ANNOUNCE: 2007 Summer School in Longitudinal Data Analysis and Missing Data - U of Bologna

[entrepreneurship-phd at lists.uni-due.de]

2007 Summer School in Longitudinal Data Analysis and Missing Data
taught by Prof. Paul D. Allison
organized by
Management Department
University of Bologna
Bologna, June 4-8, 2007*
Deadline: May 15th 2007

*School Objectives*

The course will introduce statistical methods and techniques required
for the analysis of longitudinal data and for managing missing data
problems. In the first three days of the school participants will learn
how to apply methods for data which varies both across units and over
time; these include models for "panel" data, "time-series
cross-sectional" data, and the like. Topics will include fixed- and
random-effects models, GLS-based approaches to panel data, GEE models,
random coefficient models and dynamic models with lagged dependent
variables. Linear, logistic as well as count data models will be
covered. The last two days will be devoted to methods and techniques for
handling missing data. Participants will learn how to apply
state-of-the-art maximum likelihood and multiple imputation techniques
in order to address problems that cannot be reliably solved with
traditional approaches like listwise deletion or regression imputation.
These new methods for handling missing data have been around for at
least a decade, but have only become practical in the last few years
with the introduction of widely available and user friendly software.
While underlying theory will be thoroughly discussed, the greatest
emphasis will be on application and interpretation of models and results
(see full program for further details). Participants are therefore
strongly encouraged to bring their own datasets for analysis. The
computer statistical packages SAS and AMOS will be used extensively for
data analysis.

*Instructor*

Paul Allison is Professor and Chair of Sociology at the University of
Pennsylvania, where he teaches graduate methods and statistics. He is
widely recognized as an extraordinarily effective teacher of statistical
methods who can reach students with highly diverse backgrounds and
expertise. Allison is the author of many statistical books sold
worldwide such as /Fixed Effects Regression Methods for Longitudinal
Data Using SAS/ (SAS Institute 2005), /Missing Data/ (Sage 2001),
/Logistic Regression Using SAS(r): Theory and Application/ (SAS
Institute 1999), /Multiple Regression: A Primer/ (Pine Forge 1999),
/Survival Analysis Using SAS(r): A Practical Guide/ (SAS Institute
1995), /Event History Analysis/ (Sage 1984), and numerous articles on
regression analysis, log-linear analysis, logit analysis, latent
variable models, missing data, and inequality measures. A former
Guggenheim Fellow, he is also on the editorial board of Sociological
Methods and Research. In 2001 he received the Paul Lazarsfeld Memorial
Award for Distinguished Contributions to Sociological Methodology.

*Prerequisites*

Participants should enter this workshop with an active working knowledge
of the topics covered in a standard course in /Regression Analysis./ A
familiarity with the basic chi-square test for two-way contingency
tables and elementary regression and ANOVA is also presumed.

*Location, format, materials*

The 5 day course will be held at Villa Guastavillani,
(http://www.almaweb.unibo.it/about_alma.html), a wonderful 16th century
Villa located on the hills of Bologna.

Here is a typical day's schedule:

9-11 Lecture
11-12:30 Supervised computing
12:30-1:30 Lunch break
1:30-3:30 Lecture
3:30-5:30 Computing and consulting

Participants will receive a 100-page manual containing detailed lecture
notes (with equations and graphics), examples of computer printout, and
many other useful features. This document frees participants from the
distracting task of note taking. Participants may also want to refer to
Professor Allison's books, /Fixed Effects Regression Methods for
Longitudinal Data Using SAS/ (SAS Institute 2005) and /Missing Data/
(Sage 2001). The books are optional.

*Application procedure*

Deadline for applications is May 15th 2007. A complete application
package should include the following items:

* A CV indicating nationality, date and place of birth, current
affiliation and position.
* Doctoral students are required to specify the title of the doctoral
program and the year in which they are currently enrolled.
* A statement (no longer than 1 page) that describes (i) the current
research activities of the applicant and (ii) his/her broad research
interests.

