PSYC 3250 Lecture Notes - Factor Analysis, Test Validity, Base Rate
5 Aug 2014
School
Department
Course
Professor
Document Summary
Principal axis: not all var in items can be explained, will have measurement error in items. P components: there is error, all var accounted for by analysis. Factor analysis: first table of communalities has initial column= 1 and extraction column. What is the chance that he/ she has disease? o: bayes theorem: it specifies the relationship btwn positive predicted value, sensitivity, false positive rate and prevalence, se(p)/ se(p) + frp(1-p)= ppv, ppv= 1(. 001)/ 1(. 001)+ . 05(. 999)= . 02. If increase prevalence= . 10, fpr= . 05, se= 1 (perfect: ppv= 1(. 1)/1(. 1)+ . 05(. 999) = 50% pre; ppv= . 95: prev= . 0001; ppv= . 002, in all these the se is same and the fpr is the same. Ie test validity, item analysis and factor analysis. Program will extract as many factors as you have items. Most will be trivial (devoid of meaning/ misleading. Eigenvalue is amount of total variance explained by factor. There is an eigenvalue calculated for each factor extracted.
