Experimental Pricing As an Approach to Demand Analysis 27

These estimates turned out to be

 b7 = 2.009675
 bs = 2.288554
 bg = -5.584685
 blo = 2.120746
 ASSR7,8,9,10 = 6.406250

 The Determinate Model.-From outcomes of the preceding
tests, modified form (C) was inferred to be the best fitting
model.25 Accordingly, the final version of the model could be
appropriately written

 Yik-i- = 0 + Si + i + 'kj-- + 2.00 2.288554XODk-j-


 5.584685 Xi_ + 2.120746X;O1k-j- + .;k-j-.

 Estimates of the regression constant (p) and of the class
 constants (s6i, .is and .ji-'s), designated below, supplied the finish-
 ing touches to the specific model used to summarize the demand
 for orange concentrate for the given experimental data.26

 57 A, ^A
 0 = 5.71766 -1_- = -0.01042 a = -0.01740
 As ^ A,
 81 -= -0.02494 2-1- = -0.02874 a2 = 0.02513
 Ai Ai A,
 2 = 0.04973 3-2- = -0.01451 a3 = -0.00292

 3 = -0.21705 = 0.02505

 64 = 0.15449 5-1- = 0.03158

 5 = 0.12897 62- = 0.02374

 6 = 0.01189 7-1- = 0.00757
 A, Ag
 67 = -0.01709 w8-2- = -0.00924

 68 = -0.01502 e_ = -0.02502

 69 = -0.02444

 610 = -0.04662

 "2 The reduction in sum of squares due to fitting modified form (C) was
 10.368225, representing a proportional reduction in variance of 0.986. How-
 ever, the method of estimating the missing observations greatly increased