I am working in a software company. I have two questions with me. Basically we will do control charts for ceratin parameters like schedule variance, effort variance etc. We calculate the control limits by mean +/- 1*std dev.
Questions
1. Is it necessary to check the normality or not?
2. Is there any criteria for selecting How many std deviations to consider from the mean for control limits?
3. Which are the control charts are better for this type paramters as sechedule variance, effort variance ?

thanks