A famous example depicting the shortcomings of N-way structure for alternatives that are correlated
Consider a city where 50% of travelers choose car (C) and 50% choose bus (B)
Assume: CC=CB
In a stroke of genius, the manager paint half the buses red (RB) and half the buses blue (BB), while maintaining the same level of performance: CRB=CBB=CC
Thus, the probability of the car using logit formulation is now: PC=exp(−βCC)+exp(−βCRB)+exp(−βCBB)exp(−βCC)
Which given the above assumptions, reduces the probability of a user choosing a car to decrease of 50% to 33%
Applying the nested structure to this problem gives a more reasonable result: PC=1+exp(−λ1(CB−CC))1 where PB=1−PC; PRB=1+exp(−λ2(CBB−CRB)1; PBB=1−PRB; CB=λ2−1log(exp(−λ2CRB)+exp(−λ2CBB)); And λ1 and λ2 are the primary and secondary split parameters
If CB=CC, this model will correctly assign 50% to the car and 25% to each bus options