problem-7.29
problem-7.29
The data set implies that the population distribution is skewed
and not normally distributed or symmetric. Thus t.test() and
wilcox.test() are not appropriate. However, after a log
transform, it appears that wilcox.test() will be.
> wilcox.test(log(commutes), conf.int=TRUE, conf.level=0.9)
...
3.135 3.258
...
> exp(c(3.135,3.258)) # transform back
[1] 22.99 26.00
Or, that [23,26] is the confidence interval. The warning message is
about the ties in the data. Notice that the sampling statistic is based
on an assumption of a continuous distribution.