SOA 1061st Edition Lecture 13 Outline of Last Lecture I. 4 Reasons to go “On the Record” With a HypothesisII. 3 Types of Scientific HypothesesIII. Testing HypothesesIV. Measures of VarianceOutline of Current Lecture II. Error VarianceIII. Two Ambiguities Caused By Selecting Subjecting Subjects from a Varied PopulationIV. Measures of Variance Help us to Diagnose OverlapCurrent LectureError Variance● Error variance -- differences between people that have nothing to do with the IV. Pre-existing differences● Imprecise measurement● Pre-existing individual differences in..○ characteristics you are measuring (DV)○ susceptibility to being changed by the IVTwo Ambiguities Caused by Selecting Subjects from a Varied Population● Group means will almost always differ○ even if IV is not effectiveData Story Possible ExplanationsCentral tendency differs across levels of the IV ● Mean differences tells the truth (IV changes behavior?)● Mean differences is accident of population variance (IV does nothing?)Individual scores overlap across levels of the IV● Overlap tells the truth (IV does nothing?)● Overlap is accident of population errorvariance (IV change behavior?)● Groups will almost always overlap○ even if the IV is effectiveMeasures of Variance Help us to Diagnose Overlap● They measure the amount of similarity between groups○ the less overlap the better○ overlap is too much when if obscures real differences○ if the measure is big, it suggests a lot of overlap, if its small it suggests little overlap● How to avoid hypothesis decision errors? -- Replicate the experiment many times● Inferential Statistics○ advice on whether to believe mean differences○ gives operational definition of “too much
View Full Document