Hypothesis Testing for a Population Proportion

Sampling Distributions & Confidence Intervals: Proportion

Hypothesis Testing for a Population Proportion

Hypothesis testing for a population proportion is a statistical method used to determine whether the proportion of a certain characteristic in a population matches a claimed value. This process is commonly applied when dealing with categorical data, such as the presence or absence of a feature.

• Population Proportion (p): The true proportion of individuals in the population with a specified characteristic.

• Sample Proportion (\hat{p}): The proportion observed in the sample, calculated as \( \hat{p} = \frac{x}{n} \), where x is the number of successes and n is the sample size.

Process for Hypothesis Testing (Proportions)

The following steps outline the process for testing a claim about a population proportion:

1. Identify the hypothesis: Clearly state the claim to be tested.

2. Translate the hypothesis into symbolic form: Express the claim using symbols (e.g., \( p = 0.93 \)).

3. Identify the null and alternative hypotheses:

   • Null hypothesis (H0): The default assumption (e.g., \( p = p_0 \)).

   • Alternative hypothesis (Ha): The claim to be tested (e.g., \( p \neq p_0,\ p < p_0,\ p > p_0 \)).

4. Choose significance level (\( \alpha \)): Common choices are 0.05 or 0.01.

5. Convert the sample statistic into the appropriate test statistic: For proportions, use the z-score:

$z = \frac{\hat{p} - p_0}{\sqrt{\frac{p_0 q_0}{n}}}$ where \( p_0 \) is the claimed population proportion, \( q_0 = 1 - p_0 \), and \( n \) is the sample size.

6. Use either a critical value or P-value to test the null hypothesis: Compare the calculated z-score to the critical value, or compare the P-value to \( \alpha \).

7. Draw conclusions: Decide whether to reject or fail to reject the null hypothesis based on the comparison.

When to Use the Normal Distribution for Proportions

The normal approximation for the sampling distribution of the sample proportion is appropriate when:

• The sample is randomly selected.

• There is a fixed number of independent trials with two possible outcomes (success/failure).

• The conditions \( np \geq 5 \) and \( nq \geq 5 \) are both satisfied, where \( p \) is the hypothesized proportion and \( q = 1 - p \).

Test Statistic for Proportions

The test statistic for a sample proportion is calculated as:

$z = \frac{\hat{p} - p_0}{\sqrt{\frac{p_0 q_0}{n}}}$

• \( \hat{p} \): Sample proportion

• \( p_0 \): Hypothesized population proportion

• \( q_0 = 1 - p_0 \)

• \( n \): Sample size

Example 1: Testing a Claim About Antivirus Software

Scenario: 93% of computer owners are claimed to have antivirus programs. In a sample of 400 computers, 380 (95%) have antivirus software. Test the claim at \( \alpha = 0.05 \).

• Step 1: Hypotheses \( H_0: p = 0.93 \) \( H_a: p \neq 0.93 \)

• Step 2: Requirements check

  • Random sample: Yes

  • Fixed number of independent trials: Yes

  • Two categories: Yes (has antivirus or not)

  • \( np = 400 \times 0.93 = 372 \geq 5 \)

  • \( nq = 400 \times 0.07 = 28 \geq 5 \)

• Step 3: Calculate test statistic

  • \( \hat{p} = \frac{380}{400} = 0.95 \)

  • \( p_0 = 0.93,\ q_0 = 0.07,\ n = 400 \)

  • $z = \frac{0.95 - 0.93}{\sqrt{\frac{0.93 \times 0.07}{400}}}$

• Step 4: Find P-value and compare to \( \alpha \)

• Step 5: Conclusion: The P-value is greater than 0.05, so we fail to reject the null hypothesis. There is not sufficient evidence to reject the claim that \( p = 0.93 \).

Example 2: Testing a Claim About Marriage Rates

Scenario: In 2014, 62% of Americans age 18+ were married. In a recent sample of 900 people, 522 were married. Test the claim that the proportion is now less than 62% at \( \alpha = 0.01 \).

• Step 1: Hypotheses \( H_0: p = 0.62 \) \( H_a: p < 0.62 \)

• Step 2: Requirements check

  • Random sample: Yes

  • Fixed number of independent trials: Yes

  • Two categories: Yes (married or not)

  • \( np = 900 \times 0.62 = 558 \geq 5 \)

  • \( nq = 900 \times 0.38 = 342 \geq 5 \)

• Step 3: Calculate test statistic

  • \( \hat{p} = \frac{522}{900} \approx 0.58 \)

  • \( p_0 = 0.62,\ q_0 = 0.38,\ n = 900 \)

  • $z = \frac{0.58 - 0.62}{\sqrt{\frac{0.62 \times 0.38}{900}}}$

• Step 4: Find P-value and compare to \( \alpha \)

• Step 5: Conclusion: The P-value is less than 0.01, so we reject the null hypothesis. There is sufficient evidence to conclude that the proportion of Americans age 18 and up who are married is now less than 62%.

Summary Table: Steps in Hypothesis Testing for Proportions

Step	Description
1	State the null and alternative hypotheses
2	Check requirements (random sample, independent trials, two categories, \( np \geq 5, nq \geq 5 \))
3	Calculate the test statistic (z-score)
4	Find the P-value or critical value
5	Compare to significance level \( \alpha \) and draw a conclusion

Additional info: In practice, statistical software or z-tables are used to find P-values. The process is similar for one-tailed and two-tailed tests, with the alternative hypothesis determining the direction of the test.