Hi I want to find out if there is statistical significant separation between the enzyme group and non-enzyme groups for the CD type I proteins using Principle component Analysis. Using parallel analysis, I found out that 10 PCs should be retained. Using two sample Hotelling's t-squared test and F-statistics, I found out that the enzyme and non-enzyme group in my score plot is separated significantly( I used this paper as my reference for analysis: http://www.sciencedirect.com/science/article/pii/S0169743911001754). My question is: Does Hotelling's T-squared Test for two samples (PC1 and PC2 score values) will give the same F-statistics as using Hotelling's T-squared for 10 PCs ? I am assuming "No". But, can I say that by using first two principle components will give the similar results as using 10 components. I have problem on how to perform Hotelling's T-squared test for more than 3 variables in Excel or SPSS.
@ Angeles Camacho . Based on the paper that I used, http://www.sciencedirect.com/science/article/pii/S0169743911001754, I am trying to used Hotelling's T-squared test ( it measured if there is significant difference between the multivariate means of Enzyme and non-enzyme groups using 10 PCs.) I can calculate the Hotelling's T-square for two PCs. But, I am having difficulty calculating T-squared with 10 PCs.
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Enzymology
- Enzymes