In our thesis, we have used a mixed-effects logistic regression and now we want to present it as a formula, however, we are not sure how to present a mixed-effects logistic regression?

Our binary outcome is if the calf of a specific age group is sick (yes/no) and we have two explanatory variables. The first one is a continuous variable (Cq-values) made to a dichotomous one (true/false under a certain Cq cut-off). The second explanatory variable is a categorical one called age group with four levels = 1, 3, 5, and 7. We also have a random effect that is a combination between herd number (from where a calf is from) and age group (where the calf belongs to) called GroupID.

We have tried to make a formula but it does not seem right when we compare it to others. It is as follows:

logit⁡(p_ij )=intercept+A_ij+slope*x_ij+G_kij where p_ij = is the probability for a calf j in an age group i, being scored as sick given the explanatory variables A_ij is the fixed effect, age group, i = 1, 3, 5, 7 for calf j x_ij is the fixed effect, Cq values being ≤ cutoff, i = true, false for calf j G_kij is the random effect, herd number, k =1…36, in combination with age group, i = 1, 3, 5, 7 for calf j


1 Answer 1


Based on your description, the equation should look something like this:

$$\left \{ \begin{array}{l} \log\frac{\Pr(\texttt{sick}_{ij} = 1)}{1 - \Pr(\texttt{sick}_{ij} = 1)} = (\beta_0 + b_{i0})+ (\beta_1 + b_{i1})\texttt{Age}_{ij}^{(3)} + \\ \quad\quad\quad\quad (\beta_2 + b_{i2})\texttt{Age}_{ij}^{(5)} + (\beta_3 + b_{i3})\texttt{Age}_{ij}^{(7)} + \beta_4\texttt{Cq}_{ij},\\\\ b_i = (b_{i0}, \ldots, b_{i3}) \sim \mathcal N(0, D) \end{array} \right.$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.