Journal on Policy and Complex Systems
pairs of means : between Policy 1 ( No Management ) and Policy 3 ( Size ) ( d = 0.24 ), between Policy 2 ( Supply & Demand ) and Policy 3 ( d = −0.30 ), between Policy 3 and Policy 4 ( Price ) ( d = −0.31 ), and between Policy 3 and Policy 5 ( Size & Price ) ( d = −0.34 ). There was a medium difference between the following pairs of means : between Policies 1 and 2 ( d = 0.69 ), between Policies 2 and 4 ( d = −0.76 ), and between Policies 2 and 5 ( d = −0.79 ).
Medium price . The number of medium-price producers ( green bars in Figure 6 ) was significant ( F ( 4,4995 ) = 57.8 , p < 0.001 ) across the five different policies . In the figure , means that do not share a ( lowercase ) letter are significantly different , as calculated by post-hoc analysis . There were small differences between the following pairs of means : between Policy 1 ( No Management ) and Policy 2 ( Supply & Demand ) ( d = 0.46 ), between Policy 2 and Policy 3 ( Size ) ( d = −0.30 ), between Policy 2 and Policy 5 ( Size & Price ) ( d = −0.29 ), between Policy 3 and Policy 4 ( Price ) ( d = 0.43 ), and between Policies 4 and 5 ( d = −0.42 ). There was a medium ( d = 0.59 ) difference between Policies 1 and 4 .
High price . The number of high-price producers ( red bars in Figure 6 ) was significant ( F ( 4,4995 ) = 9.41 , p < 0.001 ) across the five different policies . In the figure , means that do not share a ( Greek ) letter are significantly different , as calculated by post-hoc analysis . There was a small ( d = 0.21 ) difference between Policy 1 ( No Management ) and Policy 4 ( Price ). There was also a small ( d = 0.22 ) difference between Policy 3 ( Size ) and Policy 4 .
Discussion
Of all five sourcing policies , Policy 2 ( the unmodified supply−demand selection policy ) is the strictest . However , Policy 4 ( Price ) is effectively nearly as strict as Policy 2 , since relatively few low-price producers attempt to participate in the food hub ( a reflection of real-life producer behavior ), and therefore an exception for these producers does not relax the supply−demand ratio constraint very much . By contrast , Policies 1 , 3 , and 5 represent less interference by the food hub manager . Based on the results of the ANOVA tests , all of the system performance metrics of interest were significantly influenced by the food hub manager ’ s choice of supplier selection policy . However , the effect sizes varied considerably for different pairwise comparisons of policies on each metric , which tended to reflect the difference between less restrictive approaches ( as with Policies 1 , 3 , and 5 ) and strategies that were more selective ( i . e ., Policies 2 and 4 ).
Under Policy 2 ( Supply & Demand ), both the mean number of food units traded through the food hub and the number of participating consumers are shown to be significantly less than the mean values under Policies 1 and 5 ( No Management and Size & Price , respectively ). These results suggests that , in terms of sales , the best option for the food hub may be to continue allowing any producer
42