1: As a field, we have traditionally
referred to scales as “reliable” or “unidimensional”,
but methodologists since Lord & Novick (1968) caution that instruments do
not have reliability, only scores from particular
samples do. (See also Wilkinson and the Task Force
on Statistical Inference, 1999.) Despite this, we should have a reasonable
expectation for instruments to have the same basic structure across
samples if we are to have any rational basis for the science of measurement
within the social sciences.
[return]
2: We could go on for many more pages
summarizing various historical approaches to summarizing congruence.
For the sake of parsimony we will simply refer the readers to the
above-cited resources that give thorough coverage of the issues.
[return]