The validity of a test reflects the degree to which it is measuring what it says it is measuring. Validity is often measured by correlation with a similar measure of the same construct. For example, a depression rating scale could be correlated with another questionnaire that measures depression. Differences across groups can also be used to establish validity. Does a group of depressed psychiatric inpatients score higher on the depression scale than a group of healthy subjects? For that matter, do the depressed patients score higher on the depression scale than a group of inpatients with schizophrenia? With convergent validity, measures of similar constructs will rate the same material similarly. Two measures of depression should be positively correlated. With divergent validity, measures of different constructs will rate the same material differently. A measure of depression should not be well correlated with a measure of happiness.