The National Immunization Survey (NIS) provides state-level estimates of preschool immunization coverage. These coverages are frequently presented in ranked lists, and ranks are frequently over-interpreted. In this paper, we highlight the difficulty in interpreting ranked point estimates. To demonstrate the uncertainty of ranks, parametric bootstrap methods were used to derive 90 per cent confidence intervals for ranks of state vaccination coverage levels among preschool children. We graphically compared states to a reference state. If NIS data are used to rank states, one should consider presenting confidence intervals for rank and the results of comparisons of one state with another graphically.
Published in 2005 by John Wiley & Sons, Ltd.