$ (right-tailed)

2. Calculate pooled sample proportion: $\hat{p} = \frac{x_1 + x_2}{n_1 + n_2}$

   • $x_1$, $x_2$ = number of successes in each sample

3. Calculate test statistic: $z = \frac{(\hat{p}_1 - \hat{p}_2) - (p_1 - p_2)}{\sqrt{\hat{p}(1-\hat{p})(\frac{1}{n_1}+\frac{1}{n_2})}}$

4. Determine p-value using z-score and standard normal distribution

5. Compare p-value to significance level $\alpha$ and make decision

   • If $p \leq \alpha$, reject $H_0$; if $p > \alpha$, fail to reject $H_0$

6. Interpret results in context of the problem

   • Determine if there is a significant difference between the proportions

Confidence intervals for proportion differences

• Confidence interval for difference between two population proportions:
  • $(\hat{p}_1 - \hat{p}_2) \pm z_{\alpha/2}\sqrt{\frac{\hat{p}_1(1-\hat{p}_1)}{n_1}+\frac{\hat{p}_2(1-\hat{p}_2)}{n_2}}$
  • $z_{\alpha/2}$ = critical value from standard normal distribution
• Interpretation: $(1-\alpha)100\%$ confident true difference between population proportions falls within interval
• If interval contains 0, insufficient evidence to conclude significant difference between proportions

Sample size for proportion tests

• Minimum sample size for each group:
  • $n = \frac{(z_{\alpha/2}+z_\beta)^2}{(\hat{p}_1-\hat{p}_2)^2}[\hat{p}_1(1-\hat{p}_1)+\hat{p}_2(1-\hat{p}_2)]$
  • $z_{\alpha/2}$ = critical value for desired confidence level
  • $z_\beta$ = critical value for desired power (1 - Type II error)
  • $\hat{p}_1$, $\hat{p}_2$ = anticipated sample proportions
• Round up calculated sample size to nearest whole number
  • Ensures sufficient data for accurate results