STAT 1: Intro to Statistics

accuracy, relevance, bias-resistance

After taking this Escala course, I changed my grading to be 100% summative assessments: quizzes, midterms, and a final. This change made the grades more reflective of what the students actually know rather than trying to manage and control their behavior. I believe that this is more bias resistant.

Here is the explanation from my syllabus:

Grading: We'll use mastery based grading in this class, which may be different than what you're familiar with. I'm using this system because it gives students many opportunities to learn without penalizing mistakes, and I've found it works well for this class.

We'll use the following grade weighting scheme to determine the final grade.

70% standard mastery

30% final exam

We'll assess mastery of each standard by solving a problem in an individual, in class assessment. You demonstrate mastery by earning one of the following two scores.

5/5 Correct

4/5 Correct, with only minor mistakes

If you don't earn a 4 or a 5 on your graded work, then you need to review the material and try the problem again. If you don't demonstrate mastery by the final, then the missing grade will then be recorded as a 0.

Only your highest score counts for each standard. For most standards you'll have three chances to demonstrate mastery: quiz, midterm, and final. Each standard will have a corresponding online homework to prepare you for the assessment. We'll cover approximately 2 standards per week for a total of around 24 standards.

The final exam will have a subset of the standards. I will include problems where the most students would benefit from having another chance. If we included all the problems / standards, then the exam would be too long.

Clarifying edit- You need do every question on the cumulative final exam. It doesn't matter what score you've gotten on a standard previously. It is a separate grade category from the standards.

I feel much more confident about the accuracy of the grades when using this system. There is a clear, objective, and rational standard that students have to meet to get "checked off" for each concept that we are learning. This clarity and focus has also helped me write better lectures.

Another change I made because of the Escala course is to make the questions more culturally relevant. Compare the two images below representing typical questions for constructing a confidence interval for a proportion. The textbook question is dry, generic, boring, and irrelevant. Instead, I asked the students, "Who feels that social media has had an overall positive effect on their life?" I counted up the number of students who agreed with this statement, and we made a confidence interval for the proportion of college students at CSUS who agree. We discussed the important caveat that our class is not a simple random sample.

By collecting our own data and looking at it, we see statistics in action. It becomes more real and relevant. The impacts of social media are very interesting and relevant. By asking questions like the above, I'm giving the students food for thought and drawing awareness to personal choices they make every day.

Image one: screenshot from lecture where I asked an interesting question and collected data from students.

screenshot from lecture where I asked an interesting question and collected data from students.

Image two: typical (boring) stat 1 question

Bio

I’m an Assistant Professor in the Department of Mathematics & Statistics at CSU Sacramento, where I teach classes in statistics, computing, and data science. I graduated from UC Davis in 2019, where I studied with Professor Duncan Temple Lang and spent lots of time in the Datalab. My research interests include software development, data science, and scaling statistical applications to large data sets.