SOISOS

Preparation and EDA
Data should be standardized in factor analysis
scale(crime.dat)

#standardize data
crime.dat.sd= scale(crime.dat)

To obtain number of factors to use for the factor analysis, PCA can be used
#PCA for EDA
crime.pca<-princomp(crime.dat.sd)

Bartlett scores
crime.fa.s<- factanal(crime.dat.sd, 3, scores="Bartlett", rotation="varimax")
crime.fa.s$scores

Factor Analysis

#arg(standarized data, no of factors, rotation)
factanal(crime.da.sd, 3, rotation="none"

Hypothesis Testing
H₀: The number of factor is sufficient
H_a: The number of Factor is not sufficient

Decision Rule
Reject H₀ if test statistic > Chiq(alpha, df) or if p-value is small
[Note: the test statistic and p-value can be obtained from R output of factoranal]