Extreme outliers should be deleted (multiple choice, only pick one):

A). Always; delete all cases with residuals greater than 3 in absolute value

B). Sometimes; if well-reasoned and transparently noted

C). Never; keep all data values, even if they are typos or illogical

Possible benefits of transformations include (multiple choice, check all that apply):

A). May serve to stabilize variance of residuals

B). May eliminate dependence of residuals

C) .May correct non-linearity of relationship between the response variable and the predictor variables (e.g. while the relationship between y and x may not be linear, the relationship between y.t and x.t, the transformed variables, may be linear)

D). May serve to normalize model residuals

E). May mute outliers (e.g. points that are unusual on the original scale may be less extreme on the transformed scale)