Compiling and distributing estimates of assessment-probability for all US stocks

Dr. Olaf Jensen, Department of Marine and Coastal Sciences, Rutgers University  


Stock assessment models fitted to data for managed fish species often inform our understanding of marine fish productivity and stock status. Given that local scientific and management authorities give certain species priority based on a number of scientific, political, or economic factors, the group of stocks that is assessed may not be representative of the productivity or status of marine fish populations as a whole. Calculating a “propensity score” (the probability that any marine fish would have been assessed in some other realization of the process governing selection of assessed stocks) represents one solution to the problem of a non-random selection of stocks receiving assessments. Propensity scores can be used to match stocks that have been assessed with those that have not to detect systematic differences. If a bias exists, the propensity score can be utilized as a covariate in any subsequent analyses, thus mitigating bias. Although the propensity scores for all assessed and unassessed US stocks have been estimated in previous research, these propensity scores will be most useful if made publicly and freely available online. We propose to compile propensity scores for all stocks landed in the US and create an RShiny graphical user interface (GUI). The publicly available RSHINY GUI may be used to search for individual stock propensity scores, or to download the entire database. This interface will also include the predicted probability of assessment and either the forecasted status based on the last assessment, or assessment status based on a stock with identical characteristics.