Transportation Statistics and Microsimulation

Inferences    ◾    121

respectively. is adjustment for bias is conservative; that is, the intervals

are too wide. See Kna et al. (1984) for a more accurate and complicated

approach. e impact of adjusting the condence intervals for bias makes

the intervals more realistic. As opposed to traditional condence inter-

vals, the formula also shows that the width of the intervals decreases more

slowly as the sample size increases. Accordingly, the bias-adjusted inter-

vals never decrease to zero. For many transportation applications this is

a more realistic scenario. For example, many measuring devices have a

known bias that can be readily obtained ...