22 MASS. EXPERIMENT STATION BULLETIN 389 



The tendency for uniformity throughout the milkshed in the prices paid by 

 any particular handler is further borne out by the analysis on a dealer basis of 

 differences in prices received by shippers. 



The pronounced variation in prices received by shippers would tend to dampen 

 any notion that chance could have been a substantial contributory cause of the 

 differences. On the other hand, however, one could not necessarily assume that 

 the variations were predetermined in some clandestine fashion. One might 

 suspect, if he were familiar with the market, that the organization of the market 

 and the arrangements between handlers and shippers would be a basic considera- 

 tion. Testing the assumptions by analysis of variance gave results which indicated 

 a high degree of validity. 



Analysis showed upon systematizing shippers' prices that^ 



1. Adjusting values to a comparable 3.7 percent test removed about half 

 of the variation in the prices; 



2. Adjusting next for trucking charges removed a slight amount of the 

 remaining variation; 



3. Adjusting further for seasonality had practically no effect on the 

 variation remaining at this stage. 



The effect of ratings on price depended upon whether or not the dealer planned 

 on having the sum of his shippers' ratings approximate — • 



1. His sales, 



2. His purchases. 



When ratings approximated sales, the effect of ratings was a significant cause 

 of variation in shippers' prices. (Table 16A.) 



When ratings did not approximate sales (and therefore probably approximated 

 purchases), ratings had no effect on variation in price. (Table 16B.) 



The effect of these variables should not cloud the importance of differences in 

 dealers' product-costs (based on use) as a major cause of differences in shippers' 

 prices. In arriving at the effect of butterfat, trucking charges, seasonality, and 

 ratings on shippers' prices, dealers' product-costs were held constant. The 

 importance of dealers' product-costs as a cause of variation is well emphasized 

 by the data in table 15. 



The prices received by 1174 full-time shippers of Grade B milk were used in 

 making the analysis of variance. The results were highly significant with each 

 type of price used. Not only was the variation in prices on a dealer basis mathe- 

 matically significant for each type, but with progressive refinement of the prices 

 the significance of the dealer classification became increasingly pronounced. 



With the variance ratio "F" 1.61 or greater, there was one chance in a hundred 

 that differences in dealer classification were not a major cause of price variation. 20 



Since the value of "F" was 78.6 with unrefined prices, the probable importance 

 of dealer grouping was immediately emphasized. Standardizing prices to a com- 

 mon 3.7 F.O.B. basis raised the value of "F" to 210.0 (Table 15). Removal of 

 these causes of variation, by expanding "F", helped clarify and accentuate the 

 probable influence of forces which were peculiar to dealer groups. 



From an inspection of the Mean Squares in table 15 one can also draw some 

 conclusions concerning the relative effects of butterfat test and of trucking charges 

 on shippers' prices. The amount of variation in shippers' prices within dealer 

 groups was reduced nearly 50 percent after adjusting the prices to a comparable 



2»Snedecor— p. 177, Table 10.2— Statistical Methods. 



