There are two available packages for estimating multinomial logit models in R, namely mnlogit and mlogit. I am wondering whether someone who used both can share his experience. Which one is better / more commonly used? Are there some advantages to using one of them? How do the packages deal with pannel data?