Abstract

RAMIRO VASQUEZ, Edison; CABALLERO NUNEZ, Alberto  and  HERRERA VILLAFRANCA, Magaly. Transformation of binomial variables for analysis  according to a randomized block design. cultrop [online]. 2017, vol.38, n.1, pp. 108-114. ISSN 1819-4087.

The research aimed to evaluate the transformation of variables with Binomial distribution applied to randomized block design. For the analysis was considered 5, 10 and 30 observations per experimental unit (n) and probability of success of the event (p) of  0,10; 0,20; ...; 0,90. Through the Monte Carlo method, 100 experiments with 3, 5 and 9 treatments (t) were simulated; 4 and 8 replicates (r); with 5, 10 and 30 observations per experimental unit (n) and probability of success of the event (p) of 0,10; 0,20; ...; 0;90. It was evident that the arcsine and family transformations Box- Cox power transformation angle: natural logarithm, square root and inverse were not able to solve the deviations with respect to the assumptions of ANAVA. It should be mentioned that in recent years little input regarding such research are made

Keywords : ANAVA assumptions; Monte Carlo; simulation; dichotomous variable.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )