Stor 155, Section 2, Last Time

Stor 155, Section 2, Last Time • Hypothesis Testing • Assess strength of evidence with P-value • P-value interpretation: • Yes – No • Gray – level • 1 - sided vs. 2 - sided “paradox”

Reading In Textbook Approximate Reading for Today’s Material: Pages 400-416, 424-428, 450-471 Approximate Reading for Next Class: Pages 485-504

Hypothesis Testing, III A “paradox” of 2-sided testing: Can get strange conclusions (why is gray level sensible?) Fast food example: suppose gathered more data, so n = 20, and other results are the same

Hypothesis Testing, III One-sided test of: P-value = … = 0.031 Part 5 ofhttp://stat-or.unc.edu/webspace/postscript/marron/Teaching/stor155-2007/Stor155Eg24.xls Two-sided test of: P-value = … = 0.062

Hypothesis Testing, III Yes-no interpretation: Have strong evidence But no evidence !?! (shouldn’t bigger imply different?)

Hypothesis Testing, III Notes: • Shows that yes-no testing is different from usual logic (so be careful with it!) • Reason: 2-sided admits more uncertainty into process (so near boundary could make a difference, as happened here) • Gray level view avoids this: (1-sided has stronger evidence, as expected)

Hypothesis Testing, III Lesson: 1-sided vs. 2-sided issues need careful: • Implementation (choice does affect answer) • Interpretation (idea of being tested depends on this choice) Better from gray level viewpoint

Hypothesis Testing, III CAUTION: Read problem carefully to distinguish between: One-sided Hypotheses - like: Two-sided Hypotheses - like:

Hypothesis Testing Hints: • Use 1-sided when see words like: • Smaller • Greater • In excess of • Use 2-sided when see words like: • Equal • Different • Always write down H0 and HA • Since then easy to label “more conclusive” • And get partial credit….

Hypothesis Testing E.g. Text book problem 6.34: In each of the following situations, a significance test for a population mean, is called for. State the null hypothesis, H0 and the alternative hypothesis, HA in each case….

Hypothesis Testing E.g. 6.34a An experiment is designed to measure the effect of a high soy diet on bone density of rats. Let = average bone density of high soy rats = average bone density of ordinary rats (since no question of “bigger” or “smaller”)

Hypothesis Testing E.g. 6.34b Student newspaper changed its format. In a random sample of readers, ask opinions on scale of -2 = “new format much worse”, -1 = “new format somewhat worse”, 0 = “about same”, +1 = “new a somewhat better”, +2 = “new much better”. Let = average opinion score

Hypothesis Testing E.g. 6.34b (cont.) No reason to choose one over other, so do two sided. Note: Use one sided if question is of form: “is the new format better?”

Hypothesis Testing E.g. 6.34c The examinations in a large history class are scaled after grading so that the mean score is 75. A teaching assistant thinks that his students have a higher average score than the class as a whole. His students can be considered as a sample from the population of all students he might teach, so he compares their score with 75. = average score for all students of this TA

Hypothesis Testing E.g. Textbook problem 6.36 Translate each of the following research questions into appropriate and Be sure to identify the parameters in each hypothesis (generally useful, so already did this above).

Hypothesis Testing E.g. 6.36a A researcher randomly divides 6-th graders into 2 groups for PE Class, and teached volleyball skills to both. She encourages Group A, but acts cool towards Group B. She hopes that encouragement will result in a higher mean test for group A. Let = mean test score for Group A = mean test score for Group B

Hypothesis Testing E.g. 6.36a Recall: Set up point to be proven as HA

Hypothesis Testing E.g. 6.36b Researcher believes there is a positive correlation between GPA and esteem for students. To test this, she gathers GPA and esteem score data at a university. Let = correlation between GPS & esteem

Hypothesis Testing E.g. 6.36c A sociologist asks a sample of students which subject they like best. She suspects a higher percentage of females, than males, will name English. Let: = prop’n of Females preferring English = prop’n of Males preferring English

Hypothesis Testing HW on setting up hypotheses: 6.35, 6.37

Hypothesis Testing Connection between Confidence Intervals and Hypothesis Tests: Reject at Level 0.05 P-value < 0.05 dist’n Area < 0.05 0.95 margin of error

Hypothesis Testing & CIs Reject at Level 0.05 Notes: • This is why EXCEL’s CONFIDENCE function uses = 1 – coverage prob. • If only care about 2-sided hypos, then could work only with CIs (and not learn about hypo. tests)

Hypothesis Testing & CIs HW: 6.71

Hypothesis Testing The three traps of Hypothesis Testing (and how to avoid them…) Trap 1: Statistically Significant is different from Really Significant (don’t confuse them)

Hypothesis Testing Traps Trap 1: Statistically Significant is different from Really Significant E.g. To test a painful diet program, 10,000 people were put on it. Their average weight loss was 1.7 lbs, with s = 73. Assess “significance” by hypothesis testing.

