Academic Year/course:
2016/17
8056 - Master in Current Democracies: Nationalism, Federalism and Multiculturalism
32229 - Techniques of Statistical Analysis I
Teaching Guide Information
Academic Course:
2016/17
Academic Center:
805 - Masters Centre of the Department of Political and Social Sciences
Study:
8056 - Master in Current Democracies: Nationalism, Federalism and Multiculturalism
Subject:
32229 - Techniques of Statistical Analysis I
Credits:
5.0
Course:
1
Teaching languages:
Theory: | Group 1: Pending |
| Group 2: Pending |
Teachers:
Jorge Rodriguez Menes, Sandra Bermudez Torres, Daniel Andres Ciganda Soliņo
Teaching Period:
First Quarter
Schedule:
Presentation
Techniques of Statistical Analysis I (Group 2) provides an introduction to regression analysis in the social sciences. The course builds on students’ knowledge of basic statistics to deepen their understanding of linear regression techniques, placing special emphasis on problems associated with model specification and solutions to other violations of regression assumptions. The course also provides an introduction to regression with independent and binary dependent categorical variables with logistic and probit models. The teaching approach avoids as much as possible formalistic-mathematical presentations and instead focuses on the logic that lies behind the techniques. The lectures combine a theoretical presentation of each subject with hands-on applications using STATA.
Associated skills
- General competences (Instrumental, Interpersonal and Systemic)
Instrumental Competences
Ability to analyse and synthesise
Planning and management of time
Basic computer competence and ability of using statistical software (STATA)
Information management abilities (ability to search and analyse information coming from a variety of sources)
Interpersonal Competences
Critical and self-critical ability
Team work
Ability to work in a interdisciplinary team
Ability to communicate with people that are not experts in the subject
Systemic Competences
Research abilities
Ability to work autonomously
Ability to generate new ideas (creativity)
Design and management of projects
- Knowledge of the basics of linear regression analysis
- An understanding of the assumptions underlying classical linear regression models, and of the importance for testing for, and applying remedial measures to, deviations from such assumptions
- Knowledge of regression techniques for categorical dependent variables
- Interpretation skills, as crystallized into the ability to describe results from intermediate statistical analyses
Prerequisites
Knowledge of elementary statistics
Contents
- There will be three basic content blocks.
1. The Classical Regression Model
Deterministic, probabilistic & observational models. Metric, significance & strength of linear relations. Estimation: OLS & ML. Practical significance an: Beta coefficients: total, part, and partial correlations; the R2 and its decomposition. Statistical significance: t and F tests. Model building: J & LR tests. Regression with dummy variables (ANOVA); regression with nominal & continuous independent variables (ANCOVA). Interaction effects.
2. Deviations from regression assumptions
Regression assumptions. Model miss-specification: wrong regressors, wrong functions (non-linearities and parameter inconsistencies). Tests and remedial measures for miss-specification. Inefficiencies: heteroskedasticity, autocorrelation, & multicollinearity. Testing and correcting inefficiencies with weighted and generalized least squares.
3. Regression applications: ANOVA, ANCOVA, & Regression with categorical dependent variables
Logit & probit models; logistic regression: metric, goodness of fit, significance & strength of relationship; maximum likelihood (ML) estimation; likelihood ratio & Chi-square; pseudo R2 measures. Multinomial models.
Teaching Methods
Each weekly 3-hour session will be divided into two parts: a lecture-type, 1:45 minute’s session where the main aspect of the weekly topics will be laid out, and a 1 hour practical session where students will practice their newly acquired knowledge with Stata.
Evaluation
- Assessment will be based on the 0-10 scale
0-4.9 – Fail
5 -6.9 – Pass
7-8.9 – Good
9-10 – Outstanding
- 2 take-home tests at weeks 5 (due on week 7) and 10 (due two weeks after classes end), with an emphasis on problem solving and interpretations of results using real data. Each will count 50% of the final mark.
- Every week, students will be given sets of computer exercises to complete at home (due in class on the following week). The problems will not be marked, i.e., but for every two sets not returned on time, 1 full mark will be deducted from the final mark.
