You have a five marker, and two box-plots. What do you compare?
Why isn’t stratifying by what you are investigation a good idea?
because you can artificially bias results
if the calculated skew from a data set is positive, what does that mean about the data spread about the median?
data below the median is less spread than data above.
is using names-out-of-a-hat appropriate sampling method for 650 names?
No. 650 name sin a box woul not mix very well and therefore would not give every name and equal chance of being selected.