Hypothesis Testing Traps Trap 1: Statistically Significant is different from Really Significant See Class Example 25: Trap 1 http://stat-or.unc.edu/webspace/postscript/marron/Teaching/stor155-2007/Stor155Eg25.xls P-value = 0.0099 Strongly Statistically Significant Careful: Is this practically significant?

Hypothesis Testing Traps Trap 1, e.g: Is this practically significant? NO! Not worth painful diet to lose 1.7 lbs. Resolution: Hypo. testing resolves question: Could observed results be due to chance variation? Answer here is no, since n is really large.

Hypothesis Testing Traps Trap 1, e.g: Is this practically significant? Answer here is no, since n is really large. But this is different from question: Do results show a big difference?

Hypothesis Testing Traps Trap 2: Insignificant results do not mean nothing is there, Only: Didn’t have strong enough data to actually prove results. E.g. Class 25, Trap 2 http://stat-or.unc.edu/webspace/postscript/marron/Teaching/stor155-2007/Stor155Eg25.xls

Hypothesis Testing Traps Trap 3: Try enough tests, and you will find “something” even where it doesn’t exist. Revisit Class Example 21, Q4 http://stat-or.unc.edu/webspace/postscript/marron/Teaching/stor155-2007/Stor155Eg23.xls We saw about 5% of CIs don’t cover. So, (using CI – Hypo Test connection), expect about 5% of tests to choose HA, (and claim “strong evidence) even when H0 true.

Hypothesis Testing Traps Strategies to avoid Trap 3: • Scientific Method: Form Hypothesis tests once. • For repeated tests: use careful adjustments: (beyond scope of this course) Get help if needed

Hypothesis Testing Traps HW: 6.74 (about 50) 6.82 (0.382, 0.171, 0.0013) 6.83, 6.84 (0.0505, 0.0495)

Some stuff for next time (next slides)

Hypothesis Testing View 3: level testing Idea: instead of reporting P-value, choose a fixed level, say 5% Then reject H0, i.e. find strong evidence… When P-value < 0.05 (more generally ) (slight recasting of yes-no version of testing) HW: 6.53 (careful, already assigned above)

Hypothesis Testing View 4: P-value shows only half of the decision problem: Graphical Illustration: Truth H0 true HA true Test Result: Choose H0 Choose HA

Hypothesis Testing View 4: Both sides of decision problem: Small P-values Small Type I error What about Type II error? (seems part of problem) a. Simplistic Answer: Don’t care because have put burden of proof on HA.

Hypothesis Testing View 4: What about Type II error? • Deeper Answer: Does matter for test sensitivity issues, and test power issues E.g. how large a sample size is needed for a given test power (treated above)?

Hypothesis Testing View 4: Terminology: P{Type I Error} is called level of test 1 - P{Type I Error} is called specificity 1 – P{Type II Error} is called power of test 1 – P{Type II Error} is called sensitivity

And now for something completely different… A statistician’s view on politics… Some Current Controversial Issues: • North Carolina State Lottery • Replace Social Security by Individual Retirement Plans Debate is passionate, (natural for complex and important issues) But what is missing?

And now for something completely different… Review Ideas on State Lotteries, from our study of Expected Value Not an obvious choice because: • Gambling is (at least) unsavory: • Religious objections • Some like it too much • Destroys some lives

And now for something completely different… State Lotteries, not an obvious choice: • The only totally voluntary tax: • Nobody required, unlike all other taxes • Money often used for education • Good or bad, given state of economy??? • Highest tax burden on the poor • Poor enjoy playing much more • Higher taxes on poor better for society??? • Tendency towards “rich get richer”???

And now for something completely different… What about Individual Retirement Plans: Main Benefit: On average individual investments return greater yields than government investments So can we conclude: • “Overall we are all better off”??? • Since more total money to go around?

And now for something completely different… Very common mistake in this reasoning: • Notice “on average” part of statement • Should also think about variation about the average???

And now for something completely different… Variation about average Issue 1: • Should think of population of people • Average is over this population • Except some to do great • And expect some to lose everything • What will the percentage of losers be? • What do we do with those who lose all? • What will that cost?

And now for something completely different… Variation about average Issue 2: • Also are averaging over time • Overall gains of stock market happen only over this average • Some need $$$ when market is down • How often will this happen? • How do we deal with it?

And now for something completely different… Main concept I hope you carry away from this course: Variation is a fundamental concept • Look for it • Think about it • Ask questions about it (Vital to informed citizenship)

And now for something completely different… Australian joke about Variation: Did you hear about the man who drowned in a lake with average depth 6 inches?

And now for something completely different… Australian joke about Variation: He understood “average”, but not variation about the average

And now for something completely different… Really have such lakes? Yes, in Australia

And now for something completely different… Suggestions of such issues (politics, controversy…) for discussion are welcome….

Stor 155, Section 2, Last Time