My students sit at tables instead of individual desks. To reduce the incidence of cheating, I create two versions for each test. I’ve learned in the process that similar looking problems actually can vary quite a bit in their effectiveness.
Here’s my first example:
-4 – 9 =
-9 – 4 =
Similar problems, same answer, yet about 20-30% of my students will answer one correctly, and the other incorrectly.
I’ll leave the which is which and why as exercises to the reader.