Further investigation of whether correlations amongst data are invalidating the conclusion of a statistically significant trend in Antarctic minke body condition over time
Working Paper
2013
Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
University of Cape Town
Faculty
License
Series
Abstract
A statistically significant decline in body condition in the Antarctic minke whale over the JARPA period was reported in Konishi et al. (2008). Subsequently, however, questions were raised in the IWC Scientific Committee as to whether the model used had adequately accounted for the data structure in JARPA (de la Mare, 2011), i.e. whether neglect of correlations in the data had led to negative bias in the confidence interval estimated for the decline. Earlier a jack-knife approach with year as the sampling unit was used to account for within-year correlations, and showed that while the confidence interval estimates reported originally had been negatively biased, the estimated decline remained statistically significant at the 5% level. This approach did not, however, take account of possible betweenyear correlations. This paper investigates the possibility that such correlations could have biased these interval estimates appreciably. The results show no sign of appreciable interannual auto-correlation, and indicate that the decline reported by Konishi et al. (2008) remains statistically significant at the 5% level. We suggest that his constitutes sufficient statistical evidence to confirm this significance, so that this matter might now be regarded as resolved by the Scientific Committee.
Description
Reference:
Konishi, K., & Butterworth, D. (2013. Further Investigation of Whether Correlations amongst Data are Invalidating the Conclusion of a Statistically Significant Trend in Antarctic Minke Body Condition over Time. In Scientific Committee Annual Meeting 2013 (SC65A).