Level Dependent Heteroschedasticity in Factorial Experiments: Aliasing Patterns and Implications