Skip to main content

Table 3 Predictive ability of standard (SMM) and recursive mixed models from the analyses of cross-validation subsets

From: Modeling relationships between calving traits: a comparison between standard and recursive mixed models

 

Comparison criteria

Modela,b

  

SMM

RMM1

RMM2

RMM3

GL

     
 

Average mean squared error

19.559

19.559

19.558

19.558

 

Pearson's correlation

0.424

0.424

0.424

0.424

CD

     
 

Average mean squared error

0.824

0.823

0.824

0.823

 

Pearson's correlation

0.448

0.450

0.449

0.450

SB

     
 

Average mean squared error

0.111

0.111

0.111

0.111

 

Pearson's correlation

0.150

0.172

0.170

0.170

  1. aBoldface numbers indicate the best performance by criterion of comparison; bRMM1: recursive mixed model (RMM) assuming that the relationship between residuals is due to the recursive relationships between the gestation length (GL) phenotype and the liabilities to calving difficulty (CD) and stillbirth (SB); RMM2: RMM assuming that the relationships both between residuals and between herd-years are due to the recursive relationships between the phenotype of GL and the liabilities to CD and SB; RMM3: recursive mixed model assuming that phenotypic correlations of the system are uniquely caused by the recursiveness