APPENDIX II 



Formula Used to Calculate Standard Errors for Estimated Registration Rates 



The formula (Cochran, 1963, section 6.11) used to derive standard error follows: 



/ r, 



var - R\ ^ — 

 h h r, 



+ Y, 



- - 2YP, S S 

 n, h m, n, 



h h h 



for individual strata, and for pooled estimates 



r. 



7 \^"R, 

 var(Y) = ER2j^ ^ 



S2 



+ Y 



h n. 



h=l 



h 



2YP, S S 

 h m^^ 



To estimate weighted means for pooled strata, we used 



r. 



m. 



hi 



Z h 



h=l R i^l rh 

 h=l R i=l 



hi 



where 



Yj^^ = estimated registration rate in stratum h; Y = overall estimated rate; 



- actual total number of sample units in stratum h; 



R = E R, , or the total number of sample units; 

 h=l 



r, = the number of selected sample units in stratum h; 



= the number of registered groups; 



m^^^ = the number of registered sample groups from sample unit i, stratum h; 



'hi 



the total number of sample groups from sample unit i, stratum h; 



S2^ = variance of m^^ (and similarly for n^^) 

 Pj^ = estimated correlation of mj^^ and i^^- 



43 



