Question: Where we use data normalization in multivariate analysis.
Situation: When I analiyse all descriptives statistics I don't find any problem, but the multivariate normality test in LISREL does not run. Additionaly when I try to use Robust Method and run covariance matrix and asypmtotic covariance, apparently its determinant is not possitive and it is not possible to calculate the model.
Someone told me that I can solve the situation using data normalization in Lisrel, but I have no support to do it.
Normalization is a family of transformations that are computed sample-wise. Its purpose is to “scale” samples in order to get all data on approximately the same scale. It is applied in cases where the data are collected with a method (system) where the detector signal is a function of sample mass (i.e. most GC detectors) or of source power (i.e. Raman spectroscopy) instead of sample concentration.
Few types of normalization are area, unit vector, mean, maximum, range and peak normalization
Dear Camilo
1. Once one want to aggregate the parameters by any suitable options say index and the parameters of interest are having different range of values (varied scale), which may distort the true significance of aggregation. So Normalisation is simply putting your all parameters range same so that relative contribution has been observed in more pragmatic ways.
2. In multivariate analysis, you may use any one transformation suitable to your data. this may be square, log, or standard normal variate. In this case, you wish that your data should follow Normal distribution.
Regards
Simply, do not normalize your data
Article TO DETERMINE SKEWNESS, MEAN AND DEVIATION WITH A NEW APPROAC...