Results of GLM with negative binomial distribution and log-link function for count of lethal acute toxicosis as predicted by herd-cohesion-factor (HCF) and stocking-density (SD).

β coefficients are from the same GLM without the interaction present. GLM fit: Fisher scoring iterations = 1; residual deviance = 516.94 on 476 degrees of freedom; AIC = 1686.3.此处的β系数取自未纳入交互项的同一广义线性模型（Generalized Linear Model, GLM）。该模型的拟合结果如下：费希尔计分迭代次数为1；残差偏差为516.94，自由度为476；赤池信息准则（Akaike Information Criterion, AIC）值为1686.3。