Statistics Collaboratory


Collaboratory short courses and presentations are available without charge to faculty, graduate students, and staff of UCR. We invite members of the community to contact the organizers (listed with each, below), if they are interested in attending.

SAS Day at UCR

SAS Day Slideshow Presentation

Thursday, November 20, 2008 1:00-5:00PM
Bourns Hall A265

Topic: An Introduction to SAS ODS Statistical Graphics in SAS 9.2

Presenter: Mike Patteta, Senior Instructor (from SAS Institute Inc.)
Time: 1:00-1:50PM

Graphics are essential in modern data analysis. With ODS Statistical Graphics in SAS 9.2, the statistical procedures create graphs as automatically as they create tables. There are also new SAS/GRAPH procedures that streamline the process of making graphs for exploring data. The talk introduces statistical users to ODS Statistical Graphics. Examples illustrate basic functionality, which requires minimal syntax, and typical graph management tasks.

Topic: Analyze Survey Data in SAS Using Replication Methods

Presenter: Anthony B. An, Senior Research Statistician (from SAS Institute Inc.)

Increasingly, statisticians analyze data that come from probability-based sample surveys. The SURVEYMEANS, SURVEYFREQ, SURVEYREG, and SURVEYLOGISTIC procedures properly analyze complex survey data by taking into account the sample design to make statistically valid inference about the study population. Beginning in SAS 9.2, In addition to Taylor series method, these SAS procedures for survey data analysis provide replication methods, such as BRR and jackknife methods for variance estimation. They give more flexibility and power to SAS users to analyze their complex survey data. This presentation discusses how to estimate variances with jackknife method, or BRR method, or Fay’s BRR method in these survey procedures and how to create replicate weights, and to estimate the variance when you provide your own replicate weights. The talk also covers other variance estimation issues related to replication methods. 

Topic: The SAS GLIMMIX Procedure

Presenter: Phil Gibbs, Manager of the Advanced Analytics group, SAS Institute Inc.
3 – 3:50PM

The new GLIMMIX procedure allows generalized linear models to be fit while including random effects. We will explore the syntax for the procedure, graphical output from the procedure including the new DIFFOGRAM plot, common uses of the modeling methodology, and pitfalls to avoid when using the procedure.

Topic: Determining Power and Sample Size in SAS 9.2

Presenter: Mike Patetta, Senior Instructor (from SAS Institute Inc.)
4 – 4:50PM

“How many subjects do I need” is a common question in the planning of a statistical study. The question is an important one because the sample size should be large enough that an effect of a magnitude of scientific interest might also be statistically significant. To determine the appropriate sample size, the researcher must compute the power of the statistical test used in the study. It is recommended that you examine power across many different scenarios by varying the parameters that affect power. In this talk, you will learn how to estimate the parameters in a power analysis, how to conduct a sensitivity analysis of the parameters, and how to determine power and sample size in SAS 9.2. Examples will include the two-sample t-test, a multiple regression model, a two-way ANOVA model with covariates, and a logistic regression model.

Analyzing data as graphs: Representation, analytical questions, and statistical problems

Professor Robert A. Hanneman
Department of Sociology

University of California, Riverside

Wednesday, November 8, 2006 3:00-4:30PM

Many phenomena examined in physical, biological, social and cultural sciences can be usefully conceptualized as graphs or networks of objects (e.g. atoms or individual people) connected by relations (e.g. attraction or economic exchange). Representing data in this “relational” way suggests new approaches to understanding the behavior of the individuals and the populations of which they are a part. The “observations” in relational data – by definition – are not “independent” replications. This raises a number of issues for sampling, estimation, and inference that are not characteristic of non-graph data.

An Introduction to GIS and Spatial Modeling in the Social Sciences

Professor Robert Nash Parker
Director, Presley Center for Crime and Justice Studies
University of California, Riverside

Monday, March 3, 8:30-9:30 pm
Tuesday, March 4, 8:30-9:30 pm
Wednesday, March 5, 8:30-9:30 pm

There is increasing interest in many of the social sciences in GIS (Geographic Information Systems) and spatial modeling. GIS systems allow for the creation of databases that contain geospatially linked data from a variety of traditional (US Census) and non-traditional (address based event data) to address new kinds of questions and test a variety of new hypotheses.