Bibliography and information resources
Week
|
Classroom activity
|
Readings & Incidences
|
1
|
Lecture
|
Simple regression (1)
Basic notation. Deterministic & probabilistic models. Error term. Expected values. OLS parameter estimation.
|
Baum: chapters 2 & 3
Agresti & Finlay: Chapter 9
Kennedy: chapters 1 & 2
|
Lab
|
Working with Stata: Basics operations with files & variables
|
2
|
Lecture
|
Simple regression (2)
Betas, Covariance & Correlation. R2. Hypothesis testing
|
Lab
|
Producing and interpreting regression output with Stata
|
3
|
Lecture
|
Multiple regression
Notation. OLS estimation. Part & partial correlations. Multiple Determination. Adjusted R2. T tests.
|
Baum: chapter 4.1 – 4.4, 4.6
Kennedy: chapter 4
Agresti & Finlay: Chapter 11.1 to 11.3, 11.7 & 11.8
|
Lab
|
Multiple regression with Stata: basic output
|
4
|
Lecture
|
Model building strategies
Complete and Reduced Models. F & J tests Deductive and inductive strategies. Forward & Backward selection
|
Baum: chapter 4.5
Agresti & Finlay: Chapter 11.6, 14.1
Kennedy: chapters 3 & 5.1-5.2
|
Lab
|
Simple and complex hypothesis-testing with Stata
|
5
|
Lecture
|
Regression with categorical independent variables
ANOVA & ANCOVA. Marginal and interaction effects
|
Baum: chapter 7.1 – 7.2
Agresti & Finlay: Chapter 12 & 13
Kennedy: chapters 8 & 11 (except ridge regression & with principal components)
|
Lab
|
Regression with dummy & continuous variables in Stata
First take-home test handed in to students
|
6
|
Lecture
|
Specification error
Omission of relevant variables & wrong functional forms. Causal Models. Exogeneity and endogeneity
|
Baum: chapter 5
Agresti & Finlay: Chapter 10, 14.4
Kennedy: chapters 5.3 - 5.5, 6 & 7 (excl. limited depende nt var)
|
Lab
|
RESET test. Non-linear functions & linear transformations
|
7
|
|
Inefficiencies
Non-spherical disturbances (heteroskedasticity) & multicollinearity. WLS, Robust standard errors
|
Baum: chapter 6.2
Agresti & Finlay: Chapter 14.2 & 14.3
Kennedy: chapters 8.1-8.3 & 11 (except ridge regression & with principal components)
|
Lab
|
Tests of heteroskedasticity in Stata, Weighted Least Squares
|
8
|
Lecture
|
Regression with binary dependent variables
Logit models. Estimation, testing, and interpretation. Odds ratios. Wald and LR tests
|
Baum: chapter 10.1
Agresti & Finlay: Chapter 15
Kennedy: chapter 15.1
|
Lab
|
Logistic regression in Stata
|
9
|
Lecture
|
Model building and testing with logistic regression
Logistic regression with continuous and dummy variables Two- and three-way interaction effects
|
Baum: chapter 10.1
Agresti & Finlay: Chapter 15
Kennedy: chapter 15.1
|
Lab
|
Regression with dummy & continuous variables in Stata. LR and Bayesian Model Selection
Second take-home test handed-in to students
|
10
|
Lecture
|
Probit models
The Probit transformation. Estimation, fit and testing. Marginal Effects
|
Baum: chapter 10.1
Kennedy: chapter 15.1
|
Lab
|
Probit regression with Stata
|
Baum Christopher F. 2006. An Introduction to Modern Econometrics Using Stata. Stata Press
Agresti, A. & Finlay, B. 1997 Statistical Methods for the Social Sciences. 3rd edition. Prentice Hall.
Kennedy, Peter. 2003. A Guide to Econometrics, Fifth Edition. The MIT Press.
- Complementary bibliography
Wooldridge, Jeffrey M. 2003. Introductory Econometrics. A Modern Approach. 2nd Ed. Thompson
Stock, James H. & Watson, Mark W. 2003. Introduction to econometrics. Boston: Pearson Education.
Chatterjee, S., Hadi, A. S., and Price, B. 2000. Regression Analysis by Example. Third Edition. New Yor: John Wiley and Sons, Inc.
Gujarati, Damodar. 2002. N. Basic Econometrics, Fourth Edition. McGraw-Hill.
Maddala, G.S. 2001. Introduction to Econometrics, Third Edition. Wiley.