SELECTING THE BEST OXE OF SEVERAL BINOMLA.L POPULATIONS 00/ 



ALTERNATIVE SPECIFICATION 



If the experimenter has some a priori knowledge about the processes, 

 then he will prefer to specify the following three quantities in order to 

 determine the number n of observations he is required to take from 

 each process. 



Specification: He specifies pm and p[2] (0 ^ p[2] ^ 

 ^ 1) in the neighborhood of his estimate of the 



4: 



P[l] 



probabihties associated with his processes. He also speci- 

 fies a probability P* (0 ^ P* < 1) that he would hke 

 to guarantee of making a correct selection whenever the 

 true p[i] ^ p[i] and the truep[2] 



< 



* 



(12) 



Table VI — Number of Units Required per Process to Guaran- 

 tee A Probability of P* of selecting the Better of Two Bi- 

 nomial Processes when the True p[i] ^ p* ] and the True 

 P[2] ^ p*2] . (Alternative Specification, k = 2) 



Table VII — Number of Units Required per Process to Guar- 

 antee a Probability of P* of Selecting the Best of Four Bi- 

 nomial Processes when the True p^ ^ pfu and the True 

 P[2] ^ p*2] . (Alternative Specification, k = 4) 