The workshop, to be offered in the evenings on Monday, March 3rd, 8:30-9:30 pm; Tuesday, March 4, 8:30-9:30 pm, and Wednesday, 8:30 -9:30 pm, will assume no knowledge of GIS or Spatial modeling. The workshop will take place in the UCR Statistical Consulting Collaboratory, Rm 2228, Sproul Hall. We will begin with an introduction to geocoding and the production of thematic maps using ARCView GIS software, and continue with a discussion of database building for spatial analysis. We will next cover the basic statistical approach to geospatial data developed by geographers, economists, and public health. Applications and examples will be run by participants in the workshop using a specialized software package that runs under Mathematica and interactively performs a number of tests, calculations, and regression like models, with a number of examples presented.

Applied Survey Data Analysis Using SUDAAN

Professor Augustine Kposowa
Department of Sociology
University of California, Riverside

Survey research in many disciplines uses complex multi-stage sampling. Basic package software (e.g. SPSS, SAS) may compute incorrect estimates of standard errors with data of this type -- which can result in incorrect (and often "false positive") confidence for inference and wrong conclusions about hypotheses.

This workshop series will discuss the proper estimation of error variance in complex samples, and teach the use of SUDAAN software to get correct results. SUDAAN is widely used in epidemiological studies, and applies well to work with complex samples from many fields of study. RTI Software has kindly donated licenses for their state-of-the-art software to support this short course. SAS software will also be used.

The course will meet once each week from 5pm until 6:30pm on Tuesdays from April 1st (week 1 of the spring quarter) until April 15th (3rd week of the spring quarter) in the UCR Statistical Consulting Collaboratory located in Sproul Hall 2nd Floor, Room 2228. Enrollment is limited, and those who enroll should plan to attend all scheduled sessions. Attendees should have a basic knowledge of survey sampling, and of basic linear models. If you would like more information, or would like to apply for admission to this free workshop series, please contact Professor Kposowa.

An Introduction to Social Network Analysis using UCINET

Professor Robert Hanneman
Department of Sociology
University of California, Riverside

Social network analysis focuses on patterns of relationships among actors, rather than on the attributes of the actors themselves. This "relational" approach to understanding social structure has proven to be very useful in studying a wide variety of phenomena ranging from how individual attitudes are formed to patterns of inequality in the world system. Social network analysis applies basic ideas from the mathematical theory of graphs to understand the "structure" of social relations.

This workshop will focus on working with network data to provide insights into such things as diffusion, power and dependency, and social roles. Following the instructor's free on-line textbook, it will show how UCINET software from Analytic Technologies can be used to manipulate and analyze network data. Related analytic and graphical software (e.g. Pajek, NetDraw) will also be examined.

The course will meet once each week from 5pm until 7pm on Tuesdays from May 6th (week 6 of the spring quarter) until June 3rd (week 10 of the spring quarter) in the UCR Statistical Consulting Collaboratory, located in Sproul Hall 2nd Floor, Room 2228. Enrollment is limited, and those who enroll should plan to attend all scheduled sessions. Attendees should have a knowledge of basic statistics, but advanced preparation is not required. If you would like more information, or would like to apply for admission to this free workshop series, please contact Professor Hanneman.

Using the Collaboratory - Running Long Simulations on the Server

Mark Nicolay
Instructional Technologist
Computing & Communications

A workshop for faculty and graduate students who will be using Gauss software to run long simulations. The procedure for logging on to the server, uploading, and retrieving files will be demonstrated.

Wednesday, February 19, 11 am
Sproul 2228

More Information

General Campus Information

University of California, Riverside
900 University Ave.
Riverside, CA 92521
Tel: (951) 827-1012

Career OpportunitiesUCR Libraries
Campus StatusDirections to UCR

Collaboratory Information

Statistical Consulting Collaboratory
1438 Olmsted Hall

Tel: (951) 827-6062
Fax: (951) 827-3286
E-mail: karen.xu@ucr.edu