How to interpret fixed effects from a binary logit model in which all dependent variables are categorical? Marginal effects does not work

The study is on survey data from UK households with a lot of missing variables. I ran a logit regression to predict the probability of Y (binary variable)and estimated the corresponding marginal effects to interpret the coefficients. However, When I ran the fixed effects (to account for individual households effects) I cannot interpret the coefficients through marginal effects because marginal effects assume that either:

(1) There are no fixed effects (2) Linearity of the independent variables. <-- Which is not the case given that they are all categorical