How to Incorporate Heterogeneity, Scale, and Effective Sample Size Into the Spatio-Temporal Analysis of Environmental Data Sets: Theoretical and Practical Aspects

• Activity Number: 564
• Type: Topic Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:45 PM
• Sponsor: International Indian Statistical Association
• Author(s): Pierre Dutilleul*+
• Companies: McGill University
• Keywords: Spatio-temporal heterogeneity ; Scale ; Effective sample size ; Autocorrelation ; Heteroscedasticity ; Statistical analysis of environmental data

Abstract

Heterogeneity (of the mean, of the variance, and due to autocorrelation) is a characteristic of most environmental datasets collected in space, time, or both. Heterogeneity generally occurs at several scales, defined relatively to the extent of the sampling domain. The use of an effective sample size is one way to take heteroscedasticity (heterogeneity of variance) and autocorrelation into account in estimation and testing, and requires the preliminary identification of the different components of heterogeneity present in the data at the predominant scales of variation. I will explain how the three concepts of heterogeneity, scale and effective sample size are inter-related, and will discuss the statistical implications of these inter-relationships from the theoretical perspective as well as with examples using simulated and real data.