Instructor(s)
Lectures and computer lab: prof. Jacques Hagenaars
email: Jacques.A.Hagenaars "at" uvt.nl

Computer lab help (and author of this page): Aleš Žiberna
email: ales.ziberna "at" fdv.uni-lj.si

Basic information

Basic information is available at web page of Graduate study of Statistics at University of Ljubljana. The most important information is:

• Course description.
• Preliminary schedule (time table and planed topics by days) The time table shows times when classrooms are reserved. We will usually finish earlier (about an hour).

If anyone wants to receive potential news (like important new content of the web page) let me know on (ales.ziberna "at" fdv.uni-lj.si).

Materials

• Basic literature in listed in Course description. The most important parts of the literature are available in FORA (photocopy shop on FDV). You have to demand the materials for prof. Hagenaars' course. These are:
  • Hagenaars J.A. (1990) Categorical Longitudinal Data: loglinear panel, trend, and cohort analysis, Chapters 2 and 3. Newbury Park: Sage
  • Hagenaars, Jacques A. and Allan McCutcheon (Eds.) (2002), Applied Latent Class Analysis. New York: Cambridge University Press, Chapter 9
• In addition, we suggest the following book: Agresti, Alan (2002) Categorical data analysis, 2nd ed. New York: Wiley.
• Transparencies from two prof. Hagenaars' courses, covering similar topics, one on loglinear models and one on Latent class analysis.
• Lem related links (the software we are using):
• Web page
• LEM manual
• LEM's author, Jeroen K. Vermunt, PhD thesis. Useful for understanding Loglinear (and other) models and LEM .
• A zip file with LEM + examples (original source)
• Data:
• Examples from the materials available in FORA + this table
• Data for examples from book Applied Latent Class Analysis and just datafile for example in Chapter 9.
• Main dataset - European social survey (ESS) dataset from round 3 (year 2006) mainly for Slovenia. The data are available on ESS web page. You have to register to download the data, but registration is free. Data for Slovenia for year 2006 in SPSS format are also available here:
• With user defined missing values
• without user defined missing values (they are converted to system missings - more convenient for import into R)
• Variables and quetions with coding
• Your data.
• R-code for data preparation for LEM. (modified 18.4.2008 - now also suitable for missing data)

Assigment

• Write a paper on approximately 10 pages (not more than 15!!!). When you send the paper, also send datafile, input and output files (they do not count in the 10 pages).
• Deadline: June 2, 2008
• Your paper should include:
  • problem specification; underlying 'theory'
  • data description (sample, variables, data, etc.)
  • (process of) model specification
  • interpretation of results
• You have to use:
  • both structural manifest variables and indicators
  • at least 2 indicators
  • at least 3 structural variables
  • at least 2 variables with more than 2 categories
  • lat least one latent variables
  • at least 1 three-variable effect (regardless if it is significant or not), which you also have to interpret
• If you want to deviate from these instructions or cannot complete the assignment by June 2, send an e-mail to prof. Jacques A. Hagenaars and specify the reason (if you want to extend the deadline also suggest an alternative deadline that you will definitely meet).