Modelling the growth curve of Santa Ines sheep using Bayesian approach

Growth models are used to understand the relationships in production during the life of an animal, being an abstraction of their natural dynamics. In this context, the objective of this research was to fit a curve for weight of Santa Ines sheep using frequentist and Bayesian approaches, present strategies for eliciting prior distributions for the latter and compare the results obtained with each one. Growth data from a literature study was used as sample. The parameter estimates were obtained using nonlinear least squares in the frequentist approach and using Monte Carlo method via Markov Chains algorithms in the Bayesian approach. Noninformative and informative prior distributions were used in the Bayesian approach, with prior information coming from other six studies. A methodology for eliciting informative prior distributions was provided. Prior information contributed to more precise estimates of sheep weight. It was seen that predominance of prior information may produce inconsistent interval estimates. Although the values of the parameters estimated by the two approaches were similar, the use of the Bayesian approach, together with the prior distributions, allowed for good and more precise estimates when compared to the frequentist approach.