Kitada and Tezuka: Survey designs for estimating recreational fisliery catch 



239 



Y-Y.. 



n.>Ni-niN2 



(>'i-y2'- 



showing that the two methods provide different estimates 

 with the extent of the difference, depending upon the 



sizes of the strata, the sample sizes, and the estimates of 

 the sample means. According to Cochran (1977), method 

 2 works well enough if the sample allocation is propor- 

 tional because a simple random sample distributes itself 

 approximately proportionally among strata. With propor- 

 tional allocation N^/n^ = NJn.^; therefore the difference of 

 the two estimators is 0. In our case study, the annual 



