A model for failure of a chicken embryo to survive incubation

Proper assessment of factors contributing to failure of an egg to hatch, i.e., infertility and embryonic mortality, is important in poultry production. A model consisting of the sum of two cumulative logistic distributions was proposed previously to describe the distribution for time of mortality during incubation; model parameters, including probabilities of infertility and mortality, were estimated by the method of least squares. The objective of this paper was to improve the previous model and method of estimation by evaluating alternative distributions and methods; we propose four recommendations. First, probabilities of infertility and mortality should be estimated as observed proportions rather than as model parameters. Second, parameters of the distribution for time of mortality should be estimated using a diphasic Weibull distribution rather than a diphasic logistic distribution. Third, parameters of the distribution for time of mortality should be estimated using noncumulative proportions rather than cumulative proportions. Fourth, parameters of the distribution for time of mortality should be estimated by maximum likelihood rather than by least squares. The minimum Hellinger distance, however, is a good alternative to maximum likelihood to estimate distribution parameters if the distribution of mortality is not known exactly or if the data contain outliers.