BIO3011 Lecture Notes - Lecture 8: Box Plot
BIO3011 Lecture 8 Post hoc tests
• Problem of running month as a predictor variable -> form of numbers -> r will think is a
linear relationship -> may not be linear
• Before anything else, not a bad idea to plot boxplot
• If don’t think its linear then turn into a factor eg look at month
• If significant then boxplot and to a tukeys test
• When plotting a tukeys test, if CI crosses 0 then not significant
-need to get rid of y axis title and put something sensible eg. 6 (x10-12)
-April is nonsignif to everything except august
• for every extra comparison, p values are penalised -> adjusted -> can be unreliable
A,B
A,B
A
A,B
A
find more resources at oneclass.com
find more resources at oneclass.com