Cattell's scree test is one of most simple ways of testing the number of components or factors in a correlation matrix. Here we plot the eigen values of a correlation matrix as well as the eigen values of a factor analysis.
scree(rx,factors=TRUE,pc=TRUE,main="Scree plot",hline=NULL,add=FALSE,sqrt=FALSE)
VSS.scree(rx, main = "scree plot",sqrt=FALSE)
a correlation matrix or a data matrix. If data, then correlations are found using pairwise deletions.
If true, draw the scree for factors
If true, draw the scree for components
if null, draw a horizontal line at 1, otherwise draw it at hline (make negative to not draw it)
Title
Should multiple plots be drawn?
If TRUE, take the sqrt of the eigen value before plotting
William Revelle
Among the many ways to choose the optimal number of factors is the scree test. A better function to show the scree as well as compare it to randomly parallel solutions is found found in fa.parallel
Following a suggestion from Marco Del Giudice, I added the sqrt option for version 2.2.12. -
fa.parallel
VSS.plot
, ICLUST
, omega
scree(attitude)
#VSS.scree(cor(attitude)
Run the code above in your browser using DataLab