The course is dividend in two modules. Module A (6 credits) covers the theory and application of the linear regression model and of the binary logit model. Module B (6 credits) deals with the theory and application of generalized linear models. In particular: models for categorical response variables, models for counts, multilevel models.
The models will be used for the analysis of real data using Stata, with particular attention to the choice of the model and interpretation of the results.
Fahrmeir, L., Kneib, T., Lang, S., Marx, B. (2013). Regression. Models, Methods and Applications. Springer.
Learning Objectives
Knowledge of the properties and potentialities of the main statistical models. Ability to select the most appropriate model depending on the research aims and the available data. Ability to estimate the selected model, to evaluate the model adequacy and to interpret the results. Ability to write a short report on the analysis.
Prerequisites
Basic knowledge of calculus, matrix algebra, probability and statistical inference.
Compulsory prerequisite exam: "Statistical Inference".
Teaching Methods
Classroom lessons and lab activities.
Further information
Type of Assessment
The exam consists of a written test with questions about theory and exercises. Exercises require the elaboration (using a pocket calculator) of the results derived from the application of statistical models, discussing the specification choices and providing relevant interpretations. During the course individual homeworks are assigned (to be performed using data analysis software) that contribute to the final mark only at the first examination taken. The oral exam is optional for students who completed at least 50% of the homeworks.
EXAM DURING THE COVID-19 EMERGENCY: whenever it is forbidden to conduct the exam at the University, written exam is replaced by a test on Moodle, while the oral exam is conducted through Google Meet.
Course program
Module A
1. The linear regression model: a review with applications
1.1. Simple linear regression
1.2. Multiple Linear regression
1.3. Non linear functions
1.4. Interactions
1.5. Estimation
1.6. Inference
1.7. Model selection
1.8. Diagnostics
2. Binary responses and logistic regression
2.1 Estimation and inference
2.2. Goodness of fit and model choice
2.3 Prediction and ROC curve
Module B
3 .Theory of generalised linear models
3.1. exponential family
3.2. generalised linear models
3.3. estimation
3.4. inference
4. Models for categorical responses
4.1. Ordinal responses
4.2 Nominal responses
5. Models for counts and Poisson regression
6. Multilevel models
7. Quantile Regression