divides construct validity into two further sub-headings, namely construct under-representation and construct-irrelevant variance. The former suggests that a test is too narrow and omits essential target language and construct. The latter, conversely, describes the test as being too broad, with too many items that are not relevant to the construct.