I have data on 30 mental health symptoms assessed upon admission to psychiatric care and every 3 months afterwards. So individuals in the sample have been assessed between 2 and 10 times. I am looking at multiple factor analysis as a procedure for data reduction. I was wondering if it is possible to perform a clustering procedure to create groups based on results (factor scores) from a MFA? Thanks in advance!
Factor analysis (FA) is a suitable method to reduce your raw data's number of columns/symptoms, whereas clustering (CA) divides your sample into subgroups.
In FA the reduced data has representative features and variables have some correlation between them, whereas in CA the reduced data has representative cases, don't have outliers and the variables are not correlated.
Factor analysis will reduce data dimensionality by obtaining a few linear combinations of the data (factors) that explain a large proportion of the variation in the data. Generally the factors are constructed to be orthogonal or independent, so if you apply cluster analysis on the factors, the cluster algorithm will try to obtain groups of factors that have minimum distance, however the factors were constructed to be independent, so it might not be very effective. If your aim is to look for structure/groups, you should apply cluster analysis directly on the original data.
- Badges
- Science topic
- Similar topics
- Psychology