Multivariate analysis of influence of coffee intake on animal weight

Abstract

The multivariate analysis methodology does not impose any constraints, as there is no assumption about the structure of the variance and covariance matrix of the errors, making it a plausible solution for the longitudinal data set. The objective of this work was to analyze, by the Mauchly test, if the estimated structure of this matrix has a specific structure called Huynh and Feldt, as well as to use the multivariate analysis to evaluate if there was significant interaction between the temporal periods and the six diets Used to feed animals (rats) in an experiment developed at the Department of Pharmacy of the Federal University of Minas Gearis. Wistar rats were submitted to six feeding diets, using a completely randomized design with 10 replicates. The animals were weighed weekly for 15 weeks after the fourth week of birth. The results showed that animal weights are correlated over time. When applying the Mauchly test it was diagnosed that the structure of the covariance matrix did not meet the condition defined by Huynh and Feldt, which justified the use of multivariate analysis methodology, and this analysis indicated a significant interaction between the six diets and the periods Evaluated. In the unfolding of the interaction it was evidenced that the use of coffee in the diet of normonutrient rats resulted in a lower average weight gain of the animals when compared to the same diet without the use of coffee.