Abstract: This paper put forward a general approach to measure test validity. First, according to the measure aim, the test constructor raised a certain set including various possible fitting models which is called“Criterion Structure Space”.Then each model in the criterion structure space was examined respectively with confirmatory Factor Analysis,If an assumed model did fit the observed data best,the proportion of the variance which is caused by the factor of the model to the total variance of the observed variable could be defined as validity of the test.

Key words: validity, criterion structure space, confirmatory factor analysis