Upcoming Events:

Friday, May 11, 2007

and

Monday, May 14, 2007

WORKSHOP ON LONGITUDINAL DATA ANALYSIS

Registration Required (click here to print registration form)

Friday, May 11th (8:30am-5pm)                 Monday, May 14th (8:30am-5pm)
Presented by Alan Acock                                            Presented by Walt Stroup

• On average, do college students tend to increase or decrease the days they use alcohol after their freshman year?
• Does this "drinking trajectory" vary much among students, or is the process consistent across students?
• Does marital satisfaction tend to increase, decrease, or stay the same in the first ten years of marriage?
• As countries become more democratic, do citizens become more philanthropic?
• Are increases in immigrant populations in cities associated with increases in new churches?
• Will a new math program increase the speed of learning math skills faster than an existing program?

All of the questions above are about change. Modeling change presents methodological challenges. The increasing availability of longitudinal data (a.k.a. repeated measures, panel data, over time data) and appropriate methods for modeling change (also called time, trajectories, latent growth curves, latent trajectories, multi-level models) provide exciting possibilities for social and behavioral researchers. Getting started with longitudinal data can be overwhelming both conceptually and practically. The SSP Core is offering two independent  full day workshops introducing latent growth curves/over time model analyses using practical examples. We will explore what questions modeling change can answer, and the different types of modeling techniques available for different questions.

Friday, May 11th 2007
The first workshop, lead by family researcher Alan Acock, will include an introduction to latent variable modeling (structural equation modeling) and learning the practical details of estimating latent growth curves with Mplus software. Participants will learn how to set up data, how to handle missing data, and how to do preliminary analyses of data to get an idea of the patterns that may emerge. In the next step, participants will estimate overall average trajectories (e.g., does the dependent variable tend to increase or decrease, or is the latent trajectory non-linear?), if there are differences in trajectories (e.g., do some children learn words faster than other children?), and why trajectories might differ (e.g. why do some adolescents start using alcohol and increase rapidly, while others have a more gradual increase in drinking days?).  Working with growth mixture models, growth models with binary outcomes and with outcomes that are counts will be introduced.

Monday, May 14th, 2007
The second workshop, led by Statistician Walt Stroup, will review basic change models/growth curve modeling from a SAS software perspective, and will extend the approach to nominal and categorical outcomes (e.g. depression episodes, offending, using substances, having a child). This workshop will also include models that are less about trajectories, and focus on how changes in predictors (e.g. changes in physical health) are associated with changes in outcomes (e.g. changes in mental health). These change models (sometimes called simply multi-level models of change, other times called fixed effects pooled time series) focus on the effects of changes in predictors on changes in outcomes.

Both workshops will emphasize practical applications based on working examples drawn from social science research. They will provide “hands-on computer time” for participants to run models, interpret output, and ask questions. Participants can come simply to get exposure in how to think about how to conceptualize testable questions about change and how to answer them with longitudinal data, as well as to gain technical insights for successfully running models and interpreting output. Participants can bring their own data and ask questions about their projects during breaks.

The workshops are designed for investigators who have never used either software program (all instructions and software will be available during the workshops) and for investigators who have used the software but want to know more. Follow up help for more complicated or advanced projects will be available through the SSP Core in the weeks and months following the workshops.

We anticipate that attending the workshops will help investigators to conceptualize and begin analyzing their data, and will also help investigators to more successfully be able to consult with experts when necessary to complete their projects.

A little background on the presenters:

Alan Acock (Ph.D., Washington State University) is Professor and former Chair of Human Development and Family Sciences at Oregon State University. He has also taught at Louisiana State University, Virginia Tech, and the University of Southern California. Alan has published 4 books, 20 book chapters, and 120 articles. He is a Fellow of the National Council on Family Relations, a winner of the Reuben Hill Award, several awards for teaching, and his book on Family Diversity and Well-Being received the 1995 Choice Award for Outstanding Academic Book. Alan has held elected offices in the American Sociological Association and the National Council on Family Relations. His substantive research has been on the effects of family structure on the well-being of family members and on intergenerational relations.  He has served on editorial boards of several substantive journals including the Journal of Marriage and Family. His methodological research has focused on structural equation modeling and missing values.   He has one book on Stata and is currently writing a book with Peter Lachenbruch on Advanced Statistics Using Stata.   His current statistical research is on applications of relative distributions to longitudinal interventions.  He serves on the editorial board of the Journal of Structural Equation Modeling.

Walt Stroup is Professor and Chair of the UNL Department of Statistics. He has been on the UNL faculty since 1979. He has extensive experience consulting and collaborating with academic and industry researchers in a wide variety of disciplines (psychology, education, public health, biological science, economic development in developing countries, agriculture, natural resources, pharmaceuticals, semiconductors, etc.). His research specialties are statistical modeling, especially generalized and mixed (a.k.a. hierarchical) models, and design of experiments and quasi-experiments. He was involved in a multi-state regional project that led to SAS’s development of its mixed model software (PROC MIXED, NLMIXED, and GLIMMIX). He has worked as a statistical and user advisor to SAS in the development and refinement of these PROCs. He has co-authored textbooks SAS for Mixed Models (1st and 2nd editions) and SAS for Linear Models, 4th ed. – books that have won awards for technical communication and have become the standard texts in their area for practical application on linear and mixed model analysis. He has taught a variety of undergraduate- and graduate-level modeling courses and mixed model short-courses and workshops to a variety of academic and industry audiences in the U.S., Europe, and Africa. He is co-founder of the SSP Core Facility.