Note: This is a prototype (storytelling) report for the chemistry course instructor. In this hypothesized use case, this report would be delivered to the instructor after the week 3 of the course.
Weekly Report: Student Learning Profiles in Your Classroom
How have students' quiz score changed from week 1 to week 3?
Chemical equations
Basic statistics of the score distribution (i.e., median, quantiles, min, max) are available when hovering over the boxplots.
Overall, students seem to make fair improvements in their ability to manipulate chemical equations. But does this hold for all student subgroups? Let's check!
Quiz score profiles of different subgroups of students
We discovered 4 clusters of students by analyzing their quiz scores in week 1 and 3 (see Technical Appendix). Here are the quiz score profiles of each cluster:
❗️These plots display normalized (rather than the actual) quiz scores (❓what does this means) Tips for interpreting normalized scores:
- positive or negative: whether the actual score is above or below the mean
- (absolute) magnitude: how much (in standard deviation units) it is above or below the mean
View 1: Score distribution in boxplots for the four student clusters
View 2: Scatter plots for comparing pairs of scores (colored by student clusters)
Comparing week 1 and week 3 scores for each student cluster
Comparing scores of concept and equation quiz for each student cluster
What may these quiz score profiles indicate?
- Cluster 2 is currently doing great (good performance in both the concept and the equation quizzes in week 1 and 3) -- 📧 Congrats for their achievements
- Cluster 1 also achieved good quiz performance in week 3, although they had some problem with manipulating chemistry equations in week 1 -- 📧 Congrats for their achievements
- Cluster 3's scores are around or moderately below the average in the four quizzes -- What advising from instructors could help boost their performance? 📧 Targeted scaffolding message
- Cluster 4 seem to have troubles with both the chemistry concepts and chemical equations since week 1 -- What support could be provided? 📧 Targeted scaffolding message
- ⚠️ These quiz score profiles are of course not static -- 🤔 How to support those who lag behind to catch up, and how to keep the momentum for those who are doing great?
- Pedagogical Implications: What may these suggest about the relative difficulty of the knowledge components in this course? Would students need more help in learning chemistry concepts or manipulating chemical equation?
- ⚠️ These clusters are detected by machine learning algorithms but is up to human for interpretation. Trust your own judgement and feel free to contact us about your feedback.
Technical Appendix
Clustering Procedures
- Exploring and cleaning data: No outliers were identified. However, there were 5 missing values in the score records (Student N276373 has null values in “conceptswk1” and “equationswk1”. Student N947943 has null values in “conceptswk3” and “equationswk3”. Student N505894 has a null value in “equationswk3”). These missing values were replaced by estimates based on related variables (e.g., predicting conceptswk1 based on the values of conceptswk3). After handling missing values, we further normalized the scores to put each score variable on a comparable scale with a mean of zero and a standard distribution of one based on its distribution.
- Clustering: K-means clustering was applied to identify student clusters. The variables included for clustering analysis are conceptswk1, conceptswk3, equationswk1, equationswk3. The number of clusters (k=4) was determined based on the Elbow method as well as visual inspection.