As was pointed out on page 308 , two ways of increasing the size of a two-way classification experiment are

(a) to double the number of blocks,

(b) to replicate the entire experiment. Discuss and compare the gain in degrees of freedom for the error sum of squares by the two methods.

Data From Page 308

Transcribed Image Text:

Level & rejection regions for testing p = po when B = specified bound on Type I error probability Alternative hypothesis p po P Po P Po Rejection region Xx where x1 is the largest integer with B(x1; n, po) B Xx2 where x2 is the smallest integer with 1 - B(x2; n, po) B Xx or X x2 where x1 is the largest and x2 the smallest integer with B(x1; n, po) B and 1 - B(x2; n, po) B