The act of measurement involves comparing an object to some
standard of measurement. The standard is sometimes called an anchor
or a gold standard. Without this standard, measurement will be
invalid or meaningless. For survey data, anchors can be external,
as when voter turnout reports are validated with information from
public records, or when health self-assessments are compared to
direct medical tests. Anchoring vignettes provide a comparatively
inexpensive way of creating an anchor within the survey context
itself. The idea is to compare respondents' self-assessments to the
respondents assessments of hypothetical people described in short
vignettes that have known characteristics, and to use the latter to
adjust the former.
By describing levels on a particular concept that are fixed
across respondents, vignettes provide scale anchors that enable
interpersonal comparisons. In combination with the chopit
statistical model or our nonparametric alternative, anchoring
vignettes can be used to adjust for differences in the way
individuals use ordinal response categories.
Anchoring vignettes also may be used as a bridge between
different items on the same domain. For example, having vignette
ratings on two different items relating to mobility allows for
comparisons of the response category cutpoints on the two items and
for both items to contribute to the estimation of respondents'
mobility levels. Related items in different survey instruments may
also be linked through the use of vignettes.
The technique is also very useful for measuring complicated
concepts that are hard to define fully in theory, but can be defined
with reference to examples.
