Describing interactions in dystocia scores with a threshold model.

Field data on calving difficulty scores provided by the American Simmental Association were subjected to two methods of analysis: ordinary least-squares analysis and maximum likelihood with an assumed threshold model. In each analysis, the model included the interaction of sex of calf X age of dam. This interaction was readily apparent in the data (observed scale): within the youngest dams 58% of the heifer calves and 37% of the bull calves were born unassisted vs 96% and 92%, respectively, in the oldest dams. The objective was to determine if this interaction would be greatly reduced or would disappear on the underlying scale of a threshold model. The least-squares estimate of the sex difference was greatest within the youngest age-of-dam group (18 to 24 mo) and steadily declined with increasing age of dam, approaching zero for dams 6 yr and older. In contrast, the estimates of the sex difference from the threshold analysis were remarkably similar across ages of dam. It was concluded that observed interactions in calving ease data could be adequately described by a threshold model in which the effects of age of dam and sex of calf act additively on the underlying variable.