Although factor analysis (MCA and PCA) are used to detect and represent underlying structures in a data set, can we apply these statisticals tests to justify the heterogeneity of the samples and consequently the impossibility of constituting similar groups in the studied population ? This will be indicated by: low correlation between variables and between individuals, large number of the obtained factors and the low inertia provided by each factor.
Thanks for your reply.
Dear Djallel Eddine Gherissi ,
Although I don't understand all the terms you have used in your question I am still going to venture an answer:
- High heterogeneity between sample members will obviously make it difficult to detect factors. It may also indicate that your questions do not have a sufficient range of distinctive responses.
- It is the relative correlation within and between potential factors that is important, i.e. items in a factor need to have a higher average correlation with each other than the average correlation between factors. This might lead to different EFA strategies, e.g. oblique rotation instead of orthogonal.
- You should also test the adequacy of your sample size.
Please see my study guide for more information.
I think it can not. Beacuse low explained variance stem from not only heterogenetiy but also your sample. In addition, it stems from your scale or assesment tool.
Although factor analysis can be used for examine relationship among individuals scores, it is used to examine the relationships between variables more generally. Because when it is necessary to examine the relationships between individuals, a lot of variables will be needed. (Q type and R type factor analysis).
Best
No.
For that to work you would need to show that your sample size was sufficient based on the variability in the data and show that you had measured (or controlled for) all relevant variables.
- Badges
- Science topic