p value

Introduction to P-Value Method

• The video is focused on the concept of p-values in hypothesis testing.

• The p-value helps in making a decision regarding the null hypothesis (H0).

Definition of P-Value

• The p-value is defined as:   - The probability of obtaining results at least as extreme as observed results, under the assumption that the null hypothesis is true.

Explanation of the Definition

• Consider null hypothesis (H0): Mbewe (population parameter) ≤ 14.

• Alternative hypothesis (H1): Mbewe > 14.

• The null hypothesis is presumed true unless there is compelling evidence to suggest otherwise.

• In this example, the null hypothesis can be simplified to:   - H0: Mbewe = 14

Concept Visualization

• According to the Central Limit Theorem, the sample means (X̄) follow a normal distribution centered around the hypothesis value (here, 14).

• If a sample statistic (X̄) is 16.1:   - If 16.1 is merely marginally greater than 14, it is not surprising.   - However, if it is significantly far from 14, there may be skepticism regarding the null hypothesis.

Understanding Significance of Sample Statistic

• A very high or low sample statistic in relation to 14 raises doubts about the null hypothesis.

• This discrepancy will help determine the p-value:   - In right-tailed tests, it represents the probability of observing results greater than the sample statistic (X̄).

Decision Making in Hypothesis Testing

• Decisions on the null hypothesis are made by comparing the p-value to the significance level (alpha).

• If p-value ≤ alpha, reject H0.

• If p-value > alpha, fail to reject H0.

Example

• Given:   - H0: Mbewe ≤ 14   - H1: Mbewe > 14   - p-value = 0.058   - Significance level (alpha) = 0.01

• Since 0.058 > 0.01, fail to reject the null hypothesis.

Understanding Alpha

• Alpha (α) is the threshold for significance in hypothesis testing, often set at:   - α = 0.05 (standard)   - α = 0.01 (more stringent)   - α = 0.10 (less stringent)

• Alpha represents the probability of making a Type I error (rejecting a true null hypothesis).

Calculating P-Values

• P-values can sometimes be provided directly; other times, they must be calculated based on test statistics:

Types of Tests

1. Right-Tailed Test

• For right-tailed tests:   - Z-test statistic calculated from sample data.   - P-value = area to the right of the test statistic on the Z-distribution, computed as:     $P = 1 - ext{normdist}(Z, 0, 1, ext{TRUE})$

2. Left-Tailed Test

• For left-tailed tests:   - P-value equals the area to the left of the test statistic:     $P = ext{normdist}(Z, 0, 1, ext{TRUE})$

3. Two-Tailed Test

• Two-tailed tests account for extreme values on both sides:   - P-value = area in both tails, calculated by:     $P = 2 imes ext{normdist}(|Z|, 0, 1, ext{TRUE})$

• This addition captures outcomes that are either significantly higher or lower than the hypothesized value.

Example Calculations

• Example: Test statistic Z = 0.95 for a right-tailed test:   - P-value calculation:     $P = 1 - ext{normdist}(0.95, 0, 1, ext{TRUE})$   - Result: P-value ≈ 0.1711 (rounded)

• Example: Test statistic Z = 2.95 for a two-tailed test:   - Calculate left tail first,     $P = 2 imes ext{normdist}(-2.95, 0, 1, ext{TRUE})$   - Or for right tail,     $P = 2 imes (1 - ext{normdist}(2.95, 0, 1, ext{TRUE}))$

Conclusion

• Understanding p-values and their implications in hypothesis testing is crucial for statistical analysis.

• The next video will discuss conducting a complete hypothesis test from start to finish.