The objective of this work was to determine the method of stand correction more adequate to estimate productivity in cornfields of 33 varieties in eight situations. The methods tested were: without correction, rules of three, method of Zuber, covariance of medium stand, covariance of ideal stand, proposed by Cruz, proposed by Vencovsky & Cruz, the method proposed in this work, which consisted of a stratified correction based on the combined productivity for the characteristic stand, by Scott Knott's test. The methods proposed in this work, covariance of ideal stand and that proposed by Vencovsky & Cruz were considered the most appropriate ones, because they presented lower values for the coefficient of variation and high value of F, which was confirmed by the low values of Pi and Pi multivariate in the combined analysis. The methods rules of three and proposed by Cruz showed the highest averages of productivity, which overestimate the real value of the average production. The method proposed in this work was efficient to correct the productivity of the corn varieties, once the most productive varieties were also those with higher potential to leave descendants, measured by the average stand.
experimentation; data analysis; statistical analysis; stand improvement
