Panel <- Panel %>% select(country, year,opinion, x1, x2, x3)

Panel_multinomial <- mlogit.data(Panel, shape = 'wide', choice = "opinion", id ="country" , sep = "_")

regression <- mlogit(opinion ~ 1|x1+ x2+ x3, data = Panel_multinomial,
        panel = TRUE, rpar = c(x1= "n", x2= "n", x3="n"), reflevel="1")

regression <- mlogit(opinion ~ 1|x1+ x2+ x3, data = Panel_multinomial, reflevel="1")