A teacher recently shared how she uses formative assessment in her classroom. She said she give students a yes/no question to answer. If half the students answer correctly, she goes on to the next lesson.

The teacher’s desire to keep tabs on how well her students are learning is laudable. The method she’s using to do that, however, may not be producing the results she wants.

When there are only two choices, statistically speaking we can expect roughly half the respondents to pick the right answer just by guessing without any knowledge of the topic. We could demonstrate that by asking people at random to answer a question like this:

We’d expect a rough 50-50 split if people simply guess.

Let’s say our teacher finds 50% of her students know the right answer and 50% don’t. She’s fairly safe in assuming half the class does not know the answer. She cannot, however, assume that the other half does know the answer.

Statistically, half the students could pick the right answer to a yes/no question by guessing. It is entirely possible that neither the 50% who answered incorrectly nor the 50% who answered correctly actually knew the answer.

Now suppose, for the sake of argument, that the teacher’s survey results accurately reflected students’ learning: Half the class really did learn the concept she taught.

If a concept is important enough to merit an entire class period, is a 50-percent success rate on short-term learning good enough?

How do you see it?