STA221H1 Lecture Notes - Lecture 61: Factorial Experiment, Box Plot, Aisle
25 Apr 2018
School
Department
Course
Professor
Document Summary
In order to determine whether the statistical inferences that were made are valid, check the required assumptions for random errors. Plot each of the following residual plots for each factor- level combination. State the assumptions about random errors that each of these residual plots checks. See slides 12, 26 to 28 for stating and checking these required assumptions: side-by-side boxplot of residuals for each factor-level combination. Assumption 1: the response distributions for each factor-level combination (treatment) is normal. Aligned box plot of residuals for each factor-level combination demonstrate that not all treatment appear to be normally distributed. However, anova is very robust when the assumption of normality is not satis ed exactly. That is, slight to moderate departure from normality do not have much e ect on the signi cance level of anova f-test: plot of residuals verses tted values within each factor-level combination. Assumption 2: the normally distributed response distributions have equal (constant) variances for all treatments.
