Closed ayushpatnaikgit closed 1 year ago
Not entirely sure actually
> dclus1 = svydesign(id=~dnum, data = apiclus1, fpc = ~fpc)
> dclus1$allprob
fpc
1 0.01981506
2 0.01981506
3 0.01981506
4 0.01981506
5 0.01981506
6 0.01981506
7 0.01981506
...
> dclus2 = svydesign(id=~dnum, data = apiclus1, weight = ~pw)
> dclus2$allprob
pw
1 0.02954472
2 0.02954472
3 0.02954472
4 0.02954472
5 0.02954472
6 0.02954472
7 0.02954472
8 0.02954472
...
Don't think this is correct https://github.com/xKDR/Survey.jl/blob/735613ee1450c2cfe7731d7739db9a08a829c2e6/src/SurveyDesign.jl#L85
For example, in
dclus1
, it says the sample size is 15, but it's actually 183, the number of rows. Clustering doesn't affect the sample size, only stratification does. I think we should have the else condition.@smishr