Hypothesis Testing: Concepts, Procedures, and Applications

Hypothesis Testing

Introduction to Hypothesis Testing

Hypothesis testing is a fundamental statistical procedure used to make inferences about population parameters based on sample data. It involves formulating statements (hypotheses) about a population and using probability theory to determine whether the observed data provide sufficient evidence to reject a specified claim.

• Hypothesis: A proposition or statement regarding a characteristic of one or more populations.

• Hypothesis Testing: A procedure based on sample evidence and probability to test statements about population characteristics.

• Importance: Hypothesis testing is foundational for advanced statistical topics, including correlation, regression, and chi-square tests.

Key Terms and Concepts

• Null Hypothesis (H0): The statement being tested, usually representing no effect or status quo.

• Alternative Hypothesis (H1 or Ha): The statement we consider if evidence suggests H0 is unlikely.

• Types of Tests:

  • Two-tailed Test: Tests for deviation in both directions (e.g., H1: μ ≠ μ0).

  • Right-tailed Test: Tests for values greater than a specified value (e.g., H1: μ > μ0).

  • Left-tailed Test: Tests for values less than a specified value (e.g., H1: μ < μ0).

• Level of Significance (α): The probability threshold for rejecting H0, commonly set at 0.05, 0.01, or 0.10.

• Type I Error (α): Rejecting a true null hypothesis.

• Type II Error (β): Failing to reject a false null hypothesis.

General Steps in Hypothesis Testing

1. State the hypotheses: Formulate H0 and H1.

2. Select the significance level (α): Choose the probability of making a Type I error.

3. Determine the appropriate test statistic: Depends on the parameter and sample size.

4. Find the critical value(s) or p-value: Based on the chosen α and the sampling distribution.

5. Make a decision: Compare the test statistic to the critical value(s) or compare the p-value to α.

6. State the conclusion: Interpret the results in the context of the original claim.

Hypothesis Tests for a Population Proportion

Testing a Population Proportion

When testing claims about a population proportion, the sample proportion is compared to the claimed value using the normal approximation (when sample size is large enough). Both the traditional (critical value) and p-value approaches are used.

• Test Statistic Formula: $z_0 = \frac{\hat{p} - p_0}{\sqrt{\frac{p_0(1-p_0)}{n}}}$

• Conditions: The sample size should be large enough for the normal approximation to be valid (np and n(1-p) both ≥ 5).

• Critical Value Approach: Compare the calculated z-value to the critical value from the standard normal table.

• P-value Approach: Calculate the probability of observing a test statistic as extreme as, or more extreme than, the observed value under H0.

Example: Testing a Drug Manufacturer's Claim

• Claim: More than 94% of patients are healed within 8 weeks.

• Sample Data: 213 out of 224 patients healed.

• Hypotheses: H0: p = 0.94; H1: p > 0.94 (right-tailed test)

• Significance Level: α = 0.01

• Test Statistic Calculation: $\hat{p} = \frac{213}{224} \approx 0.9509$ $z_0 = \frac{0.9509 - 0.94}{\sqrt{\frac{0.94 \times 0.06}{224}}} \approx 0.69$

• Critical Value: z = 2.33 (for α = 0.01, right-tailed)

• Decision: 0.69 < 2.33; do not reject H0.

• Conclusion: Insufficient evidence to support the manufacturer's claim at the 1% significance level.

• P-value: 0.2451 (greater than α); do not reject H0.

Additional info: Confidence intervals can be used to support hypothesis test conclusions. If the claimed value falls within the confidence interval, the null hypothesis is not rejected.

Hypothesis Tests for a Population Mean

Testing a Population Mean

When testing claims about a population mean, the t-distribution is used if the population standard deviation is unknown and the sample size is small. For large samples, the normal distribution may be used.

• Test Statistic Formula (t-test): $t_0 = \frac{\bar{x} - \mu_0}{\frac{s}{\sqrt{n}}}$

• Conditions: The sample should be random, and the population should be approximately normal if n is small.

• Critical Value Approach: Compare the calculated t-value to the critical value from the t-distribution table (degrees of freedom = n - 1).

• P-value Approach: Calculate the probability of observing a test statistic as extreme as, or more extreme than, the observed value under H0.

Example: SAT Critical Reading Scores

• Claim: Students who learn English as well as another language simultaneously score worse on SAT Critical Reading.

• Population Mean: μ = 501

• Sample Data: n = 100, \( \bar{x} = 485 \), s = 116

• Hypotheses: H0: μ = 501; H1: μ < 501 (left-tailed test)

• Significance Level: α = 0.10

• Test Statistic Calculation: $t_0 = \frac{485 - 501}{116 / \sqrt{100}} = \frac{-16}{11.6} \approx -1.38$

• Critical Value: t = -1.290 (for α = 0.10, df = 99)

• Decision: -1.38 < -1.290; reject H0.

• Conclusion: Sufficient evidence to support the claim that these students score worse.

• P-value: 0.0855 (less than α); reject H0.

Additional info: Distinguish between statistical significance (result unlikely due to chance) and practical significance (result is meaningful in real-world terms).

Hypothesis Tests for a Population Standard Deviation

Testing a Population Standard Deviation

To test claims about a population standard deviation, the chi-square distribution is used. The procedure is similar to tests for means and proportions, but the test statistic and critical values are based on the chi-square distribution.

• Test Statistic Formula: $\chi^2_0 = \frac{(n-1)s^2}{\sigma_0^2}$

• Conditions: The sample should be random, and the population should be approximately normal.

• Critical Value Approach: Compare the calculated chi-square value to the critical value(s) from the chi-square distribution table (degrees of freedom = n - 1).

• P-value Approach: Calculate the probability of observing a test statistic as extreme as, or more extreme than, the observed value under H0.

Additional info: The chi-square test can be right-tailed, left-tailed, or two-tailed, depending on the alternative hypothesis. Refer to chi-square tables for critical values.

Summary of Hypothesis Testing Procedures

Sections 10.2 – 10.4 cover hypothesis tests for proportions, means, and standard deviations. The general steps are similar, but the test statistic and distribution used depend on the parameter being tested. Section 10.6 provides further discussion on Type II error and the power of a test.

Table: Summary of Hypothesis Testing Procedures

Parameter	Test Statistic	Distribution	Conditions
Proportion (p)	$z_0 = \frac{\hat{p} - p_0}{\sqrt{\frac{p_0(1-p_0)}{n}}}$	Standard Normal (z)	np, n(1-p) ≥ 5
Mean (μ)	$t_0 = \frac{\bar{x} - \mu_0}{\frac{s}{\sqrt{n}}}$	t-distribution (df = n-1)	Population normal or n large
Standard Deviation (σ)	$\chi^2_0 = \frac{(n-1)s^2}{\sigma_0^2}$	Chi-square (df = n-1)	Population normal

Conclusion Statements

• Always state the conclusion in the context of the original claim.

• Indicate whether there is sufficient evidence to reject the null hypothesis at the chosen significance level.

Additional Notes

• Proper rounding is important in reporting results.

• Tables for critical values (z, t, chi-square) are essential tools; refer to the textbook appendix as needed.

• Understanding both the traditional (critical value) and p-value approaches is necessary for homework and exams.