Laserfiche WebLink
<br />Table 6. Simulation results demonstrating small sample bias and lack of precision. The <br />simulations were accomplished using CAPTURE. The data were generated with model M(t)!, <br />where t = 3 and the time varying capture probabilities (P) specified by a beta distribution (1.5, <br />2.5, lower bound, upper bound capture probabilities). Upper and lower bound capture <br />probabilities were 0.01 to 0.04, 0.02 to 0.08, and 0.05 to 0.15. 100 replications for each trial. <br /> <br />AveP <br /> <br />N <br /> <br />Ave Nhat <br /> <br />Rei <br />bias <br /> <br />Ave SE <br /> <br />CV <br /> <br />0.021 200 61 -0.70 34 0.56 <br /> 500 252 -0.43 278 1.10 <br /> 1000 573 -0.50 118 0.21 <br />0.043 200 137 -0.32 66 0.48 <br /> 500 446 -0.11 162 0.36 <br /> 1000 886 -0.11 276 0.31 <br />0.088 200 215 0.08 108 0.50 <br /> 500 518 0.04 142 0.27 <br /> 1000 1034 0.03 208 0.20 <br />1 M(t) was chosen over other models in Table 5 with higher selection criteria because our <br />experience has shown capture efficiency varies greatly from pass to pass due to changing flow <br />and turbidity. <br /> <br />25 <br />