AmeriDendro 2013 registration is open: follow the link on the Overview page.
Mean sensitivity isn't! A description of mean sensitivity as a useful statistic in dendrochronology
Presented by Dr. Andrew BUNN on 15 May 2013 from 18:00 to 21:00
Type: Poster Presentation
Session: Poster Session + Reception
Board #: 39
Abstract There is a small but substantial literature showing mean sensitivity to be a poor statistic for describing the properties of tree-ring data. Yet, it continues to be widely used. We show in a logical and straightforward way that mean sensitivity is an inefficient estimator of the coefficient of variation and that mean sensitivity can be expressed as a function of the variance and first-order autocorrelation of a time series. Mean sensitivity is proportional to variance in cases without high autocorrelation but is harder to interpret than either variance or autocorrelation. Thus, it is hard to justify the use of mean sensitivity in many, or any, circumstances. We suggest that is is more worthwhile to describe tree-ring data in terms of a few simple time series properties familiar to most dendrochronologists and give examples of how to do so.
Room: Salon E
- Prof. Esther JANSMA Utrecht University
- Dr. Mikko KORPELA Aalto University and University of Helsinki
- Dr. Robert WESTFALL USDA Forest Service, Pacific Southwest Research Station, Albany, CA, USA
- Dr. James BALDWIN USDA Forest Service, Pacific Southwest Research Station, Albany, CA, USA