in the goodness-offit measures between the three nonlinear models for M. longissimus α-tocopherol concentration when differences between experiments were not included as random effects. For the Gompertz and exponential functions, the goodness-of-fit increased when experiments were included as random effects in one of the model parameters, whereas for the logistic model,inclusion of a random experimental effect did not improve the goodness-of-fit (Table 4). Using more than one random effect in the parameters did not lead to convergence.