Week 12: Wrapping up and Exam prep

POL272 Quantitative Methods for Social Science Research

Sebastian Koehler

2026-04-16

Recap: Binary Dependent Variables

  • What is a binary dependent variable?

  • How is a linear regression with a binary dependent variable different from a linear regression with a non-binary dependent variable?

  • How do we interpret the coefficients in a linear regression with a binary dependent variable?

Recap: Statistical significance

  • What is statistical significance?

  • What is the difference between statistical and substantive significance?

  • How do we test for statistical significance in the context of linear regression?

Statistical Significance with Binary Outcome Variables

  • So far we have discussed statistical significance of coefficients in the context of linear regression with a non-binary dependent variable. But what if we have a binary dependent variable?

  • In that case, we can still use the same approach to test for statistical significance. We can run a linear regression with a binary dependent variable and then test for the statistical significance of the coefficients using the same approach as before.

Statistical Significance with Binary Outcome Variables

  • We need to be careful when interpreting the results of a linear regression with a binary dependent variable. The coefficients in a linear regression with a binary dependent variable represent the change in the probability of the outcome variable being 1 for a one-unit change in the independent variable.

  • Statistical significance in a linear regression with a binary dependent variable does not necessarily imply substantive significance. We need to interpret the coefficients carefully and consider the context of the study to determine whether the results are substantively significant or not.

An R example

  • Let’s see how that works using an example.

  • Download the file exercise_5.R and the STAR.csv dataset and follow along.

Recap

  1. Introduction and R
  2. Data, Datasets, Computing Means
  3. Estimating Average Treatment Effects with Experiments
  4. Survey Research and Observational Data
  5. OLS Regression with non-binary dependent variable
  6. OLS Regression with binary dependent variable
  7. Controlling for Confounders/ Multiple Linear Regression
  8. Probability
  9. Hypothesis Testing and Statistical Significance
  10. Assumptions of Linear Regression

The Exam

  • The exam will be held online on the 20 May 2026 from 10:00 to 17:00 BST.

  • We will use the Cadmus platform. The link will be available on QMPlus under the assessments tab at the time and date of the exam.

  • Cadmus is a platform for online exams, which allows you to write and submit your exam online. In Cadmus, you will have access to the exam papers and the required data. You will be able to write your answers in a text editor where you can type and copy content (e.g. R code) that you need to submit as part of your exam.

The Exam cont’d

Be aware

Cadmus will automatically submit your exam at 17:00, so you don’t have to worry about submitting it yourself. But that also means whatever is in Cadmus at 17:00 will be submitted. Whether the answer is complete or not.

  • You can work on the exam for as long as you like during the 7 hours, but make sure to manage your time wisely and leave enough time to review your answers before the deadline.

  • Make sure everything is on Cadmus before the deadline.

To avoid trouble

I strongly recommend working in Cadmus when drafting your answers!

The Exam cont’d

What is assessed?

I want to see that you can apply the concepts we have covered in the course to a new dataset and that you can interpret the results of your analysis correctly. So you need to demonstrate that you can:

  • understand the concepts
  • correctly identify which concepts to apply
  • apply the concepts in a new context using R
  • interpret the results correctly
  • write a clear and concise answer to the question

The Exam cont’d

  • The exam will be an open book exam, which means you can use any materials you like (e.g. lecture slides, seminar materials, notes, textbooks, online resources) but you cannot collaborate with anyone else (e.g. classmates, friends, family, …) during the exam.

  • The exam will consist of two questions, which will be similar to the exercises we have done in the seminars but will use different data. You will have to write code, run the analysis, and interpret the results.

  • Both questions carry equal marks.

The Exam cont’d

  • You will be given a dataset and a codebook with the exam questions together with a short explanation of the study which the data are from. You will have to use the dataset to answer the questions. The dataset will be in .csv format and you will have to import it into R.

  • You need to run your analysis in R and RStudio. Then discuss the results in writing to answer the question. You may be asked to create a graph in which case you would need to add the graph as well as the R code.

  • Do not forget to copy and paste the R code to the end of the Cadmus document.

The Exam cont’d

  • All answers need to be supplied in Cadmus. Only information on Cadmus will be marked.

  • In the seminars, we will work with Cadmus to familiarise you with the platform and to give you an opportunity to ask questions about the exam. You will also have the opportunity to ask questions about the exam in the Q&A session of the lecture.

  • If you are unable to attend the lecture, you can still access the Cadmus materials on QMPlus from 4pm today.

Marking criteria

The following criteria will be used to mark your exam:

  1. Correctness of the R code (e.g. does the code run without errors, does it produce the correct output, etc.)

  2. Interpretation of the results (e.g. do you correctly interpret the coefficients, do you correctly interpret the graphs, etc.)

  3. Clarity of the writing (e.g. is your answer well-structured, do you use clear and concise language, etc.)

  4. Understanding of the concepts (e.g. do you demonstrate a good understanding of the concepts covered in the course, do you apply the concepts correctly, etc.)

On the day of the exam

  • Make sure you are familiar with the Cadmus platform and how to use it. We will have a seminar on Cadmus where you can ask questions about the platform and the exam.

  • Make sure you have a good internet connection and a quiet place to work during the exam.

  • Make sure you have all the materials you need for the exam (e.g. lecture slides, seminar materials, notes, textbooks, online resources) ready and easily accessible during the exam.

  • Make sure you have access to R and RStudio on your computer during the exam.

How to prepare for the exam

  • Make sure you understand the concepts covered in the course and how to apply them in R.

  • Go over the lecture slides and review the seminar materials. Make sure you understand how to run the analyses we have covered in R and how to interpret the results.

  • Practice writing R code and interpreting the results. You can use the exercises we have done in the seminars as practice, but also try to find other datasets and apply the concepts we have covered to them.

How to prepare for the exam

  • Work in groups. Ask each other questions, discuss the concepts, and practice writing R code together. This will help you to understand the material better while preparing for the exam.

  • Make sure to ask questions if you are unsure about anything. You can ask questions of general interest on the QMPlus forum. Don’t hesitate to ask if anything is unclear.

  • Try to come up with your own questions and answer them. This will help you to understand the material better and help prepare you optimally for the exam.

Tackling the Exam

  • Read the questions carefully and make sure you understand what is being asked. Make sure to read the codebook and the description of the dataset carefully as well.

  • Plan your time wisely. You have 7 hours to complete the exam (which is plenty), so make sure to allocate your time accordingly. Give yourself time for planning your answers before you start writing.

  • Familiarise yourself with the dataset and the codebook before you start writing your answers. This will help you to understand the data and the questions better and will make it easier for you to write your answers.

Tackling the Exam

  • Run the analysis which you planned and make sure to check your code for errors.

  • Interpret the results of your analysis and write down your interpretation in a clear and concise manner.

  • Answer the questions based on the results and your interpretations. Make sure to answer all parts of the question and to provide a clear and concise answer.

  • Only answer the question. I am not interested in all the cool stuff you can do in R.

Questions?

Time for your questions!