Multiple linear and nonlinear regression analyses of factors causing calving difficulty.

Calving difficulty (CD) was analyzed as the dependent variable from observations on 592 first-calf, 2-year-old heifers that were either Angus X Hereford (AH), Charolais X AH or Simmental X AH. Independent variables were birth weight (BW), pelvic area (PA), BW.(33), PA.(5), cow weight (CW), cow condition score (CS) and sex of calf (S). CD was analyzed as score (1 = no difficulty through 4 = extreme difficulty) CD-I, % CD (1 vs 2 + 3 + 4 , CD-II) or % high CD (1 + 2 vs 3 + 4 , CD-III). R(2) values from CD-II and CD-III multiple correlation (R) analyses were consistently lower (.10 to .13) than from the CD-I analysis. When only BW and PA were included in the model, R(2) values for CD-I were .36, .39 and .39 for the linear, quadratic and cubic analyses, respectively. When BW, PA, CW, CS and S were included in the model, the R(2) values for CD-I were .40 and .42 for the linear and quadratic analyses (cubic effects were not tested). In the linear analysis with BW.(33) and PA.(5), the R(2) for CD-I was .36. In all analyses, the order of importance of the factors studied was BW and PA, with a much lower contribution from CW and S. CS did not account for a significant increase in R(2) in any analysis. CD remained below 50% whenever the PA BW ratio exceeded 8.4. This ratio is a convenient practical method to relate PA and BW to CD. Conclusions are that the linear effects of BW and PA account for the majority of the identified variation in CD (R(2)) and that small increases in R(2) can be obtained by adding CW, CS and S to the model and/or by expanding the model to include quadratic effects.