All Collections
Course Design
Student understanding and misconceptions
Student understanding and misconceptions

Thinking about the mistakes or misconceptions that people often have when learning the material in your course.

Amy Peterson avatar
Written by Amy Peterson
Updated over a week ago

When designing your course, you'll need to think about common mistakes that you think students will make when taking your course. These can be programming mistakes, conceptual misunderstandings, or simply examples of things that are unintuitive. This will help you tackle those problems head-on in your video and coding exercises. Read on for tips and tricks to help write this list of mistake and misconceptions.

What types of mistakes do the students make?

There are several areas in which things can go wrong. The most obvious type of mistake is a coding error (since the student will get the answer wrong when they complete a coding exercise). More serious but harder to detect problems are failures to understand concepts and misunderstandings about the meaning of jargon.

Categorize your responses

In order to think of things that can go wrong, it can be easier to narrow it down and deal with specific topics. What are the potential programming problems? What are the statistical problems? What domain-specific knowledge is hard to learn? Which bits or jargon will be unclear?

I can't think of anything

  • To inspire, take a look at The R Inferno, a classic (by programming standards) text on things that can go wrong when learning R.

  • Try explaining a concept or definition from the course out loud to someone with no data science background. Note the point at which they begin to look confused, or their eyes glaze over.


From a course on generalized additive models. The difference between prediction intervals and confidence intervals.

From a course on business process mining. The structure of "process data" might be confusing to students new to the field, especially the distinction between events and activities.

From a course on tree-based models. Not knowing which are the best hyperparameters to optimize for the different tree-based models.

Did this answer your question?