I’m analyzing six popular brands with respect to seven attributes on the scale of 1 to 10 where 1 stands for Least Preferred and 10 stands for Highly Preferred. I have taken these seven parameters (independent variables) from well-established scale. While running discriminant analysis I observed high correlation within independent variables (ranging between 0.879 and 0.715) in Pooled Within-Groups Correlation Matrix. So, I can see multi-collinearity is likely to be a problem. Please suggest.
If the variables are numeric you can use correlation and then clustering to pair all the variables highly correlated to each other.
Thanks for your reply Dr. Safdar. As i shared there is high correlation (ranging between 0.879 and 0.715) between these variables and clustering them is again giving similar results.
Multi-collinearity is always refereed to how a respondent perceive/ understand two aspects (variabl;es) into the same contextual meaning. For this, Variance Inflation Factor (VIF) is calculated by using correlation coefficient (R) of the two aspects (variables), where VIF =1/(1 - R2). Generally, if VIF >= 5, it is meant there is a problem of Collinearity for aspects/variables. The Multi-collinearity issue can be simply inferred with the correlation coefficient R, i.e., if approximately R>=0.9 (for positive relationship of the aspects/variables) or R=5 for Multi-collinearity, then this can be said that (1 - R2)= 5 in calculation. This implies that
(1 - R2)= 0.81, the R >= 0.9 (for + relation) or R
- Badges
- Science topic
- Similar topics
- Biological Science
- Kinesiology
- Running