# Matching Tests

An interesting coincidence happened today in class. My co-teacher was handing the students back their tests. She wanted each student to mark another’s test. However, she accidentally handed a student back her own test. She casually said to me “weird, what are the chances of that?” Her question implied that this was a rare event. The answer is quite the opposite.

My co-teacher had a 63%, or $\frac{e-1}{e}$ to be precise, chance of handing a student back his or her own test. If you are wondering, e stands for a special number:

e = 2.71828…

This percentage is not good. It means that, if you randomly distribute tests to students to mark, about two thirds of the time at least 1 student will end up with his or her own test!

How in the world did I come up with that answer? Let me back up a bit. Suppose we had 3 students and we wanted to give them back their tests to mark so that no student had his or her own test. What options do we have? One option is:

Student A: test B

Student B: test C

Student C: test A

Another option is:

Student A: test C

Student B: test A

Student C: test B

Hence, we have 2 options where we will hand back the tests with no matches. How many options do we have in total?

3! = 3*2*1 = 6

Thus, there is a 2/6 = 33.3% chance that everyone has a different test. In other words, there is a 66.6% chance that at least one student receives his or her own test to mark (which would not be good for academic standards).

What if we had 4 students? You can do the math, and you will find that there are 9 options where the students have different tests. For example:

Student A: test B

Student B: test C

Student C: test D

Student D: test A

Again, basic counting principles dictate that there will be 4! = 24 different possible combinations. Hence, there is a 9/24 = 38% chance each student will have a proper test to mark. We could continue on like this; painstakingly grinding out the probabilities for each situation until we reached a class size of 25. However, there is a pattern we can exploit. Consider the following table:

 Number of Students Fraction of proper distributions 3 2/6 = 33% 4 9/24 = 38% 5 44/120 = 37% 6 265/720 = 37% 7 1854/5040 = 37% 8 14833/40320 = 37%

Do you see the pattern? Once the class size gets to 5, the percentage becomes stable. Based on the above data, we could predict that a class of 25 students will have roughly the same percentage, 37%. Indeed, one can prove mathematically that this percentage will be true for any size class, large or small.1 Thus, my advice to teachers is as follows. Be careful distributing tests to students to mark. You are playing a game with a 37% chance of winning. And any gambler knows, those are not good odds.

1See next blog post.