Correction for guessing

Description

This article explains the guessing probability of a question. In closed questions, candidates have the chance to guess the correct answer. The correct answer is always in the answer options, whether in an answer alternative (multiple-choice question), in a combination (sorting question) or somewhere on the image (hotspot question).

The guess probability indicates the probability of the candidate answering a question correctly by answering randomly. The guess score indicates the expected number of points, based on the total number of points to be gained from the question looking at all possible answers, not just the correct.

For all questions, the guessing score consists of the number of bonus points * the guessing chance, plus possibly the expected score per possible answer (right or wrong) * the guessing chance. This expected score per possible answer is truncated to 0 within the interaction, so that a council score does not become negative. (After all, the actual score obtained on a question is also truncated to 0.)

Application per question type

Multiple-choice questions

One correct answer

For a single correct answer, the probability depends only on the number of answer options. With four answers and one correct option, the guessing probability is 25%.

Multiple answers

If there are several correct answers, the guessing chance and score are calculated by including all combinations to be filled in. The guessing probability for this question type is therefore a lot lower than when only one answer is entered.

Matching/Matrix questions

One association

This is effectively a multiple-choice question with one answer, but with distractors distributed across two dimensions: rows and columns. Guess probability and guess score are therefore calculated in the same way.

One or more associations per row

This is effectively multiple multiple-choice questions (possibly with one answer), in one. The guess probability and guess score per row are therefore calculated in the same way. The guess probability across all rows is multiplied; the guess score added up.

Sort questions

Sorting questions calculate the number of different sequences a candidate can fill in. An example:

Hot spot questions and Drag & Drop questions

First, the area per circle or hotspot area is calculated (approximately), both with and without overlap with one or more higher areas. After this, which answers are possible (no one chosen area, or just two out of three), and what is the probability of giving exactly this answer, are considered. An area can only be chosen once. There is therefore no need to take into account that the candidate indicates the same location twice. An example:

Graphic associate question

It first calculates how many lines can be drawn within the specified areas. An example:

This is otherwise effectively a multiple-choice question with multiple answers (possibly limited), where the possible lines are the answers. The guess probability and guess score are otherwise calculated in the same way.

Open questions/fill-in questions/upload questions

The guess probability and guess score for open questions, fill-in questions and upload questions are always 0.

Combined questions

For combined questions, which consist of multiple interactions, the guess probability and guess score are collected for each underlying question. The guess probabilities are multiplied, and the guess probabilities summed.

Although for questions with a single interaction, the expected score per possible answer within the interaction itself is truncated to 0, as also mentioned in the introduction, for compound questions this is not done at interaction level, but at question level.

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