On using the DomWorld model to evaluate dominance ranking methods

Authored by Han de Vries

Date Published: 2009

DOI: 10.1163/156853909x412241

Sponsors: No sponsors listed

Platforms: NetLogo

Model Documentation: Other Narrative Mathematical description

Model Code URLs: Model code not found

Abstract

Recently, the DomWorld model was used to evaluate five dominance ranking methods. The suitability of the DomWorld model for this purpose is however not without question. The characteristic unidirectionality of most dominance behaviour observed in many monkey species is not found in DomWorld. Besides this, the current paper shows that the additive dominance value updating method in combination with the relative win chance, P(ij) = DOM(i)/(DOM(i) + DOM(j)), gives rise to unrealistically large changes in win chance after fights among low ranking individuals. It is shown that this can be resolved by replacing the additive update rule by a multiplicative one. Moreover, this combination of relative win chance and multiplicative update rule is equivalent to the combination of a sigmoidal win chance and additive update rule as employed in the Elo-rating method. It is also shown that, contrary to Hemelrijk's recommendation, David's score is to be preferred to the average dominance index. The paper concludes with presenting a differentiated list of recommendations on the use of ranking methods that takes into account the required premises and different aims for which these methods have been developed.

Tags

Agent-based model ranking method dominance loser effect winner effect