You state the null, but not any alternative.
This thread focused on -anova-, which
is reasonable for that null, but not
so natural for any ordered alternative.
In addition to -nptrend-, there is the Jonckheere-Terpstra
test as implemented by Joseph Coveney as -jonter- from
SSC. (I have a Mata-based version not yet public.)
However, there is a much larger question. Why the
grouping? What's wrong with regression or other modelling
of weight as a function of height?
Nick
[email protected]
[email protected]
> I want to check if decile groups have significantly different
> values for
> property B if you sort them by property A.
>
> Example:
>
> A group of 100 people, sort according to height, create 10
> groups according to
> deciles. For each resulting group 1 to 10 then calculate the
> average weight.
> The null hypothesis I would like to check would be
> weight1=weight2=....=weight10
>
> Any suggestions on the test statistic and how to implement in Stata?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
