|
When the ADG
predictions were based on the SRNS estimates of MEI and on the ME
requirements for maintenance and gain of Luo et al. (2004), the RMSEP
increased to 34.9 g/d, mainly due to a fairly large systematic bias (19.2% of
the MSEP), which made the regression line significantly different (P <
0.03) from the equivalence line (Y = X) (Figure 3, right). In particular,
this model over predicted the ADG at high observed ADG, and under predicted
the ADG at low observed values
|