Stat 217 – Day 13Last Time – Confidence IntervalsConfidence IntervalExam 1Exam 1 ResourcesExam 1 AdviceSome advice during examSome big, big ideasMain TopicsTest of SignificanceSlide 11ExampleQuestions?Review SessionStat 217 – Day 13ReviewLast Time – Confidence IntervalsWhen we can apply the CLT large enough sample sizerandom sampling from large populationthe “empirical rule” predicts 95% of our statistics fall within 2 standard deviations one each side of the population parameterSo 95% of the time our population parameter is within 2 standard deviations of our sample statisticConfidence Interval“margin-of-error” = .04 or 4 percentage pointsExam 1May use one 8.5 x 11 (both sides) page of self-produced notesMixture of multiple choice, short answer, longer questions (see quizzes, labs)Bring a calculator (not a cell phone)Access to the computer (e.g., applets, JMP)Exam 1 ResourcesReview handoutReview questions/solutionsReadings 1-8Self-check videosWhat Went Wrong ExamplesQuiz solutions (and blank copies)Access to pre-labsSelf-tests, practice problemsSome grading comments on quizzes, investigations labs, annotated handouts (lecture notes page)Exam 1 AdviceReview handout, problems onlineWork problemsReview labsStart 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, variableWhat see in sample (descriptive) vs. saying something beyond the sample (inferential)Statistic vs. ParameterInterpretation of p-valueStatistical significanceInterpretations, reasoningProperties, “what if” questions…How are you deciding this?Main TopicsSample from a random process (e.g., coin toss, dolphins, cancer dog)Parameter: = probability of “success”Statistic: sample proportionRandom sample from a finite large population (e.g., Gallup poll)Parameter: = population proportion of “successes”Statistic: sample proportionConsider sampling, nonsampling biasesTest of SignificanceTest a conjecture about Assume null hypothesis is trueLook at the distribution of the statistic when the null hypothesis is trueSimulation (Coin Tossing, One Prop Inf)Central Limit Theorem (Reese’s Pieces)If observed value is in the tail of the distribution (small p-value), reject the null hypothesis. Otherwise “fail to reject.”FTR: Not convincing evidence against HoConfidence IntervalWant to estimate from the sample dataCould test all the values and make an interval of the ones that are not rejectedNot practicalWhen the CLT applies, estimate + margin-of-errorestimate + 2 standard errorsI’m 95% confident that the parameter is between these two valuesProcedure works 95% of the timeExampleQuestions?Review Session7-9pm, Building 53, Room
View Full Document