This preview shows page 1. Sign up to view the full content.
Unformatted text preview: esearcher conducDng a study is to get signiﬁcant results, but only if the research hypothesis is really true If the research hypothesis is false, we do not want to get a signiﬁcant result (that would be a Type I error!) It helps in planning studies If power is too low (i.e., you don’t have a good probability of supporDng your research hypothesis if it is indeed true), you might: ▪ Choose not to do the study (wasDng your Dme and $) ▪ Or change the design to increase your power It helps when evalua@ng the results of studies If a study has non
signiﬁcant results, you might ask: did the study have enough power? 4 9/29/13 It’s complicated… typically researchers use computer programs and specialized staDsDcal so_ware to calculate power We will learn the approach (for simple diﬀerence between means) in order to develop a conceptual understanding of issues related to staDsDcal power The Basic Concept First, you ﬁnd the Z score on the PopulaDon 1 distribuDon corresponding to the cutoﬀ score on the comparison distribuDon The probability of exceeding this Z
score (which is the power) can be found using a normal curve table 1. Gather the needed informaDon: a) The mean and standard deviaDon of Pop 2’s distribuDon of means (the comparison distribuDon) b) The predicted mean of PopulaDon 1 ▪ Basing predicDon on previous research 5 9/29/13 2. Figure the raw score cutoﬀ point on the comparison distribuDon needed to reject the null hypothesis a) On the comparison distribuDon, shade the alpha region (region more extreme than the criDcal value); this is the alpha region 3. Figure the Z score for this cutoﬀ point, but on the distribuDon of means for PopulaDon 1 a) On the distribu...
View
Full
Document
This note was uploaded on 03/24/2014 for the course PSY 21201 taught by Professor Bernard during the Winter '13 term at SUNY Stony Brook.
 Winter '13
 bernard
 Psychology

Click to edit the document details