M1 <- logLik(polrModel)  #'log Lik.' -1748.180691 (df=8)
M2 <- logLik(multinomModel)  #'log Lik.' -1734.775727 (df=20)
G <- -2*(M1$1 - M2$2)) #I used a block bracket here in the real code
# 26.8099283
pchisq(G,12,lower.tail = FALSE) #DF is #of predictors
#0.008228890393     #THIS P-VAL TELLS ME TO REJECT PROPORTIONAL