Adaptive experiments with a multivariate Elo-type algorithm

Philipp Doebler*, Mohsen Alavash, Carsten Giessing

*Corresponding author for this work
6 Citations (Scopus)


The present article introduces the multivariate Elo-type algorithm (META), which is inspired by the Elo rating system, a tool for the measurement of the performance of chess players. The META is intended for adaptive experiments with correlated traits. The relationship of the META to other existing procedures is explained, and useful variants and modifications are discussed. The META was investigated within three simulation studies. The gain in efficiency of the univariate Elo-type algorithm was compared to standard univariate procedures; the impact of using correlational information in the META was quantified; and the adaptability to learning and fatigue was investigated. Our results show that the META is a powerful tool to efficiently control task performance in a short time period and to assess correlated traits. The R code of the simulations, the implementation of the META in MATLAB, and an example of how to use the META in the context of neuroscience are provided in supplemental materials.

Original languageEnglish
JournalBehavior Research Methods
Issue number2
Pages (from-to)384-394
Number of pages11
Publication statusPublished - 01.06.2015

Research Areas and Centers

  • Academic Focus: Center for Brain, Behavior and Metabolism (CBBM)


Dive into the research topics of 'Adaptive experiments with a multivariate Elo-type algorithm'. Together they form a unique fingerprint.

Cite this