set obs 100
set seed 3755
gen group=_n
gen r=uniform()
gen r0=r if _n==1
replace r0=r0[_n-1] if _n>1
gen y1=2*rnormal()+7*uniform()+20 if r0>0.5
replace y1=1+4*rnormal()+20 if r0<=0.5
gen y2=2*rnormal()+5*uniform()+20 if r0>0.7
replace y2=3+3*rnormal()+20 if r0<=0.7
gen y3=2*rnormal()+6*uniform()+20 if r0>0.6
replace y3=2*rnormal()+3*uniform()+20 if r0<=0.6
drop r r0
stack y1 y2 y3, into (y)
rename _stack g
egen b=seq(),t(100)
anova y g b
regress
predict e,residual
swilk e
gen ee=abs(e)
anova ee g b
数据和我做的步骤，因为残差不齐，需要friedman秩和检验，friedman秩和检验需要装一个插件，怎么装？
需要两两比较，test_b[g[1]]=_b[g[2]]  一直提示1 invalid，求各位尽快解答，这是作业！