Testing Hypotheses about Proportions

Introduction to Hypothesis Testing for Proportions

Hypothesis testing for proportions is a fundamental statistical method used to determine whether the observed proportion in a sample provides enough evidence to draw conclusions about the population proportion. This process is widely applied in business, economics, and social sciences to test claims or assumptions about population characteristics.

Formulating Hypotheses

Null and Alternative Hypotheses

Every hypothesis test begins with two competing statements:

• Null Hypothesis (H0): Specifies a population parameter and proposes a value for that parameter. For proportions, it is typically written as H0: p = p0.

• Alternative Hypothesis (HA): Contains the values of the parameter that are considered plausible if we reject the null hypothesis. It can be two-sided (HA: p ≠ p0) or one-sided (HA: p < p0 or p > p0).

Sampling Distribution and Standard Deviation

Standard Deviation vs. Standard Error

When testing hypotheses, we use the model (hypothesized) value for p, not the estimated value from the sample. This gives us the standard deviation of the sampling distribution, not the standard error. The formula for the standard deviation of the sample proportion under the null hypothesis is:

Conditions for Hypothesis Testing

Assumptions and Success/Failure Condition

• The sample must be random.

• The sample size should be less than 10% of the population (to ensure independence).

• There must be at least 10 expected successes and 10 expected failures (Success/Failure Condition).

One-Proportion z-Test

Test Statistic and Calculation

The one-proportion z-test is used to test hypotheses about population proportions. The test statistic is calculated as:

Where is the sample proportion, is the hypothesized population proportion, and is the standard deviation under the null hypothesis.

P-Values and Statistical Significance

Interpreting P-Values

• P-value: The probability of observing data as extreme as (or more extreme than) the observed value, assuming the null hypothesis is true.

• A low P-value (typically below the chosen alpha level) provides evidence against the null hypothesis.

• A high P-value indicates insufficient evidence to reject the null hypothesis, but does not prove it true.

Alpha Levels and Critical Values

Significance Level (α) and Critical Values

The significance level (α) is the threshold for deciding whether a P-value is considered "small." Common alpha levels are 0.05, 0.01, and 0.001. The corresponding critical values for the standard Normal model are:

Confidence Intervals and Hypothesis Tests

Relationship Between Confidence Intervals and Hypothesis Tests

• A confidence interval provides a range of plausible values for the population parameter.

• A two-sided confidence interval at C% confidence corresponds to a two-sided hypothesis test with α = 1 - C%.

• If the null value falls outside the confidence interval, we reject the null hypothesis at the corresponding significance level.

Types of Errors in Hypothesis Testing

Type I and Type II Errors

• Type I Error (α): Rejecting the null hypothesis when it is actually true.

• Type II Error (β): Failing to reject the null hypothesis when it is actually false.

The probability of a Type I error is set by the alpha level. The probability of a Type II error depends on the true value of the parameter, the sample size, and the effect size.

Power of a Test

Definition and Importance

• Power: The probability that a test correctly rejects a false null hypothesis (Power = 1 - β).

• Power increases with larger sample sizes and larger effect sizes.

• Reducing both Type I and Type II errors is best achieved by increasing the sample size.

Summary Table: Key Concepts in Hypothesis Testing for Proportions

Worked Example: Testing a Proportion

Scenario

A survey of 100 CEOs finds that 60 think the economy will improve next year. Is there evidence that the rate is higher among all CEOs than the 55% reported by the public at large?

• Hypotheses: H0: p = 0.55; HA: p > 0.55

• Conditions: Random sample, sample size < 10% of population, at least 10 successes and 10 failures

• Test Statistic:

• P-value: 0.1446 (from Normal table)

• Conclusion: Fail to reject the null hypothesis. There is no evidence that the proportion of CEOs who think the economy will improve next year exceeds 55%.

Best Practices and Common Pitfalls

• Do not base hypotheses on observed data; hypotheses should be set before data collection.

• Always check the necessary conditions before proceeding with a test.

• Distinguish between statistical significance and practical significance.

• Report effect sizes and confidence intervals along with P-values for comprehensive interpretation.

Technical Notes

For confidence intervals, the standard error is estimated from the sample. For hypothesis tests, the standard deviation is calculated using the hypothesized value. When the sample proportion and hypothesized value are close, results are similar; otherwise, use the appropriate formula for each context.