Planar localisation analyses: A novel application of a centre of mass approach

Sound localisation is one of the key roles for listening, and measuring localisation performance is a mainstay of the hearing research laboratory. Such measurements may consider both accuracy and, for incorrect trials, the size of the error. In terms of error analysis, localisation studies have frequently used general purpose univariate techniques in conjunction with either mean signed or unsigned error measurements. This approach can make inappropriate distributional assumptions and so more suitable alternatives based on directional statistics have also been used. Here we investigate the use of a variety of methods, assess their performance, and comment on their use and availability. We also describe a novel use of a 'centre of mass' approach for describing localisation data jointly in terms of accuracy and size of error. This spatial method offers powerful, yet flexible, statistical analysis using standard multivariate analysis of variance (MANOVA).