SCS Short Courses, Fall 2000

SCS Courses Introduction to SPSS for Windows Data Analysis Using SAS Data Analysis and Statistical Graphics Using S-Plus Mixed Models for Longitudinal and Hierarchical Data Analysis Using SAS and S-Plus

Data Analysis Using SAS

Basic introduction: Sessions One and Two provide an overview of SAS and its underlying logic, an explanation of the use of the Display Manager System to run a SAS job, an introduction to the SAS Data Step for reading, transforming, and storing data, and a demonstration of how statistical analyses may be performed in SAS Insight.

Further topics: Sessions Three and Four will concentrate on SAS programming techniques to modify data and enhance SAS output. More statistical procedures will be introduced for general linear models.

Introduction to SPSS for Windows

This course presents the basics of the Statistical Package for the Social Sciences (SPSS) including an introduction to computing concepts and the essentials of SPSS. Other sessions will demonstrate various ways of formatting data and will cover data modifications, transformations and the use of SPSS system files.

Session One will introduce the computing concepts of SPSS, the different facilities for reading data into an SPSS spreadsheet, and saving SPSS data files for future use. At the end of the first session, participants should be able to run simple programs, including some statistical procedures.

Sessions Two and Three will cover basic data modifications, transformations and other functions available in SPSS. More statistical procedures will be introduced, with an emphasis on the use of graphical methods to examine univariate and bivariate relationships.

Session Four will cover Analysis of Variance and Least Squares Regression. Graphical techniques will be demonstrated.

The following zip file: f99spss.zip (last updated October 1, 1999) contains SPSS data sets and scripts used in the course. Copy this file (in binary mode) to an appropriate directory and use a zip extractor (such as pkunzip or WinZip) to expand the zip archive into the original files.

Mixed Models for Longitudinal and Hierarchical Data Analysis Using SAS and S-Plus

Mixed effects models are extensions of ordinary linear regression that allow random as well as fixed regression coefficients. These models can be used for many types of data that are otherwise difficult to analyze: complex longitudinal data (e.g. with time-varying covariates or with observations recorder at different times for different subjects), unbalanced repeated measures data, multilevel data and unbalanced block designs.

This course will present a brief review of linear models and an overview of the basic theory and applications of mixed models. The fitting of mixed models will be illustrated with a number of examples covering a range of applications. The NLME version 3.3 software of Douglas Bates and Jose Pinheiro will be used for its flexibility in visualizing data for mixed models. NLME is written in S-Plus. Every model fitted with NLME will be translated into PROC MIXED or PROC NLMIXED in SAS.

The course will consist of lectures supplemented with hands-on exercises in a computer lab to allow participants to gain experience in the use of these methods.

The following topics will be covered:

Hierarchical models: Introduction to hierarchical data structures. Why use hierarchical models? The structure of the linear mixed model: fixed effects, random effects, variance and covariance components. Some applications: one-way ANOVA, regression with means as outcomes, random coefficient model, intercept and slopes as outcomes. Key concepts: shrinkage, borrowing strength, avoiding bias, contextual versus compositional effects. Model building and diagnostics. Consequences of measurement error and how to adjust for it. Introduction to non-linear hierarchical models and hierarchical models for categorical responses.

Longitudinal models: The classical starting points: univariate and multivariate repeated measures. Why do we need more? How mixed models are used to fit longitudinal data. Modeling correlation. Model construction and diagnostics. Missing data patterns. Modeling panel attrition. Logistic regression for binary responses. Non-linear models for binary and categorical responses. Markov transition models.

Data Analysis and Statistical Graphics Using S

Many feel that S is the language of choice for the development of new statistical tools and for interactive analysis of data. A commercial version, S-Plus, is available on Unix workstations and on PCs under Windows. At York, S-Plus has been installed on the "phoenix" Unix system, in the Hebb Lab in Psychology and in the Gauss Lab in Mathematics and Statistics.

The purpose of this course is to learn how to use S in a Unix or in a Windows environment. The course will have both lecture and hands-on components. Participants will have access to an account to run S-Plus. The three sessions of the course will cover approximately the following material:

• I The Unix or Windows Environment. Unix: Logging in, editing with vi, directory structure, basic file manipulation, introduction to the X-Window environment. Microsoft Windows: The S-Plus graphical user interface. Overview of S.
• II Basic Use of S: Data input, manipulating data, printing, history mechanism, S arrays and data frames, random number generation, arithmetic operators; functions for manipulating data structures: apply and table; help facilities.
• III Programming and Graphics in S: Writing S functions, one-and two-dimensional graphs, interactive graphs, trellis graphics.
• IV Introduction to the use of statistical functions in S: Regression and regression diagnostics based on the linear model function, lm().
The main reference for the course will be "Modern Applied Statistics with S-PLUS, 3rd edition" by W. N. Venables and B. D. Ripley (Springer, 2000).

SCS home page