REDUCING TIME IN RELIABILITY STUDIES 



187 



had an effect on the efficiency. To make the procedures more comparable 

 the values for r = 3 and r = 4 in Table I were averaged with values p 

 and 1 — p computed so as to give a probability of exactly 0.800 at a = a*. 

 The corresponding values for the average experiment time were then 

 averaged with the same values p and 1 — p. The nonsequential pro- 

 cedures so altered will be called "adjusted procedures." The efficiency 

 of the adjusted Ri relative to Rz is given in Table VI. 



In Table VI the last row gives the efficiency of the adjusted procedure 

 7^2 relative to Rz . Thus we can separate out the advantage due to 

 the replacement feature and the advantage due to the sequential fea- 

 ture. Table VII gives these results in terms of percentage reduction of 

 average experiment time. 



We note that the reduction due to the replacement feature alone is 

 greatest for small n and essentially constant with a while the reduction 



Table VII — Per Cent Reduction in Average Experiment Time 

 DUE TO Statistical Techniques 



(a* = 2,P* = 0.800, ^ = 0) 



