如何为每个已识别的子组运行 gllamm？

如何解决如何为每个已识别的子组运行 gllamm？

我必须编写一个 ado 文件来使用 gllamm 进行亚组荟萃分析。即使我在每个命令分别为每个子组运行的循环中有 glamm 命令，在结果中 gllamm 执行中包含的研究数量是数据集中的研究总数，因此它不为每个子组单独运行，但为每个子组运行相同的 gllamm 命令。有什么建议可以解决这个问题吗？

local i=1
while `j'<=`ngroups' {  
qui count if (newby==`j')
    if r(N)==0{
    di "Subgroup analysis can not be completed"
    }
    qui sum `1' if newby==`j'
    scalar `k'=r(N)
    di in ye "The number of studies included in this Meta-analysis is " `k'
qui levelsof `2',local(slev)
foreach i in `slev' {
    if `i'<=0 {
        di as err "Varlist s should only contain positive values" 
        exit 125
        }
}
    eq het: `2'
    constraint define 1 `[s1]'`2'=1

    gllamm `1',i(id) s(het) nats constraint(1) level(`l') adapt prior(gamma,scale(10000) shape(2))
  
local j=`j'+1
}
}

解决方法

local i=1
* presumably local j = 1 before this 
* presumably local ngroups defined before this 
while `j'<=`ngroups' {  
    qui count if (newby==`j')
    if r(N)==0 {
        di "Subgroup analysis can not be completed"
        * !!! you need to get out of the loop here 
    }
    qui sum `1' if newby==`j'
    * summarize,meanonly better style 
    scalar `k'=r(N)
    di in ye "The number of studies included in this meta-analysis is " `k'

    qui levelsof `2',local(slev)
    * !!! shouldn't this also be for the current group of newby 
    * !!! why don't you insist here that you only want positive values 

    !!! you're already using i as a local macro name 
    foreach i in `slev' {
        if `i'<=0 {
             di as err "Varlist s should only contain positive values" 
             exit 125
        }
    }

    eq het: `2'
    constraint define 1 `[s1]'`2'=1

    * !!! this is the same every time Stata gets to here 
    * !!! so long as local macro 1 is the same 
    * !!! you may need to slap -if- conditions on the command 
    * !!! Stata won't select a subset just because you are inside a loop 
    gllamm `1',i(id) s(het) nats constraint(1) level(`l') adapt prior(gamma,scale(10000) shape(2))
  
    local j=`j'+1
    }
}