Understanding Statistical Models and their Social Science Applications

Understanding Statistical Models and their Social Science Applications

Stanford University

Description:

Statistical methods employed in the social science applications emphasizing mainly on the cause and effect determinations. The topics of the course are as follows:

• Multilevel models


• Analysis of covariance


• Path analysis


• Methods of matching and propensity score


• Longitudinal data


• Instrumental variables


• Mediating and Moderating variables


• Compliance

It's an intermediate level statistical course.

Overview of the course:

It is for the students who have gone through the instruction of intermediate-level in statistical methods and topics such as logistic regression, multiple regression and log-linear models. The course's content should offer the combination of at least some the previous instruction in statistical methods. The main task is to implant critical analysis and some introspection for the statistical methods uses that is common in medical applications and social sciences. The course focuses on understanding the vital information provided by the statistical model in both experimental and non-experimental social science setting, where more important is the non-experimental one.

Outline of the course:

• Week 1:


• An Introduction to the course


• Regression models and its properties


• Week 2:


• Experimental vs. observational studies


• Formulation of Neyman Rubin Holland


• Week 3:


• It contains Path analysis and causal modeling


• Display of Multiple regression explained through pictures


• Week 4:


• What is Multilevel data


• The Contextual effects


• Aggregation of biases


• Random effects models


• Week 5:


• The various uses and forms of the ANCOVA that includes regression continuity design.


• Week 6:


• Methods of Instrumental variable


• Simultaneous equations


• The reciprocal effects


• Week 7:


• Compliance and experimental protocols


• What are encouragement designs


• Week 8:


• Methods of Matching and propensity score


• Week 9:


• Experimental and non-experimental designs:


• Lord's paradox


• Repeated Measures Anova


• Value-added analysis


• Measurement of change


• Interrupted time-series


• Dead Week overflow and course summary


• Case studies discussion

Textbooks. :

The course contains online materials and auxiliary texts and was created by text of David Freedman. The students are greatly benefitted from this course because it encourages them to read some statistical literature and the research reports to enhance the texts.

Homework and exams:

Homework assignments are performed on a weekly basis based on the following class content that will be posted with solutions in the next class. However grades are not provided.

Statistical computing

Majority of the computing will be done using the R software and sometimes SAS, Matlab and Mathematica will also be used. Students can opt for any of the software they find comfortable.

Case Studies in Cause and Effect:

There are several case studies in Freedman's textbook:

1. Attention Deficit Disorder and TV



To see whether TV is bad for Attention Deficit Disorder and for this shall we use the structural equation model. Thus there exists any meaningful relationship between symptoms of attention-deficit disorder and television exposure



2. Money and Happiness



To see whether money can bring happiness i.e. do people get happy when they become richer?



3. God and IQ



Is it true that high IQ leads to transformation of academics to atheists? The reasons why who are theists are probably the ones with lower IQ.