On July 18, 1994, U.S. News and World Report published its annual rank
ing of America's best hospitals. The rankings were based on a model de
veloped by the National Opinion Research Center (NORC). National ranki
ngs are a type of program evaluation that deserve careful scrutiny wit
h regard to their methods. While problems with measurement validity ar
e discussed the principle concern here was the model used to compile a
final ranking of hospitals. The News-NORC model is typical of ranking
schemes that usually reduce higher levels of measurement to lower one
s in order to compute final scores. Such simple tabulation techniques,
however; distort the ratios between the scores and bias the results.
A type of percentaging, using part/whole percentages (P/W%) war used t
o show the differences between hospital rankings when data are kept at
higher levels of measurement, such as interval or ratio levels. Perce
ntaging methods produce more valid results when comparatively evaluati
ng programs whether the goal is to produce national rankings or some o
ther outcome.