Lab notes for Statistics for Social Sciences II: Multivariate Techniques

Overview

Lab notes for Statistics for Social Sciences II: Multivariate Techniques for the course 2016/2017. The subject is part of the BSc in International Studies and the BSc in International Studies & Political Science, both from Carlos III University of Madrid.

Syllabus

The notes are available at https://bookdown.org/egarpor/SSS2-UC3M.

Here is a broad view of the syllabus:

1. Course introduction
   1. Some course logistics
   2. Software employed
   3. Why this software?
   4. Installation in your own computer
   5. R Commander basics
   6. Datasets for the course
   7. Main references and credits
2. Simple linear regression
   1. Examples and applications
   2. Some R basics
   3. Model formulation and estimation by least squares
   4. Assumptions of the model
   5. Inference for the model coefficients
   6. Prediction
   7. ANOVA and model fit
   8. Nonlinear relationships
   9. Exercises and case studies
3. Multiple linear regression
   1. Examples and applications
   2. Model formulation and estimation by least squares
   3. Assumptions of the model
   4. Inference for model parameters
   5. Prediction
   6. ANOVA and model fit
   7. Model selection
   8. Model diagnostics and multicollinearity
4. Logistic regression
   1. More R basics
   2. Examples and applications
   3. Model formulation and estimation by maximum likelihood
   4. Assumptions of the model
   5. Inference for model parameters
   6. Prediction
   7. Deviance and model fit
   8. Model selection and multicollinearity
5. Principal component analysis
   1. Examples and applications
6. Cluster analysis
   1. k-means clustering
   2. Agglomerative hierarchical clustering
7. Appendix A: Glossary of important R commands
8. Appendix B: Use of qualitative predictors in regression
9. Appendix C: Multinomial logistic regression
10. Appendix D: Reporting with R and R Commander
11. Appendix E: Group project

List of animations

Important: the links below point towards https urls with auto-signed SSL certificates. That means that you most likely will get a warning from your browser saying that "Your connection is not private". Click in "Advanced" and allow an exception. https has been considered since it allows to include the apps in both http and https websites.

• Simple linear regression
  • Regression coefficients, least squares and distance choice
  • Randomness of the regression line
  • Coverage of the confidence intervals
  • Confidence intervals for prediction
  • ANOVA decomposition
  • Dealing with nonlinear relationships
• Multiple linear regression
  • Least squares and distance choice
  • Assumptions of the linear model
  • ANOVA decomposition
• Logistic regression
  • Logistic curve and maximum likelihood fit
  • Randomness of the logistic regression curve
  • Confidence intervals for prediction

Contributions

Contributions and feedback on the notes are very welcome. Either send an email to [email protected] or (preferably) fork the repository, make your changes and open a pull request.

Acknowledged contributors:

• Pablo Brugarolas Navarro (mentioned problems about the presence of NA's for stepwise selection)

License

All material in this repository is licensed under CC BY-NC-SA 4.0.