WSS NEWS

News about activities of the Washington Statistical Society
A Chapter of the American Statistical Association

---

November 1998

---

Contents:

• 1999 FCSM Research Conference
• Annual Holiday Dinner!!!
• WSS Short Course
• The Roger Herriot Award
• A Short Course On Logistic Regression
• Seminar and Open Forum
• SIGSTAT Schedule 1998/1999
• Employment Opportunities
• Short Courses
• Seminars
• Area Meetings and Courses (Adobe Acrobat PDF file)
• Note From The WSS NEWS Editor
• WSS People

1999 FCSM Research Conference

ABSTRACTS DUE DECEMBER 1, 1998

To have a paper or demonstration considered for presentation, an abstract of at least one page will be required by December 1, 1998. Submit both a hard copy and an ASCII text file by Internet or on a 3 1/2" IBM-PC compatible disk. Limit columns to 7" or less. Include the presenter's name, affiliation, mailing address, telephone and fax numbers, and E-mail address, if available. To submit abstracts or obtain additional information, contact: Ruth Detlefsen, Bureau of the Census Bureau, Room 2641-3, Washington, D.C. 20233; Phone: 301-457-2665, Fax: 301-457-1343, and E-mail: fcsm@ccmail.census.gov. Additional information and periodic updates concerning the conference may be found on web site http://www.bts.gov/fcsm.

Return to top

Annual Holiday Dinner!!!

Return to top

WSS Short Course

Return to top

The Roger Herriot Award

Drs. Harrison and Tucker were recognized for their successful joint leadership of the Federal Interagency Research Subcommittee charged with identifying, developing, and interpreting research which proved essential to arriving at an informed decision on revising the standards that govern the collection and dissemination of race and ethnicity data by all federal agencies. In this capacity, they played major roles in designing the Supplement on Race and Ethnicity to the May 1995 Current Population Survey, and in designing two major tests (over 90,000 households) of race and ethnicity items conducted in 1996.

The work led by Roderick and Clyde resulted in a major reporting change in the 2000 Decennial Census - allowing individuals to designate themselves as belonging to one or more racial categories. Both Roderick and Clyde, as co-chairs of the research subcommittee, directed a team of some 40 people from nearly 20 federal agencies in this research, results of which will affect reporting in future surveys and censuses.

The Roger Herriot Award for Innovation in Federal Statistics is sponsored by the Social Statistics and Government Statistics Sections of the American Statistical Association, and the Washington Statistical Society.

Return to top

A Short Course On Logistic Regression

If you are a research or statistical professional, this short course, Logistic Regression Analysis, is relevant in research of interest to you such as: prediction of survival/non-survival in a case-control study of low birth-weight infants using a variety of indicators such as gestation period, APGAR score, mother's age, etc.; prediction of acceptance/non-acceptance of an offer of admission to College where the predictors include demographic, aptitude, and achievement variables as well as characteristics of the offer (e.g., presence/absence of scholarship aid); prediction of which individuals respond/do-not-respond to a mail survey in terms of certain experimental conditions such as an offered incentive and follow-up procedures; and prediction of business success/failure during the first five years after inception using predictors such as the type of business venture, capitalization, and quality of management. Logistic regression analysis (LRA) is related to multiple regression analysis (MRA) in that a set of independent, or explanatory, variables is used to predict success on some outcome, or dependent variable. But, unlike MRA, LRA is appropriate when the outcome variable takes on only two distinct values. Computational procedures for LRA are available in popular packages of statistical programs and this has made the procedure readily available to professionals involved in real-world research settings. Both SPSS and SAS will be used for exemplary applications.

Scope of The Workshop:

Day 1, 12/10/98: Morning: [9:00am to Noon] Overview of the theoretical model and methods of estimation; introduction to computer programs for LRA; interpretation of simple models. Afternoon: [2:00pm to 4:30pm] Use of coded and product variables in modeling; introduction to and illustration of model comparisons based on traditional chi-square procedures. Evening: [6:30pm to 9:00pm] Optional hands-on data analysis

Day 2, 12/11/98: Morning: [9:00am to Noon] Review and summary of Day 1 with emphasis on questions from participants; introduction to and illustration of information theoretic approaches to model comparisons. Afternoon: [2:00pm to 4:30] Treatment of special problems such as missing data and outlying predictor values; relation of LRA to log-linear models when all predictors are categorical; extension of LRA to cases with three or more values for the outcome variable.

