Growth curves with multivariate θ generalized normal distribution for cardiac dysfunction in rats

Abstract

There is a necessity to study the behavior of some characteristic in the same sample unit over time in many situations, as accumulated dose of some nutrient. In practice, data structure of this nature generally establishes nonlinear behaviors in the parameters. The objective of this work is to accomplish a comparative study on the performance of four models of growth curves in four experimental groups, considering multivariate θ generalized standard errors with homoscedastic structure with the presence of lag-1 autocorrelation. Maximum likelihood estimation is used and simulation techniques are implemented to prove the methodological attributes used the experiment.