PSY 231 1st Edition Lecture 15Outline of Last Lecture I. Inferential StatisticsII. Type 1 Error RiskIII. How Inferential Statistics Generates its AdviceOutline of Current Lecture II. Risk of Belief ErrorsIII. Alpha Different from .05IV. Question & AnswerCurrent LectureRisk of Belief Errors- P estimates type 1 error risk, given that you believe the scientific hypothesis (reject the null)- Alpha is set to discourage belief in the scientific hypothesis(SH)o So alpha should be a low number- Type 1 error risk = p- Type 2 error risk = about 1- p- Example: p= .09o Fail to reject the SH = 9% chance type 1 erroro Rejects the SH = roughly 91% chance of type 2 error- Alpha = .05 means we accept…o No more than 5% of type 1 error risko Up to 95% risk of type 2 error- Belief in SH is potentially dangerous because beliefs influence real world decisionsAlpha Different from .05- P lower than .05 – more worried than usual about type 1 error o Harder for results to be statistically significant- P higher than .05 – more worried than usual about type 2 erroro Easier for results to be statistically significantThese notes represent a detailed interpretation of the professor’s lecture. GradeBuddy is best used as a supplement to your own notes, not as a substitute.- The bigger p gets the lower the risk for type 2 error- In all cases:o Alpha is lowo Type 1 risk is < type 2 risk- Varies: degree of aversion to type 1 errorsQuestion & Answer- Need to know p value and results in order to determine statistical significanceo EX: SH= DV will changeo P= .001o You can make a decision because p shows you that the results were realo EX 2: SH= DV decreaseo P=.001o You cannot make a decision because you do not know what the effect was, you need to know the effect for a one tailed hypothesiso EX 3: SH= DV decreaseso P=?o Results = DV decreaseso Without p we cannot make a hypothesis decision because we do not know how reliable it is without knowing po EX 4: SH= DV increaseso P=.99o Results =?o You can reject the hypothesis because no matter what the results are they are unreliable seeing p is so high- Most reliable p=o N= bigo Mean difference = bigo Standard deviation (variance) = small- Small & unreliable=o N= really bigo Mean difference = smallo SD= bigo Small and unreliable can still be statically significant- Systematic error variance—error variance you control, usually in just one group- Unsystematic error variance – error variance you don’t control and effects all
View Full Document