Linearity: The relationships among variables are linear. Multivariate Normality: The variables are normally distributed. Sample Size: A sufficiently large sample size is required, typically at least 5-10 times the number of variables. Correlation: There should be meaningful correlations among the variables.