Statistical analysis helps you create a good (digital) test. One of the statistical concepts you can use is the a-value. But what exactly is the a-value of a question? And what use is a-value in a (digital) test?

The a-value is the percentage (or proportion) of candidates who choose a distractor for the question. A multiple-choice question has several answer options. In addition to the correct answer, there are a number of “*distractors.* As a tool, the a-value allows you to measure the quality of distractors. Thus, each distractor in the test has its own a-value. This a-value represents what proportion of candidates choose this distractor.

*There is a clever mnemonic to remember what the a value stands for. In fact, you can think of the A-value as the “ Distractor Value. *

The purpose of a (digital) summative test is to make a final assessment. As an examination body or training institute, you obviously want to measure knowledge and insights in the best possible way. For example, you want the results of a test to accurately reflect candidates who prepared well or less well. It is therefore wise to look at and analyze statistical data at the item level. A useful statistical variable is the a-value, which gives you insight at the item level and allows you to improve the quality of items.

Suppose you have an exam with 20 multiple-choice questions. If so, not every question may be equally strong test-wise. By providing insight into the a-value, you know what percentage of candidates choose a distractor. A sample question is as follows:

Question: what are the colors of Optimum Assessment’s logo?

- Blue and green
- Yellow and black
- Red and black
- Red and Yellow

Answer 3 is correct and constitutes the “key answer. The remaining answers (1, 2 and 4) are distractors. Suppose the statistical analyses show that answer option A is hardly chosen by candidates. In that case, the answer option does not find a good fit with the question.

As a result, the probability score increases. The probability score is the score a candidate can achieve, if he guesses/games the right answer, but does not actually know. An item with a low a-value contributes to your test not reflecting the knowledge level of your candidates as well. You can choose to change the distractor in that case.

The a value is between 0 and 1. Suppose you have one correct answer option that is chosen in half the cases by the exam candidates, then there is still a value of 0.5 distributed among the distractors. So you always calculate an a-value over answer options.

Want to learn more about a-value or other statistics related to digital testing? Take a look at our knowledge base. You can also subscribe to our newsletter or follow us on LinkedIn to always stay up to date on developments in digital assessment.