AP Statistics (AP Stats): Interpreting p?value and Statistical Significance

AP Statistics – Interpreting p?value and Statistical Significance

AP Statistics Study Guide: Interpreting p-value and Statistical Significance

What This Is

The p-value is the probability of obtaining a test statistic as extreme or more extreme than the one observed, assuming the null hypothesis (H?) is true. It helps determine whether results are statistically significant—meaning they’re unlikely to have occurred by random chance. On the AP exam, you’ll use p-values to make decisions about hypotheses (e.g., "Does a new teaching method improve test scores?" or "Is the proportion of defective smartphones in a factory higher than 5%?"). Misinterpreting p-values is a common mistake, so mastering this concept is critical for both multiple-choice and FRQs.

Key Terms & Formulas

• Null Hypothesis (H?): A statement of "no effect" or "no difference" (e.g., H?: p = 0.5 or H?:-= 100).
• Alternative Hypothesis (H?): The claim we’re testing (e.g., H?: p > 0.5 [one-sided] or H?:-? 100 [two-sided]).
• p-value: Probability of observing a test statistic as extreme as the one calculated, assuming H? is true. Smaller p-value-stronger evidence against H?.
• Significance Level (?): Threshold for rejecting H? (commonly ? = 0.05). If p-value-?, reject H?; otherwise, fail to reject H?.
• Statistically Significant: Results are unlikely due to random chance (p-value-?).
• Type I Error: Rejecting H? when it’s true (false positive). Probability = ?.
• Type II Error: Failing to reject H? when it’s false (false negative). Probability = ?.
• One-Sample z-test for p: z = (p? – p?) / ?(p?(1–p?)/n)
• p? = sample proportion, p? = hypothesized proportion, n = sample size.
• One-Sample t-test for ?: t = (x? – ) / (s/?n)
• x? = sample mean, = hypothesized mean, s = sample SD, n = sample size.
• Calculator Commands (TI-84):
• 1-PropZTest: STAT-TESTS-5:1-PropZTest (for proportions).
• T-Test: STAT-TESTS-2:T-Test (for means).
• Normalcdf: 2nd-VARS-2:normalcdf(lower, upper, ?, ?) (for p-values from z-scores).
• tcdf: 2nd-VARS-6:tcdf(lower, upper, df) (for p-values from t-scores).

Step-by-Step / Process Flow

Follow these steps for any hypothesis test on the AP exam:

1. State Hypotheses
2. Define H? (null) and H? (alternative) in context.

3. Example: Testing if a coin is fair (H?: p = 0.5, H?: p-0.5).

4. Check Conditions

5. Proportions (z-test): BINS (Binary outcomes, Independent trials, np?-10, n(1–p?)-10, 10% condition).
6. Means (t-test): LINER (Linear data, Independent observations, n-30 or normal population, 10% condition).

7. Random: Data must come from a random sample/experiment.

8. Compute Test Statistic

9. Use the appropriate formula (z or t) or calculator command.

10. Example: For a proportion, use 1-PropZTest; for a mean, use T-Test.

11. Find p-value

12. Use normalcdf (for z) or tcdf (for t) if calculating by hand.

13. For two-sided tests, double the p-value from the calculator.

14. Make a Conclusion

15. Compare p-value to ? (usually 0.05).
16. If p-value-?: "Reject H?. There is convincing evidence that [H? in context]."

17. If p-value > ?: "Fail to reject H?. There is not convincing evidence that [H? in context]."

18. Interpret in Context

19. Always link your conclusion to the problem’s scenario.
20. Example: "There is convincing evidence that the new drug lowers blood pressure (p-value = 0.02 < 0.05)."

Common Mistakes

• Mistake: Interpreting the p-value as "the probability that H? is true."

• Correction: The p-value is the probability of the data (or more extreme) assuming H? is true, not the probability of H? itself.

• Mistake: Forgetting to double the p-value for two-sided tests.

• Correction: For H?:-? , multiply the one-sided p-value by 2 (or use T-Test with ?:?).

• Mistake: Using a z-test for means when the population SD is unknown.

• Correction: Use a t-test when-is unknown (unless n-30, where z is acceptable).

• Mistake: Ignoring the 10% condition for independence.

• Correction: If sampling without replacement, check n-0.10N (where N = population size).

• Mistake: Confusing "statistically significant" with "practically important."

• Correction: A tiny p-value doesn’t always mean the effect is large or meaningful (e.g., a drug lowering blood pressure by 1 mmHg may be statistically significant but not clinically useful).

AP Exam Insights

• Tricky Distinctions:
• z vs. t: Use z for proportions or means with known ?; use t for means with unknown? (unless n-30).
• One-sided vs. two-sided tests: The AP exam often tests whether you double the p-value for two-sided H?.

• p-value vs. ?: The p-value is calculated from data;-is set before the test.

• Common FRQ Setups:

• "Is there evidence that the proportion of students who prefer online learning has increased?" (One-sided test for p).
• "Does the new fertilizer increase crop yield?" (One-sided t-test for ?).

• "Is the machine producing bags with a mean weight different from 50 lbs?" (Two-sided t-test for ?).

• Calculator Pitfalls:

• Forgetting to input the correct or p? in T-Test or 1-PropZTest.
• Using normalcdf instead of tcdf for t-tests (or vice versa).
• Not doubling the p-value for two-sided tests when calculating by hand.

Quick Check Questions

1. Multiple Choice: A researcher tests H?: p = 0.4 vs. H?: p > 0.4 and gets a p-value of 0.03. If ? = 0.05, what is the correct conclusion?
2. (A) Reject H?; there is convincing evidence that p > 0.4.
3. (B) Fail to reject H?; there is not convincing evidence that p > 0.4.
4. (C) Reject H?; there is convincing evidence that p = 0.4.
5. (D) Fail to reject H?; there is convincing evidence that p = 0.4.

Answer: (A) The p-value (0.03) is less than? (0.05), so we reject H?.

1. FRQ Part: A factory claims that 5% of its lightbulbs are defective. A quality control inspector tests 200 bulbs and finds 16 defective. Is there convincing evidence that the defect rate is higher than 5%? Use ? = 0.05.
2. Step 1: State hypotheses.
3. Step 2: Check conditions.
4. Step 3: Compute the test statistic and p-value.
5. Step 4: Make a conclusion in context.

Answer: - Step 1: H?: p = 0.05, H?: p > 0.05. - Step 2: BINS: Binary (defective/not), Independent (200-10% of all bulbs), np? = 10-10, n(1–p?) = 190-10. - Step 3: p? = 16/200 = 0.08; z = (0.08 – 0.05)/?(0.050.95/200)-1.95; p-value = normalcdf(1.95, 1E99, 0, 1)-0.0256. - Step 4:* Since 0.0256 < 0.05, reject H?. There is convincing evidence that the defect rate is higher than 5%.

Last-Minute Cram Sheet

1. p-value-?-Reject H? (convincing evidence for H?).
2. p-value >-? Fail to reject H? (not enough evidence for H?).
3. One-sided test: p-value = area in one tail.
4. Two-sided test: p-value = 2 × area in one tail.
5. z-test for p: Use when testing proportions (BINS conditions).
6. t-test for ?: Use when testing means (LINER conditions).
7. Calculator: 1-PropZTest for proportions, T-Test for means.
8. Always check conditions (BINS/LINER + 10% rule).
9. Never say "H? is true"—only "fail to reject H?."
10. Interpret p-value in context (e.g., "Assuming the drug has no effect, there’s a 2% chance of observing results this extreme.").