18
improvement either with regards to dimensions not included in the index or relative more
demanding standards of the component items of the EDI.
The EDI can be used for comparative purposes, whether to compare a country
with itself or to other countries. Of these two uses, a comparison of a country with itself
at a different point in time is usually easier to interpret. After all, a country may well have
made notable improvements yet fall behind other countries if these other countries have
made even greater strides. However, in either case, it is crucial to stress that any such
comparisons should always be based on substantial and never on small differences. The
reason for this is that the EDI, as any index, has a certain degree of measurement error
and within the error bounds, it is inadvisable to make strong statements about differences.
Indeed, as was estimated through the sensitivity analysis, generous error bars for EDI
values between about .25 and .75 are roughly ±.07. Thus, any pair of cases that differ by
less than this value—for example a country with a EDI of .85 and one of .95—are simply
too close to validly distinguish. Hence, it is methodologically unjustifiable to offer an
overly precise ranking of countries, as is commonly done in the context of other indices,
which simply transforms the scores of the EDI into a ranking without taking into
consideration the degree of uncertainty associated with the EDI scores.
The identification of benchmark cases that are prototypical representations of the
features that are associated with a range of scores can help to give concreteness to the
meaning of each number. Moreover, the EDI can be used as a flag, in the sense that the
specific scores of each country invite the reader to go back to the tables on the component
items to identify precisely what feature or features account for a country’s score. In this
way the EDI can be used as a valuable analytical tool, in that it offers a summary score
Convention