Factors may correlate as high as 428 if factors are

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: ors may correlate as high as .428. If factors are used in predictors in regression analysis, their moderate correlation would create a problem of MULTICOLLINEARITY! FA solutions are supposed to minimize issues with multicollinearity, not create them! Final Model Final .428 Factor 1 i1 i2 i3 i4 Factor 2 i5 i6 i7 i8 i9 i10 Issues with this model: 1 – Hypothesized 1 factor, obtained 2 factors 2 – 3 of 10 items are cross-loaders 3 – Two factors have a moderate correlation of .428 which would create problem of multicollinearity if used as two predictors in regression analysis RECOMMENDATION: REJECT THIS MODEL!! Forced Factor Solution Forced Use only after rejection of initial solution Force the number of factors expected Interpret Forced Solution Interpret Note: There are still 2 Eigen Values above 1.0. However in forced solution, only one factor is solved for (b’c that was specified) Forced Solution: Keep all items except for los1. More sensible than initial solution. Uncorrelated factors: Orthogonal (Varimax) Rotation Orthogonal Expect 2 or more uncorrelated factors, Expect want rotation to be orthogonal (means independent) independent) In SPSS, click Varimax Rotation Example: 20-item scale measuring Example: Extraversion and Neuroticism Traits Extraversion Initial 4-factor solution Initial Forced 2-Factor Solution Forced To interpret this solution, label each item as one of the following: Clean-loader Cross-loader Non-loader Criterion: <.4 Factor Analys...
View Full Document

This document was uploaded on 02/19/2014.

Ask a homework question - tutors are online