Hi guys,
Any help would be immensely appreciated.
So imagine I have a dataset with columns: 1)names of 1200 people 2)their level of education (School/College/University) 3)their level of adaptivity to online studying (Low/Moderate/High]. I am considering the dataset my population in this case. Suppose the proportion of school:college:university students is 4:1:5.
Now, I want to use stratified random sampling to estimate how many percent of population have Low/Moderate/High adaptivity. The strata are the made based on level of education. I then could generate random realization by taking 25% of each strata (proportionate stratified r.s.) or by taking 100 realizations of each type of stratum.
I know both estimators unbiased and now I need to find the variance.
If I generated 5000 realizations I'd have a 3x5000 M matrix (each column are percentages of adaptivity levels within a sample), then var(M) would be also a 3x5000 matrix and that wouldn't tell me anything...
Could I just do variances for every type of adaptivity and sum them?
What one should do with such multi-dimensional estimators? I cannot find anything on the internet neither in books.
Thank you so much<3
Any help would be immensely appreciated.
So imagine I have a dataset with columns: 1)names of 1200 people 2)their level of education (School/College/University) 3)their level of adaptivity to online studying (Low/Moderate/High]. I am considering the dataset my population in this case. Suppose the proportion of school:college:university students is 4:1:5.
Now, I want to use stratified random sampling to estimate how many percent of population have Low/Moderate/High adaptivity. The strata are the made based on level of education. I then could generate random realization by taking 25% of each strata (proportionate stratified r.s.) or by taking 100 realizations of each type of stratum.
I know both estimators unbiased and now I need to find the variance.
If I generated 5000 realizations I'd have a 3x5000 M matrix (each column are percentages of adaptivity levels within a sample), then var(M) would be also a 3x5000 matrix and that wouldn't tell me anything...
Could I just do variances for every type of adaptivity and sum them?
What one should do with such multi-dimensional estimators? I cannot find anything on the internet neither in books.
Thank you so much<3