Sociology

LAURA BOCCI

The main aim is to endow students with multivariate statistical tools for analysing data for political, economic and social sciences applications. Practical lessons dealing with real-world examples are designed to allow students to improve abilities in collecting, analysing, interpreting and presenting findings and data.
Specifically, upon completion of the course the students will have acquired the following skills:
- Knowledge and understanding: the knowledge of the main issues and essential concepts of multivariate statistical analysis (dependence, dimension reduction, classification) and the standard methodologies to face and handle such problems (parametric and non-parametric methods, Multivariate Linear Regression, Cluster Analysis and PCA).
- Ability to apply knowledge and understanding: the ability to define a research design and to select appropriate data processing techniques for sociological research.
- Making judgements: the ability to collect and use quantitative and qualitative data relating to the social and economic sciences, and to critically interpret the results obtained using statistical evidence and sound reasoning. This particular competence develops through the presentation of applications and classroom discussions stimulated by the teacher based on analysed examples.
- Communication skills: the ability to use rigorous but comprehensible language to communicate effectively about the statistical methodologies employed and the results obtained. This skill will be developed through practical sessions and supplementary materials, including exercise sheets.
- Learning skills: the ability to independently learn data analysis techniques for professional activities or further academic study.

Basic Statistics.

Statistical sampling, and data collection techniques in social and economic research.
Matrix representations of multidimensional data. Data matrix, data cleaning, data pre-processing, covariance and correlation matrices, proximity matrices.
Graphical representations of multidimensional data.
Unsupervised learning. Multivariate statistical techniques: Cluster Analysis (hierarchical, non-hierarchical) and Segmentation; Principal Component Analysis.
Supervised learning: Multiple Linear Regression models and extensions; Logistic Regression.
Applications to real-world data using Matlab statistical software.
Case studies.

1) Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani. Introduction to Statistical Learning with Applications to R. Springer
https://www.statlearning.com/
Handouts provided by the teacher

In-class sessions comprise didactic lectures, practical sessions, hands-on exercises, demonstrations, discussion.
Lectures will be aimed at stimulating both interaction with students and their problem solving skills. Therefore, each topic will be supplemented by examples and hands-on exercises in order to facilitate the understanding of statistical tools and their use in social issues.

Lecture attendance is not mandatory but is strongly recommended given the know-how-oriented course setting.

The evaluation is performed by a final written examination, carried out during the scheduled exam sessions (3 calls in June / July session, 1 call in the September session, 2 calls in January / February session).
The written exam consists of 15 exercises with both theoretical and practical questions. The test is intended to assess knowledge and understanding and the student's ability to apply knowledge and understanding. The student must indicate the correct answer and return the calculations necessary to obtain the indicated result.
For completing the test, the students will have 120 minutes and can use a calculator and statistical tables.
In itinere evaluation will be performed. The course includes one partial exam and a project work. The partial exam consists of 9 questions including both theoretical and practical (exercises) items. For the partial exam, students will have 60 minutes and may use a calculator and statistical tables.
The project work consists of applying the techniques learned to a real dataset.
The final grade is the sum of the score obtained in the partial exam and the project work.
If a student's grade for the partial exam and project work is rejected, they must take the full exam in one of the official exam sessions.

In determining the final grade, the assessment takes into account the following elements:
1. the thought process followed by the student in solving the proposed questions;
2. the correctness of the procedure chosen by the student to get the solution;
3. the adequacy of each solution proposed by the student, considering both the type of question and his expected competences;
4. the use of a correct and proper language.
A grade of at least 18/30 is required to pass the exam. Students must demonstrate a) to have acquired a sufficient knowledge of the topics covered in the course and b) to be able to identify statistical techniques and tools - simple but adequate - for the solution of the proposed real problems.
The grade 30/30 cum laude is assigned to those students who demonstrate an excellent knowledge of all the topics covered during the course and strong critical thinking skills. Students must also demonstrate to be able to identify the most suitable statistical techniques and tools, both simple and complex, for solving real problems.

1) Given the output of the linear model, what does the regression coefficient of the independent variable X1 indicate?
2) A tour operator wants to launch a new package for an organized trip on the market. To test the appeal of the package, they conduct a sample survey on a sample of 65 customers. In the sample, 29 customers stated that they appreciated the package. (a) Calculate the confidence interval for the proportion of satisfaction with a confidence level of 99%. (b) Considering the estimate of the proportion of customer satisfaction and the margin of error obtained in point (a), what should the sample size be if one wanted to halve the margin of error at a 99% confidence level?

• Topic 1 (8 hours). Sampling
  Theory 
  • Books: Handouts provided by the teacher 


• Topic 2 (6 hours). Data Matrices and Transformations. Eigenvalues. Eigenvectors.  Distance and
  Similarity  Matrices. Introduction to Matlab statistical software 


• Topic 3 (6 hours). Unsupervised
  Learning. Hierarchical Clustering: methods and properties.


• Topic 4 (6 hours). Unsupervised
  Learning. Non-Hierarchical Clustering: methods and properties.


• Topic 5 (6 hours). Unsupervised
  Learning. Principal Component Analysis


• Topic 6 (10 hours). Supervised
  Learning. Multiple Linear Regression and extensions.


• Topic 7 (6 hours). Supervised
  Learning. Regression
  with a binary dependent variable: Logistic Regression.