The problem is quite simple. You should not mistake success rates for the marks you give to the students. A good mark is often defined by tradition to a precise number, for instance, in France, we consider that 10/20 is sufficient. In China, it’s 60/100. Each country has its own tradition. This is not the point. But you must consider the nature of the test. It’s quite obvious that getting a 50% success rate on a true or false test is an absolute disgrace, as it’s the score that you’d get through random answers.
If you pass a MCQ test with 4 choices for each question, a random answer would provide you with a 25% success rate. So, we must precise our expectations in order to avoid any misinterpretation. The easiest way is obviously to set the minimum level at 75% or more, depending on the subject and the difficulty of the questions.
There is another solution, commonly used is many Business Schools around the globe, that allows you to translate the success rate into a more traditional mark.
We can consider that a random answer is utterly worthless. Therefore, we’ll try to use probabilities to give a 0 (zero) instead of 25% to the students who only guess the answers instead of learning and thinking.
It’s rather simple.
You just have to give :
3 points for a good answer
– 1 for a mistake
0 when the student doesn’t answer at all (and admit that he’s ignorant).
A random answer will get the following results.
Over 4 questions, in 3 cases, the answers will be wrong. That’s – 3
In 1 case, the answer will be right. That’s + 3
All in all : 0
You can further discourage guessing by choosing a harsher scoring.
2 points for a good answer
-1 for a bad answer
0 if the student doesn’t answer at all.
It means that the student has some interest to guess only if he can eliminate 2 false solutions before he guesses between the last 2.
It’s a good system, providing that you explain it to the students. It makes MCQs much more serious.