An issue of guaranteed representation in a variant of Hamilton's method (Q6596264)

This paper studies a variant of the Hamilton apportionment method that guarantees representation for all candidates who receive more than a certain percentage of the vote in an election with no candidate with 50\% or more votes. This is done by first apportioning one delegate for each of these candidates and then dividing the remaining positions by Hamilton method. The paper studies the probability that Hamilton's method and this guaranteed representation version of Hamilton's method produce different apportionments, even when Hamilton's method gives at least one delegate to each candidate. This question is answered applying a geometric development for the case of two or three candidates and any number of delegates, assuming equally likely election outcomes. For more than three candidates, simulation results show how this probability diminishes with increasing the number of candidates. Simulation results determine also how increasing the threshold to eliminate candidates will affect the value of the probabilities of the two methods produce the same apportionment.