This course offers an introduction to data analysis in Social Science using linear in variables models. The emphasis is on the interpretation of the model and the application of statistical inference techniques to solve relevant practical problems. The course discusses in detail how to make inferences under non-standard situations, relevant in Social Sciences, due to the nature of the variables in the model (qualitative, transformed to allow nonlinear relations or non-observable,) or to the nature of data. The rigorous formal justification of the applied statistical inference techniques is out of the scope of this course. The background in Probability, Statistics, Algebra and Calculus offered in Mathematics I & II and Statistics I & II is more than enough for this course.
The course has two objectives. On the one hand, understand the interpretation and all the methodological aspects of estimating causal relationships between variables in different contexts. This includes the interpretation of causal relationships in linear and non-linear models in variables, binary regression models, and models with endogenous explanatory variables. It also includes learning the fundamentals of inferences based on least squares, maximum likelihood, and instrumental variables. On the other hand, the student will be able to make inferences in the studied models using real data with the help of the GRETL program. The student will be examined for each and every one of these aspects.
A very important aspect of the course consists of using Econometrics software packages. The most used in class is GRETL, but we also use E-Views. It is essential that the student has a personal computer with at least GRETL installed. The midterms exams, and possibly the final, require using GRETL. Students must attend all classes, both magistral and reduced, with their personal computers.
At the end of the course, the student will acquire a good working knowledge on the interpretation of the linear regression model, discriminating between alternative specifications by means of statistical inference, and using GRETL for estimation and hypothesis testing.