Reliability of a test can be defined as the measure of how consistent your test is in measuring the technical skills of a candidate accurately. If a test is reliable, it has good questions which makes it a good assessment to test the skills of candidates.
To calculate the reliability of a test, the following conditions should be fulfilled:
- A minimum of 50 candidates should have taken the test
- A minimum of 10 questions should be in the test
What makes a question good?
A question in a test becomes a good question when it accurately assesses the technical skills of a candidate. For example, a good question contains proper language and implementation of the concept, the question is based on.
What makes a question bad?
If a question has a low-reliability score, it is a bad question. The reliability of the question can be low due to the following reasons:
- The content of the question is not proper or clear. This includes the following:
- The language of the question might be confusing.
- The test cases of the question might not be clear.
- The code stubs might be wrong.
- There might be some missing libraries or dependencies in the question.
- The test is too difficult for the candidates.
- The questions of the test and the skills of the candidates do not match which means that the candidates are not having the skills to solve the questions asked in the test.
- The test is taken by too few candidates.
There might be other reasons too for a low-reliability score of the question. It is recommended to delete or hide such questions from the test.