All applications will be accepted up to the limit of 20. In case of more
than 20 applications the following criteria will be used for selection:
(1) preference for academic researchers in social sciences; (2)
preference for PhD students and junior researchers with a basic training
in statistics; (3) relevance of the School to the applicant's research
program; (4) date of application. Applicants selected out in the first
step of the process will be included in a waiting list and possibly
accepted later. Applications should be sent exclusively to the following
address: bolognaschool at gmail.com <mailto:bolognaschool at gmail.com> The
School is coordinated by Raffaele Corrado (raffaele.corrado at unibo.it
<mailto:raffaele.corrado at unibo.it>) and Simone Ferriani
(simone.ferriani at unibo.it <mailto:simone.ferriani at unibo.it>).

*Registration*

Registration fee is *Euro 450*. The money will be used to cover daily
lunch catering services, prepare the course material, purchase the SAS
licenses. Selected participants will receive details on the payment
procedure. Registration will be confirmed as soon as the payment is
received. Cancellation less than 2 weeks prior to the workshop is
subject to a Euro 150 late withdrawal fee. Participants are expected to
make their own arrangements for housing. Special discounted rates are
available through the Management Department.//


*COURSE PROGRAM*

*Longitudinal Data Analysis*

*A Course on Regression Methods for Panel Data (3 Days)*

------------------------------------------------------------------------

*Course Outline*

1. Opportunities and challenges of panel data.
a. Data requirements
b. Control for unobservables
c. Determining causal order
d. Problem of dependence
e. Software considerations

2. Linear models
a. Robust standard errors
b. Random effects models
c. Fixed effects models
d. Hybrid models

3. Logistic regression models
a. Robust standard errors
b. Subject-specific vs. population averaged methods
c. Random effects models
d. Fixed effects models
e. Hybrid models

4. Count data models
a. Poisson models
b. Negative binomial models
c. Fixed and random effects

5. Linear structural equation models
a. Fixed and random effects in the SEM context
b. Models for reciprocal causation with lagged effects

*MISSING DATA *

*A Course on Modern Methods for Handling Missing Data (2 Days)*

------------------------------------------------------------------------

Conventional methods for missing data, like listwise deletion or
regression imputation, are prone to three serious problems:

* Inefficient use of the available information, leading to low power and
Type II errors.
* Biased estimates of standard errors, leading to incorrect p-values.
* Biased parameter estimates, due to failure to adjust for selectivity
in missing data.

More accurate and reliable results can be obtained with maximum
likelihood or multiple imputation.

These new methods for handling missing data have been around for at
least a decade, but have only become practical in the last few years
with the introduction of widely available and user friendly software.
Maximum likelihood and multiple imputation have very similar statistical
properties. If the assumptions are met, they are approximately unbiased
and efficient--that is, they have minimum sampling variance. What's
remarkable is that these newer methods depend on less demanding
assumptions than those required for conventional methods for handling
missing data. At present, maximum likelihood is best suited for linear
models or log-linear models for contingency tables. Multiple imputation,
on the other hand, can be used for virtually any statistical problem.

This course will cover the theory and practice of both maximum
likelihood and multiple imputation. Maximum likelihood for linear models
will be demonstrated with Amos, a software package designed for
estimating structural equation models with latent variables. Multiple
imputation will be demonstrated with two new SAS procedures (PROC MI and
PROC MIANALYZE) and two Stata commands (ICE and MICOMBINE).

*Course outline*

1. Assumptions for missing data methods
2. Problems with conventional methods
3. Maximum likelihood (ML)
4. ML with EM algorithm
5. Direct ML with Amos
6. ML for contingency tables
7. Multiple Imputation (MI)
8. MI under multivariate normal model
9. MI with SAS
10. MI with categorical and nonnormal data
11. Interactions and nonlinearities
12. Using auxiliary variables
13. Other parametric approaches to MI
14. Linear hypotheses and likelihood ratio tests
15. Nonparametric and partially parametric methods
16. Sequential generalized regression models
17. MI and ML for nonignorable missing data