Reliability and validity

It is desirable that tests be valid and reliable. If a test falls short in one or both components, then it is not suitable for use.

Reliability

One way to measure whether your test is reliable is to repeat the test and see if it produces the same results. It is not often that the results are exactly the same, which is why reliability is often expressed through statistical values. Cronbach's alpha is an example of this. See also the article Statistical Values for this.

Open-ended questions are less easy to test for reliability. One option is for multiple people to assess the answers. Based on the scoring key accompanying the test, the assessors must independently give the same score to the candidate's performance. The way this is done may vary from test to test. In RemindoTest, correction rounds can be set up that allow multiple correctors to look at a test. Any differences in grading can also be resolved in RemindoTest by averaging, for example.

Validity

The validity of a test has to do with what a test is supposed to measure. After taking a test, it should be clear that . These two components are referred to as content validity and concept validity. For the use of RemindoTest, these are the most relevant forms of validity.

Content validity means that the test must adequately cover the subject matter covered in . Here the link between the learning objective of the teaching and the test is essential. What knowledge should a student have acquired in the previous education? What should he be able to do with that knowledge? This must be properly mapped out before the test is drawn up. This can be done by using test templates, which can be managed in RemindoTest.

Comprehension validity means that the test tests what it is supposed to test. This seems obvious, but by no means all tests are concept valid. An example: a test on the Cold War contains an assignment to an English-language source. If students have sufficient command of the English language to fully comprehend the source, there is no problem. If that is not the case, you are not testing students' historical knowledge, but rather their proficiency in English. The question is then not conceptually valid.

Disclaimer: This text was automatically translated from the Dutch version.