Do you know what is different between covariance based and variance based method

In the research we have a flexibility to choose the tools that we will use for the purpose of research achievement. If an analysis tool is not suitable because they do not meet the criteria required, then the alternative may be used. As an example in the application of tools based covariance analysis of data such as Amos and Lisrel. This analysis tool requires that the data used in the analysis must be in normal distribution. If this is not fulfilled, whether we will stop until this stage? Certainly not, Why? because there are other tools available that is based varians analysis tools such as PLS. Do you know what the differences between them? The difference is that Amos and Lisrel objective is to try to reproduce more covariance matrix, while the PLS objective is to try to maximize the variance can be explained.