Validity and Reliability
by marsha mullins
1. Does test yield consistent results?
1.1. if results are consistent the test is valid
1.2. If it does not it is not a valid method
2. Test-retest
2.1. A test is given twice to compair the results of the first and the second test
2.1.1. consider the time between tests if too long the lower correlated the two tests will be
2.1.1.1. If too much time has passed the test will produce a reliability coefficient that reflects the students change and not the test
3. Alternative Form
3.1. when two forms of the test which are equivalent to each other can be used to find an average they should be used.
3.1.1. problem with this is that it is hard to get one good test let alone two
4. Internal consistency
4.1. designed to measure a single basic concept
4.1.1. split halves involves splitting the test into two equal halves and determining the correlation between the two
4.1.1.1. an example is even and odd to determine an internal cosistency
5. all test must answer
6. New node
7. content validity evidence
7.1. are the questions being asked asked the correct question for the learning outcomes?
7.2. is the test at the correct reading and achivement level for what is being tested?
8. does test measure what it is suppose to?
9. Construct Validity Evidence
9.1. Measures something not previously measured
9.1.1. New node
10. Criterion-related Validity Evidence
10.1. Scores are correlated with an external criterion
10.1.1. New node
10.1.2. New node
10.2. concurrent
10.2.1. validates evidence with measures that can administetred at the same time
10.3. predictive
10.3.1. tells how well the test predicts some future behavior