I have seen people post statistics questions on here and get some informative answers, so I figured I'd give this one a shot.
I am trying to find patterns in a somewhat large data set (700+ subjects) looking at gait patterns between different groups of individuals. There has been a lot of research using PCA to examine patterns in gait, but a lot of it seems to be very basic (using default programs in the stats programs and a standard recipe, ie, Kaiser criterion for extracting factors even though this is agreed to be a poor option, Varimax rotation even though oblique likely yields more accurate results, very small subject to item ratios). I was planning on doing exploratory factor analysis to examine structure of the data and identify patterns, using recommended best practices for factor analysis.
However, one issue I have is I would like to identify different patterns within my entire sample. So rather than just lumping everybody into one big factor analysis which will provide information on latent variables for the entire sample, I would like to take the data set and see if an analysis can identify different patterns within the population (ie, it finds that there are 8 distinct types of gait patterns that these 700 subjects can be grouped into, rather than finding there are x amount of factors contributing to gait when all 700 are considered). Further, I would like to differentiate between patterns of healthy vs. various pathologies.
I realize some of this sounds like it can be done with a hierarchical cluster technique. I have attempted this, but it seems a bit too "messy" as far as where to draw the line at what constitutes what pattern. I realize discriminating between different gait patterns could easily be done with discriminant analysis, if we pre-define the groups (ie, healthy, cerebral palsy, brain injury). However, I would like to put everything into an analysis and let the analysis define the different patterns, and then be able to compare between groups to see if there is a group that tends to display a given pattern.
Any ideas on the general analysis to do this?
