Convergent, Discriminant, and Construct Validity - Essay Example

Discriminant validity involves conducting a correlational test between the scores of the new measure with the scores of a measure whose construct the test measure does not assess.  The result should show little or no correlation.

Construct validity is the extent to which a measure of a construct assesses what it is intended to assess. Construct validity results when the measure possesses both convergent and discriminant validities.

Test-retest reliability involves the administration of the test measure to numerous respondents comprising a large sample at two different instances.  If the measure is reliable, then answers will be consistent for the two test times.  The set of answers in both times could be tested by statistical correlation, and if the correlation coefficient is 0.80 or above, then the measure is deemed reliable.

Internal consistency reliability is a method of testing whether all the items within a measure consistently measure the same thing.  To determine internal consistency, a series of correlation tests would be conducted among all the items in the measure with every other item.  The average interitem correlation coefficient should be higher than 0.80 for internal consistency to be reliable.

To improve upon reliability and validity, the correlational tests suggested in the foregoing may be conducted.  The test scores should be obtained from a test sample from the intended population, but which will not form part of the actual research sample.