ABSTRACT
Results from a measurement in an intervention study are typically represented as numbers, but different types of measurement have different properties. Nominal, also known as categorical, data are just used to designate qualitative information. Data on an interval scale are commonly found in behavioural studies. Here, the numbers are equally spaced, so the difference between 1 and 2 is the same quantity as the difference between 3 and 4. With interval data, we can compute an average and conduct many types of statistical tests that are not possible with ordinal data. Most widely-used language assessments have been developed using methods from classical test theory. A modification of a popular adage is “If a thing is not worth doing, it's not worth doing well.” This applies to selection of outcome measures: you could have a highly reliable measure, but if it is not measuring the right thing, then there's no point in using it.