Fees:

Registration by phone: (301) 405-3624; FAX: (301) 314-9245; e-mail: wfm@wam.umd.edu; or mail: Department of Measurement, Statistics, and Evaluation, 1230 Benjamin Building, University of Maryland, College Park, MD 20742

To Register

DMSE Seminar and Training Center
1230 Benjamin Building
University of Maryland
College Park, MD 20742

About the Instructor:

About the Location

Return to top

Seminar and Open Forum

Agenda topics for January 21 include: overview and history of official metropolitan areas and the Metropolitan Area Standards Review Project; review of findings from the 1995 "Conference on New Approaches to Defining Metropolitan and Nonmetropolitan Areas"; Discussion of alternative approaches and criteria for defining metropolitan and nonmetropolitan areas, including, among others, full national territory coverage; county and subcounty level classifications; and methods of measuring metropolitan and nonmetropolitan integration.

January 22 will be an open forum for expressing views to the Office of Management and Budget on the review of the metropolitan area standards and proposed alternatives. If you would like to present your views on this day, please call the COPAFS office (703/836-0404) by January 15, 1999. Oral statements will be limited to 15 minutes. Persons speaking are asked to bring three copies of their written statement. We unfortunately cannot accommodate last minute requests to present oral statements but will accept all written statements.

It be held at the Embassy Suites Alexandria, 1900 Diagonal Road, Alexandria, VA 22314 , 703/684-5900, across from the King Street Metro stop (Yellow and Blue lines). For overnight accommodations, call Embassy Suites 1/800/EMBASSY by December 21 for the discounted room rate of $134.00 plus tax (single occupancy). Mention that you are attending the COPAFS seminar.

The cost is $95.00 for the Seminar. There is no charge to attend the Open Forum on Friday, January 22, 1999. There is a limit of 150 participants for Thursday, January 21. For a registration form, to request a place on the list of speakers for January 22, or for more information, contact the COPAFS office: 703/836-0404 or copafs@aol.com.

Return to top

SIGSTAT Schedule 1998/1999

November 18: Proc-StatXact - A demo of this new SAS PROC from Cytel Software Corp., the developers of StatXact and LogXact for exact statistical inference. Proc-StatXact consists of 11 external SAS procedures which augment the existing small-sample capabilities of SAS.

The following 6 procedures provide exact inference for continuous data: PROC GOF - Goodness-of-fit testing; PROC ONESAMPL - One-sample testing; PROC PAIRED - Two related sample testing; PROC TWOSAMPL - Two-sample testing; PROC KRELATED - K related sample testing; and PROC KSAMPL - K-sample testing.

The following 5 procedures provide exact inference for categorical data: PROC BINOMIAL - Binomial inference; PROC STRATIFY - Stratified 2x2 and 2xc contingency tables; PROC RXC - rxc contingency tables; PROC ASSOCIAT - Measures of association; and PROC AGREE - Measures of agreement.

December 9: PASS - PASS 6.0 is a windows-based software tool for power analysis and sample size determination. It analyzes a broad range of experimental situations, including t-tests, F-tests, proportions, survival curves, correlations, bioequivalence tests, analysis of variance tests, log rank tests, multiple regression tests, and more. PASS is a product of NCSS.

All SIGSTAT meetings are held in Waugh A, 1800 M St, NW. Directions can be found on the SIGSTAT website, www.econ.ag.gov/sigstat/. Links to the above products are on the SIGSTAT Schedule page. First time attendees should contact Charlie Hallahan, 202-694-5051 or hallahan@econ.ag.gov and leave their name.

Return to top

Note From The WSS NEWS Editor

Return to top

WSS People

Return to top

First posted October 28, 1998
Last modified October 28, 1998

Webmaster: Dan Jacobs, Maryland Sea Grant