Multivariate

Statistical Methods

 Research Master

Examination and Paper

DATE TO HAND IN PAPER AND EXAM: NOVEMBER 23^RD
DAY:  WEDNESDAY
 

EXTRA INFORMATION FOR THE EXAM
One part of a selection of the paper of Sniderman et al. was stuck in the copy-machine, so that part is missing. In order to be safe I have made the who selection of that paper available: download it.

THE EXAMS
They will be handed to you on November the 9th, after the presentations.  The following texts are required readings:
All the sheets which are downloadable from this site under:
1. Course Material,
2. Presentation Theory (except for latent growth curve modeling)
3. Presentations Practicum.
And the book chapters of one the books the are recommended, eg. from the Stevens book you should read:
4. Chapter 2: Matrix Algebra
5. Chapter 3: The Regression Model
6. Chapter 11: Confirmatory and Exploratory Factor Analysis (Note that PCA is generally discussed under the heading of Exploratory Factor Analysis, though the model is very different.)

THE PAPER
The paper should have a structure similar to:
1. Overview/Introduction
2. Theory/literature
2. End with a conceptual model
3. Data (that is who has collected the data that you use, and some sample characteristics)
3. Data - discussion of the variable in the model and with which questions are they measured?
3. Data - If necessary an overview of the analysis of variables by definition and by derivation.
4. Analyses - Discussion of the model and assumptions
5. Results
6. Conclusion/Discussion
But any structure that is internally consistent and is capable of clarifying your points is alright. If you need examples, you can look to journals like 'American Sociological Review', or 'Public Opinion Quarterly', which can be accessed via www.jstor.org, or via the digital library. But only from within the University, or with special software tools at home. But the papers below in the section literature: social capital also give a good idea about what is expected. Of course, I do not expect a publishable paper, but I do expect that you show that you understand the structure of such a paper, and that you can write in an academic way.

This course will provide an overview of and introduction to some of the many multivariate statistical methods for data analysis. These methods have one thing in common: they are applied when studying the relationship between two or more variables. The difference on the other hand is in the type of variables they can handle, which limits each techniques’ scope of application. To make a informed decision for a specific multivariate method requires insight in the statistical methods, and this course is intended to provide that insight.

There are so many different multivariate statistical methods that it is impossible to treat every method within a course of 8 weeks. The techniques that will presented are multiple regression (continuous dependent variables), logistic regression (for binary and polytomous variables), confirmatory factor analysis, and principal component analysis. In addition, a short course on matrix algebra is given. The reason for that is that the mathematics in which multivariate analysis is cast is matrix algebra. Basic knowledge of matrix algebra is also essential for those students going on to the advanced courses in structural equation modeling or regression analysis.

The course is designed for students who have finished their bachelor courses in statistics. In addition, it is assumed that students are familiar with LISREL and SPSS. Students can download the student version of LISREL (Tutorial) at the SSI-website, and a version of SPSS can be obtained from Surfspot.nl.

Below an overview is presented of the contents of each meeting. More detailed information can be found in the remaining of this syllabus.

Students participating in this course are expected to work in small groups of 3 or 4 persons and complete their own research project, about which they write a paper in journal style. In  week 2 each group has to give a presentation of their research proposal, and during the last week, it is expected that each group presents their results. The paper and presentations are evaluated together, and will approximately be 40% of the total grade. Then there will be some individual home exercises that if correct and delivered on time can yield about 15% of the total grade. The remaining of the grade, 45%, will be a classical exam.

Literature: Multivariate techniques

*

James P. Stevens (2001). Applied Multivariate STATS 4th PR. Lawrence Erlbaum Associates, Mahwah (NJ).

*

Literature: Social Capital