Analysis of Random Intercept and Slope Model (RISM) for Data of Repeated Measures from Hy-Line White Laying Hens

In animal science, sequential variation on quantitative traits during a certain time period should be precisely identified for regulating managerial conditions in animal experimental data. This study was conducted in order to investigate the effect of including some covariates on performance of covariance structures, fixed and random effects on the scope of random intercept and slope model (RISM) in order to improve model quality criteria. In repeated measurement data of laying hens, cumulative egg weight (CEW) per hen as a dependent variable was recorded per week, and treatment, time and treatment x time interaction effects were added as independent variables. Time effect was considered as a continuous variable in RISM. For better improving quality of RISM, feed intake (FT), feed conversation ratio (FCR), and egg mass (EGGM) per week were also included as covariates. Model quality criteria like-2 Res Log Likelihood, Akaike's Information Criterion (AIC), Bayesian Information Criterion (BIC), and Corrected Akaike's Information Criterion (AICC) criterion were used to identify best covariance structure among Compound Symmetry (CS), Heterogeneous Compound Symmetry (CSH), Unstructured (UN), First-Order Autoregressive (AR(1)), Unstructured correlation (UNR), Heterogeneous First-Order Autoregressive (ARH(1)), Toeplitz (TOEP) and Heterogeneous Toeplitz (TOEPH) with/without adding covariates. The explanation proportion of 90% in the dependent variable (CEW) was estimated for CSH, UNR, ARH(1), TOEPH, and UN as an outcome of adding covariates, which was prominently higher than the RISM without adding covariates. The significant differences in parameter estimates of fixed and random effects were recorded between the RISM with and without covariates. In repeated measures design, adding covariates in improving quality criteria of RISM could be recommended for data of laying hens