STA210 Regression Analysis

Course Objectives

• Analyze real-world data to answer questions about multivariable relationships.
• Fit and evaluate linear and logistic regression models.
• Assess whether a proposed model is appropriate and describe its limitations.
• Use Quarto to write reproducible reports and GitHub for version control and collaboration.
• Communicate results from statistical analyses to a general audience via a scienctific report and presentation.

Class Introduction

It is a 6-week course required for students minor in statistics. It is a small class with 8 students and most students come from other disciplines, such as political science(1), economics(1), computer science(4), biology(1), and mathematics(1).

Pedagogy

According to students’ aptitude, I design the course emphasizing more on

• comprehension of statistical methods,
• ability to apply statistical methods and models using R, and
• capability to interpret and communicate with others.

Enhance comprehension of knowledge

I emphasize on forming curiosity and cognitive learning. I would use the heredity of height to motivate the idea of regression model. I would explain the intuition behind the ordinary least square estimator in regression analysis. I will explain multi-collinearity from geometric aspect.

To enhance students’ memory of knowledge, I would introduce new information by analogy and comparison. For example, I would analogize the interpretation of coefficients in a multiple linear regression to a single linear regression, analogize the logistic regression to linear regression. Through learning cognitively, students can memorize the previous knowledge and learn new knowledge quickly and efficiently by connecting it to previous existing methods. Before each quiz, I will lead a review session to process and synthesize knowledge. For example, I would introduce geometric interpretation of coefficients in a regression in different situations, for continuous and nominal variables, for a single variable and multiple variables, for a linear regression and a logistics model. During the review session, question-driven learning is very helpful when combined with collaborative learning. I would use questions to drive students participate and review the previous materials. Almost everyone participated actively even for students previously sitting behind and prefer not engaging.

Another important method I found to enhance students’ comprehension is application. I will also encourage students to apply what they have learned to solve real-world problems, such as analogize hypothesis testing to a court case, evaluate covid-19 rapid test kit by interpretating its specificity and sensitivity. Through practicing in life situations, students are able to understand statistical results when reading scientific report and think in a statistical way.

Practice using R and communicate results to others

During the labs, students are exposed to various real datasets and are asked to using R programming to conduct regression analysis to solve the real-world problems. I found cooperative learning will be most effective since half of the class major in computer science. According to students’ aptitude, I allocate the team based on their background to encourage students with little programming background learn from those students skilled at programming. In the final project, students are asked to analyze a data-driven research question and present the findings in a report and a video. Through learning collaboratively, students are able to get trained in how to propose a statistical question and write a formal scientific report to communicate with others. The final project also helps students to synthetic what they have learned in the class and put into practice, and question themselves on the model evaluation. This transformative process can not only encourage mastery and transmit information, but also develop critical thinking skills and self-motivation and self-governing.

Course Evaluation

Here are the feedbacks from students and observers.

Credit

Most of the content is based on STA 210 - Spring 2022 by Dr. Mine Çetinkaya-Rundel and STA 210 - Fall 2021 by Dr. Maria Tackett. I modify course materials based on students’ background and add new lecture on data wrangling.