A Bayesian Adaptation of Publication Bias Compensation for Stranded Marine Mammal Growth Estimation

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Mary Shotwell*+ and Elizabeth H. Slate
• Companies: Medical University of South Carolina and Medical University of South Carolina
• Keywords: pseudo-data ; strandings ; publication bias ; detection bias

Abstract

Federal legislation restricting research on wild marine mammals places emphasis on the importance of fully utilizing stranded animals. Bias in detection of stranded marine mammals poses a statistical concern, analogous to publication bias. A Bayesian adaptation of a pseudo-data method for publication bias is described and applied to South Carolina strandings data in order to better estimate total body weight growth of bottlenose dolphins (Tursiops truncatus). A simulation study was performed to assess sensitivity of the method to sample size, detection scenarios, and guesses at the true parameters. Results of the simulation show that the new method outperforms complete case analysis as long as bias is present. The real data analysis results suggest that South Carolina bottlenose dolphins reach smaller sizes at a faster growth rate than previously estimated from complete case analysis.