Statistical analysis
Each experiment, including soymilk preparation and chemical
analyses, were replicated twice. Completely randomized designs
were used for the flavor contents and sensory scores in soymilk.
Each number represents the mean of triplicate tests. The data were
subject to analysis of variance (ANOVA) using SPSS 16.0 package
(SPSS 2007). Significant difference among treatments were analyzed
using Duncan’s multiple test with a probability level of 0.05
or less.