Insegnamento a.a. 2024-2025

Classes: 45 (II sem.)

Instructors:
Class 45: OMIROS PAPASPILIOPOULOS

Preliminaries to the course are basic probability (at the level of Chapter 6 of the above book) and very basic calculus and linear algebra. Although this course is in R, and no previous knowledge of this is required, the course assumes that the students are familiar with the fundamentals of object-oriented programming (e.g., Python) as developed in the corresponding course in the first semester. Additionally, this course is in conversation with the concurrent course in Machine Learning and each course will benefit from concepts developed earlier in the other.

The course develops hands on quantitative social science skills together with the maturity to formulate interesting questions and the confidence to use data and algorithms to produce useful insights and policy recommendations. During the course the students will learn to use the R language and they will work with tidyverse in R, which is a standard framework for data analysis among computational social scientists. Note that most of the fundamental code used in the course will be provided to the students. This course is in conversation with the concurrent course in Machine Learning and each course will benefit from concepts developed earlier in the other.

The course is organized along the following themes:

1. Introduction
   - presentation of the goals of the course, case studies, and motivating problems

2. Measurement
   - surveys, rectangular data structures, types of data, summarizing and visualizing
   - measures of association 
  
3. Introduction to R
   - Rstudio, tidyverse, markdown basics, jupyter notebooks
   - plots and ggplot
   - data summarization and visualization

4. Causal inference 
   - introduction to causal inference and randomized control trials
   - potential outcomes and introduction to randomization inference
   - case studies

5. Discovery
   - Finding hidden patterns in data: studying happiness, polarization, choices
   - From rectangular data to singular vectors: principal component analysis
   - Case studies
   - PCA in action, biplots, correspondence analysis
   - Exploration of spatial data and maps

6. Uncertainty
   - Testing for treatment effects pt 1 (randomization and p-values)
   - Testing for treatment effects pt 2 (bootstrap and p-values)
   - Case studies
   - Assigning confidence to data summaries pt 1: examples and concepts
   - Assigning confidence to data summaries pt 2: maths and models
   - Case studies
   - Multiple testing

7. Structural modelling
   - Linear models: interpretation, inference for coefficients, causal inference and heterogenous treatment effects 
   - Linear models: sensitivity and residual analysis
   - Case studies

At the end of the course student will be able to...

+ interprete and answer causal inference questions

+ develop R code for data analysis

+ formulate and evaluate structural models

At the end of the course student will be able to...

+ carry out data analysis

+ create insightful graphs and visualizations

+ pose and answer causal inference questions

+ use state of the art analysis methods for quantitative social science

• Lectures
• Practical Exercises
• Collaborative Works / Assignments

+ practical exercises are in terms of case studies of analyzing real data and social science questions

+ collaborative work is in terms of a group project that will include the collection and analysis of data for a real social science question 

	Continuous assessment	Partial exams	General exam
Written individual exam (traditional/online)		x	x
Collaborative Works / Assignment (report, exercise, presentation, project work etc.)	x

The course does not distinguish between attending and non-attending students. 

 

Evaluation:

- midterm exam, final exam, project (it is optional and provides additional points on top of those from exams)
- exam format: part multiple choice that examines fundamental knowledge, part exercises that require computing; open book exam

The course is built around the following textbook:

https://press.princeton.edu/books/hardcover/9780691222271/quantitative-social-science

but also draws from a few other places and from the research of Professor Papaspiliopoulos.

Computing will be carried out using jupyter notebooks (the same framework used also in the concurrent Machine Learning book).