Stat 217 – Day 10Last TimeComments on HW 2Slide 4Slide 5Slide 6Slide 7Lab 2 Notes (model online)Lab 2 NotesSlide 10Challenge QuestionLab 3About Exam 1Some advice for studyingSome advice during examSome big, big ideasActivity 4-19: Voter Turnout (p. 70)Slide 18Questions?Stat 217 – Day 10ReviewLast Time Judging “spread” of a distribution“Empirical rule”: In a mound-shaped symmetric distribution, roughly 68% of observations fall within one standard deviation of the mean, 95% within two standard deviations of the mean2SD = width of middle 68% of distributionZ-scores measure the relative position of an observation and provide us a unitless measuring stick for how far an observation falls from meanVery useful for comparing values from different distributionsBoxplots – visual display of five number summaryHelpful for comparing distributions (spread, center)Comments on HW 2Problem 2: Identify termsSampling frame is the list of the population used to select the sampleDoes not include the response variable information!(b) average number of words on a page of a textbook(d) tend to gain an average of 15 lbs?Comments on HW 2Act 5-14: Studies from Blink(a) and (b) only had response variables, observational studies(c) and (d) had 2 variables and the explanatory variable was randomly assigned, experimentsSo in (c) and (d) can potentially draw cause and effect conclusions“Generalizability” means can you take information from sample and apply it to the larger population? “There was not a significance difference in SAT performance in the sample so I don’t think there is in the population as well”Yes if have random sample, so maybe only in (a)Comments on HW 2Question 4: Hand hold(b) Can the status of the EV be determined by Ashleigh?Gender of participant vs. gender of researcherRandom sampling vs. random assignmentComments on HW 2Cause and effect vs. generalizing to populationYes NoYesNoWere groups randomly assigned?Were obs units randomly selected?Can draw cause-and-effect conclusionsCan generalize to larger populationComments on HW 2Question 5: AIDS testingMost of you got the table right but then read the wrong proportion from the tableOf those who tested positive, what proportion had AIDS = 4885/78515 = .062Of those who have AIDS, what proportion test positive= 4885/5000 = .977 (sensitivity)Positive test Negative test TotalCarries AIDS virus (2) 4885 (2) 115 (1) 5000Does not carry AIDS (3)73630 (3) 921370 (1) 995,000Total (4) 78515 (4) 921485 1,000,000Lab 2 Notes (model online)Comparing groupsAre people yawning a lot vs. does the yawn seed group yawn more oftenOverall proportion vs. Difference in conditional proportions 4.4% vs. 4.4 percentage pointsYawned “a lot more” vs. “yawned a lot more often”Interpreting p-value vs. conclusions from p-valueProbably want to explicitly compare p-value to some cut-offLab 2 NotesInterpretation of p-valueIf those subjects were going to yawn, regardless of which condition they were in, how often would the random assignment process alone lead to such a large difference in the conditional proportions?Each dot represents one (fake) random assignmentObservation units = 1000 fake random assignmentsVariable = difference in conditional proportionsRoughly 51% of fake random assignments (null model) saw a difference at least this largeDon’t consider this a small p-value since > .05Lab 2 NotesEffect of sample sizeChallenge QuestionWhy was “random assignment” used in the study?Why did we shuffle the cards and deal them into 2 groups?Lab 3 Randomization distributionIf everyone was going to remember the same number of letters regardless of which sequence they got, how often would the random assignment process alone lead to such a big difference in the group means?Each dot represents one random assignmentObservation units = 1000 fake random assignmentsVariable = difference in group meansWhere is the observed difference in means in this distribution?About Exam 150 minutes, 50 pointsWill include one of the self-check activitiesBring calculator, pencil, eraserCould be asked to use Minitab and/or to interpret Minitab outputNo cell phone calculators (square root)One 8.5x11 sheet of own notesBoth sides okI will supply paperSome advice for studyingReview handout, problems onlineSee also p. 627?Review lecture notes, text, hws, labsSee me for old homework, inclass activitiesWork problemsStart with ideas that we have emphasized more oftenSome advice during examIf you get stuck on a problem, move onlater parts, later problemsTry to hit the highlights in your answer (e.g., not all sources of bias, just the most serious)Be succinct (think before you write)Read the question carefullyShow all of your work, explain wellcommunication pointsRead entire question before writing anythingSome big, big ideasObservational units, variableRandom assignment vs. random samplingImplementationPurposeConsequence (Scope of conclusions)What see in sample vs. saying something beyond the sampleStatistic vs. ParameterStatistical significanceInterpretations, reasoningProperties, “what if” questions…How are you deciding this?Activity 4-19: Voter Turnout (p. 70)Statistic: .682 proportion claiming to voteParameter: .490 proportion claiming to voteWhat are some possible explanations for why these values differ?Those in sample do not represent populationThose in sample were not honestStatistics vary from sample to sample and may differ from parameter by chanceWhich of these explanations can we eliminate?No longer believe it was just “by
View Full Document