There are various aspects to validity, including:

• 'face validity' (does the test look as if it measures what it is meant to?);
• 'content validity' (does it contain a representative sample of the skills and knowledge it is meant to be assessing?);
• 'criterion-related validity' (to what extent do test results agree with those derived from an independent and dependable measure of a student's ability?);
• 'construct validity' (can it be shown that the test measures the ability it is supposed to measure?).

The latter two aspects are research procedures which are most relevant to the production of mass standardized tests (for further details, see Hughes, 1989: 23-27). For university language teaching purposes, the first two are more important.

Face validity is important because tests that are not thought to be relevant to the supposed purpose of assessment may not be accepted by students and this may reflect how they perform on them. For example, cloze and C-tests (see Module 14, section 14.5.1) may not be considered meaningful in assessing the ability to write the foreign language (FL), or a writing test may not be thought relevant to a reading knowledge course. Content validity is even more crucial: if a test ignores whole areas of a syllabus this will have a very negative effect on teaching and learning (see 'Washback', 13.1.4.5). As already noted (see 13.1.3), our assessment needs to be closely tied to learning outcomes. It is essentially a two-stage process in which, first, we identify the outcomes (knowledge or skills) we want to assess, and then decide how it is we are going to test them. Starting from this premise, we are more likely to measure what we want, that is to make our assessment valid.