Validity and Reliability

1. Reliability is when test scores show stability and consistency when given repeatedly.

1.1. Test-Retest Method is when a test is given twice within a short time period from the first to second test. Both scores are then looked at for correlation. (The longer space of time between the two tests the lower the reliablility coefficient). The problem with this method is upon retest memory and experience may be involved and could make scores less reliable.

1.2. Alternative form are tests given in two different forms and used to determine an estimate of reliabiltiy based on the scores.Both forms of a test are given to a group. This method eliminates the problem of memory and practice. If there is a large difference between the two scores it usually indicates unreliablity. This method is most often used by test publishers.

1.3. Internal Consistency is used when a test is measuring a single trait.

1.3.1. Split Halves is done when a test is split into two halves that are equivelant. It then correlates the two halves of the test.

1.3.1.1. Odd Even is used to divide up the easy and hard items on a test because most tests begin with easy items and end with hard which makes a split half test more difficult to create.

1.4. Kuder-Richardson Method determines the extent that the entire test represents a single concept.

2. Why Validity and Reliability are Important?

2.1. Validity and Reliability are important because without them you cannot get an accurate analysis of a students learning. They ensure that educational goals are met and that learning is enhanced.

3. Validity is ensuring that a test measures what it is supposed to be measuring or in other words that the test matches the instructional objectives.

3.1. Criterion-Related Validity takes scores from a test and correlates them with an external criterion.

3.1.1. Concurrent validity evidence correlates test scores and another criterion at the same time.

3.1.2. Predictive validity evidence correlates test scores with a criterion measure after a some time has passed.

3.2. Content Validity inspects tests questions to see if they correspond with what should be covered on the test. It is easiest to use in achievement tests.

3.3. Construct Validity is used when a test is measuring a criteria that has not previously been measured or no criteria exists. The relationship of the test information usually corresponds with some theory.