A comparison of the discrete RAMPs (the actual proportion of held crab that died in captivity for each Score) between the Discard- and Unobserved-mortality studies (Fig. 3) indicate probabilities of mortality are significantly higher for the Unobserved-mortality study for Scores three and six (one-sided p-values 0.004 and 0.0002, respectively, Fisher's exact test). Moreover, there was convincing evidence of a higher proportion of mortalities for the Unobserved-mortality study than would have been expected if the odds of mortality were equal for both studies, after controlling for Score (one-sided p-value 0.0002, Mantel–Haenszel test). Comparable to the logistic RAMP, a comparison of discrete RAMPs revealed similarity between studies at high and low Scores, but divergence at intermediate Scores.
3.4. Estimation of mortality rates
To evaluate how different RAMPs affect the estimation of bycatch mortality rates, we calculated rates for our research trip in the Gulf of Alaska using both the logistic and discrete RAMPs from both the Discard- and Unobserved-mortality studies. For both studies the logistic and discrete RAMPs estimated the same mortality rates. For the research trip, the discard mortality rate was estimated to be 24% from the Discard-mortality study RAMPS, and 31% from the Unobserved-mortality study RAMP