The major purpose of vegetable crops production systems is the profitable and economically viable yield of marketable products. Numerous field experiments with vegetable crops are carried out to determine biomass of weekly and heads production of marketable and non-marketable variables. However these variables possess the particular characteristic related to their presence or absence of the commercial structure of the plant at harvest time, what implicates in zero values for those variables, generate variability and affect variance analyses. This paper deals with the search of applicable transformations for variables studied in broccoli field experiments conducted in different planting times attempting to stabilize the variabilily resulting from the presence of zero values in multiple harvests. Due to the presence of zero values, broccoli experiments were analyzed using an adaptation of the Box-Cox transformation family found in Yamamura in 1999. The most adequate transformation found for marketable and non-marketable heads and side shoots of broccoli grown in winter, spring and summer was the logarithmical. For marketable and non-marketable heads and non-marketable side shoots for fall-grown broccoli the inverse fourth root and to marketable side shoots is a inverse square root was the indicated transformation.
Transformation of data; Box-Cox; variability; analysis of variance
