Boston Chapter of the American Statistical Association Short Course
www.amstat.org/chapters/boston

Individual Growth Modeling: Modern Methods for Studying Change

Judith D. Singer and John B. Willett

Harvard Graduate School of Education

"The two authors are internationally renowned social statisticians [who]… are driven by the wish to improve the quality of published empirical research… This book is a great addition to their efforts and will certainly have a substantial impact on the analyses of longitudinal data carried out in many fields." International Journal of Epidemiology

"It will come as no surprise to those familiar with Judith Singer and John Willett's didactic journal articles to learn that they have written a terrific textbook on longitudinal data analysis." Sociological Methods and Research

Date & Time Friday, May 20, 2005

8:30 AM – 9:00 AM Check-in

9:00 AM – 5:00 PM Course

Location Larsen Hall Room G-08

Harvard Graduate School of Education

Appian Way, Harvard Square Cambridge, MA

Cost $100 for chapter members, $130 for non-members, and $70 for students (ID must be presented at check-in, or send a copy with your advance registration). This will cover the cost of the course, morning coffee, lunch, and course materials.

Registration Limited to 90 participants. Mail a check (along with your name and e-mail address) for the course fee, payable to BCASA, addressed to BCASA, c/o Tom Lane, 128 Bingham Rd., Carlisle, MA 01741. Registrations will be accepted until the course fills, but should arrive no later than May 13. If space remains, on-site registration will be allowed. No refunds after May 13 unless you have someone else to fill the space. Receipts will be available at the event. Inquiries can be sent to tlane@mathworks.com.

Directions   See www.gse.harvard.edu/~admit/directions.html for directions to the Ed School campus. This website includes campus maps, subway information, and a list of local parking garages.

Abstract Based on their book, Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence (Oxford, 2003), Singer and Willett will give an accessible yet in-depth presentation of multilevel models for individual change. Using real data sets from published studies, the instructors will take participants step-by-step through complete analyses, from simple exploratory displays that reveal underlying patterns through sophisticated specifications of complex statistical models. All concepts will be illustrated using real data sets from recent studies. Implementation using a variety of software packages will also be discussed (including SAS, Stata, SPSS, Splus, MLwiN and HLM). The course’s emphasis is data analytic, focusing on five linked phases of work: articulating research questions; postulating an appropriate model and understanding its assumptions; choosing a sound method of estimation; interpreting analytic results; and presenting findings—in words, tables, and graphs—to both technical and non-technical audiences. Thoughtful analysis can be difficult and messy, raising delicate problems of model specification and parameter interpretation. The default options in most computer packages do not fit the statistical models people generally want. The course’s goal is to provide you with the short-term guidance needed to start using the methods quickly, as well as with long-term advice to support your work wisely once begun. The morning session will begin with descriptive and exploratory methods, followed by a detailed discussion of basic model specification, model fitting, and parameter interpretation. The afternoon session will extend these principles to the messy arena of real world applications, delving into topics such as centering predictors, handling variably spaced measurement occasions and varying numbers of waves, including time-varying predictors, and fitting discontinuous and non-linear change trajectories. The target audience is professionals who have yet to fully exploit these longitudinal approaches. Some participants may be comfortable with multilevel modeling, although we assume no familiarity with the topic. Although methodological colleagues are not the prime audience, they, too, should find much of interest.

Book The course is based on the first half of the instructors’ recent book, known by the acronym ALDA. You can learn more about ALDA at gseacademic.harvard.edu/~alda/. Participants are strongly encouraged to obtain copies of ALDA in advance of the workshop from either www.amazon.com or Oxford University Press www.oup.com. We are also investigating the possibility of having copies for sale at the event. Check with Tom Lane, tlane@mathworks.com, to determine if the book will be available at the course. ALDA is supported by a companion website at the UCLA Academic Technology Services, www.ats.ucla.edu/stat/examples/alda. There you can download the many data sets used throughout the book and code for reproducing all the book’s analyses, using your preferred major software package.

Instructors Judith D. Singer is the James Bryant Conant Professor of Education and John B. Willett is the Charles William Eliot Professor of Education, both at the Harvard Graduate School of Education. Singer holds a PhD in Statistics from Harvard University; Willett holds a PhD in Quantitative Methods from Stanford University. Collaborators for 20 years, their professional lives focus on improving the quantitative methods used in social, educational and behavioral research. Singer and Willett are best known for their contributions to the practice of individual growth modeling, survival analysis, and multilevel modeling, and to making these and other statistical methods accessible to empirical researchers. You can learn more about the instructors on their home pages: gseweb.harvard.edu/~faculty/singer and gseacademic.harvard.edu/~willetjo