Novel approach to the selection of Psidium guajava genotypes using latent traits to bypass multicollinearity

Multicollinearity is a very common problem in studies that employ path analysis agronomic crops, which generates unrealistic results and erroneous interpretations. This study was aimed at assessing the data obtained from guava tree full-sib based on modelling multiple regressions applying latent variables to neutralize effects of multicollinearity. Seven explanatory were measured – fruit mass (FM), length (FL), diameter (FD), mesocarp thickness (MT), peel (PT), pulp (PM), total number fruits (NTF) –, plus main dependent variable, yield per plant (YIELD). In accordance with multicollinearity scenario, eleven values tested addition constant K diagonal correlation matrix X’X. Path applied two models: all direct effect one another model regression more than chain presence variables. The multivariate methodology structural equation (SEM), uses variable prediction, provided better traditional ridge analyses.