WELCOME
   The Third Annual U.S. Environmental Protection Agency (EPA)
Conference on Statistics is sponsored by the Statistical Policy
Committee (SPC) and co-funded by the Statistical Policy Branch,
Office of Policy, Planning and Evaluation, and the  Personnel
Management Division, Office of Administration and Resources
Management.  The Conference is held solely for the benefit and
participation of EPA personnel.  Programming and  arrangements
for the Conference have been provided by  the SPC Conference
Planning Committee, chaired by Mel Kollander. Members of the
SPC Planning Committee for the  1987 EPA Conference on
Statistics are as follows:
      R. Clifton Bailey, OW       Henry Kahn, OW

      John Creason, HERL-RTP   Mel Kollander, OPPE

      Mark Goldstein, Region III   Elizabeth Margosches, OTS

      William Hunt, OAQPS-RTP   William Nelson, EMSL-RTP

                      Wayne Ott, ORD

-------
                                  CONTENTS


                                                                          Page

CONFERENCE AGENDA	  1

ATTENDEE LIST	  7

ABSTRACTS	 16

     Criteria for Statistical Evaluation of Data
        "Realistic Criteria for Statistical Evaluation," Bertram D. Litt.. 16
        "Criteria for Tolerance Limits," Richard A. Levy	 17
        "Criteria for Combining Data Sets," Herbert LaCayo	 18
        "Criteria for Selecting a Statistical Model," R. Clifton Bailey... 19

     Criteria for Health and Exposure Assessments
        "Environmental Health Models," Allan H. Marcus	 20
        "A Dose-Response Model for Teratology Data," Dorothy Wellington... 21
        "Empirical Bayes Rates for Rare Events in Small Areas,"
           Max A. Woodbury	 22
        "The Use of Statistics in Setting the Gasoline Lead
           Standard," Joel D. Schwartz	 23

     Data Presentation Devices
        "Producing Visuals," Bruce Mitchell	 24

     Poster Session
        "How Good Are Analytical Results From Your Water
           Laboratory?"  Paul W. Britton	 25
        "Evaluation and Presentation of Water Quality Data for the
           Missouri River Near Omaha, Nebraska," Norman H. Crisp,
           Joseph E. Joslin, Thomas J. Hollovay	 26
        "Small Sample Properties of Estimators of the Odds Ratio
           Under Constant Multiple Matching," Ayenew Ejigou	 26a
        "Applications of a New NOEL/RfD Statistical Procedures for
           Small Samples with Dichotomous  (Incidence) Response Data,"
           Linda Erdreich, Kenneth Brown, and Lisa LaVange	 27
        "Geostatistical Analysis," George T. Flatman	 29
        "Expert Systems to Assist in Decisions Concerning Land
           Disposal of Hazardous Wastes," Daniel G. Greathouse	 30
        "Statistical Methods for the Analysis of Toxic Interactions,"
           Richard C. Hertzberg	 32
        "Data Quality of the FY 1986 Regional Ambient Fish Tissue
           Monitoring Program," Thomas T. Holloway and Bruce Littell	 33
        "Rubberband Regression," Robert W. Jernigan	 34
        "Niagara River Study," Henry D. Kahn	 35
        "An Evaluation of Pesticide Residues in Fish Tissue,"
           David A. Parrish	 36

                                     ii

-------
Poster Session  (Continued)
   "Using Pharmacokinetic Modes To Improve Dosimetry for
      Noncontinuous Inhalation Exposures to Vapors,"
      Lorenz Rhomberg	 37
   "Modeling Rare Events in Small Populations,"
      Wilson B. Riggan and John P. Creason	 38
   "Numerical Method of Estimation of the Elemental
      Concentration of Mitochondnan Cells in Biological Tissue
      Samples," Judy A. Stober, Dennis Black, and Tammy Mills	 39
   "GRAPE:  Graphic Representations of Activity Patterns and
      Exposure," Jacob Thomas, Perng-Fei Wang, Herb Hunt,
      and Wayne Ot t	 40

SIMS:  Statistical Methodology for Environmental Issues
   "SIMS Studies:  Development of Environmentrics,"
      Donald L. Thomsen, Jr	 41
   "Role of Statistics in Environmental Monitoring and
      Regulation," Paul Switzer	 42
   "Indirect Exposure Assessment:  Implications for
      Regulatory Programs and Public Health Risks," Naihua Duan	43
   "Epidemiology in Risk Assessment for Regulatory Policy,"
      Alice Whittemore	 44
   "Investigating a Cancer Cluster in a Massachusetts Community,"
      James H. Ware,  Roger Day, Daniel Wartenberg, Marvin Zelen	46
   "Geostatistics for the Environment:  Probability Tools for
      Decisionmaking," Andre G. Journel	 48
   "Acid Deposition:   Models and Data Analysis," James V. Zidek	49

Workshops
   "Environmental Statistics," Paul Switzer	 50
   "Power of Persuasion Using Your PC," Linda A. DeLuise	 51
   "Expert System Advisor for Survey Design," William P. Smith	 52
   "Using STATGRAPHICS to Analyze Data on the PC,"
      David J. Svendsgaard	 53

Communications Within EPA	 54
                                111

-------
                  Agenda for the
Third Annual EPA Conference on Statistics

                  March 17-20, 1987
           Pavilion Tower Conference Center
                 Virginia Beach, Virginia
              Environmental Data Analysis
             Section Organizer:  William C. Nelson

                  Computer Graphics
            Section Organizer: Mark N. Goldstein

              Communications Within EPA
            Section Organizer:  William F. Hunt, Jr.

                    Poster Session
             Section Organizers: R. Olifton Bailey
                            Henry D. Kahn

-------
                               AGENDA FOR THE
                  THIRD ANNUAL EPA CONFERENCE ON STATISTICS
                              March 17-20, 1987
                              TUESDAY, MARCH 17

 3:00 p.m. - 6:00 p.m.      Registration and Check-in
                            (Main Lobby)

 7:30 p.m. - 9:30 p.m.      Opening Reception
                            (Reception Area, Conference Level)
                                WEDNESDAY, MARCH 18

 8:45 a.m. - 10:00 a.m.     KEYNOTE SESSION
                            (Lynnhaven Inlet)

     Opening Remarks and Introductions:  N. Phillip Ross, Chief, Statistical
     Policy Branch,  Office of Policy,  Planning and Evaluation, Chemical and
     Statistical Policy Division

     Keynote Addresses:
        Mary Allen,  Director, Office of Standards and Regulations, OPPE
        Vaun Newill, Assistant Administrator, Office of Research and
        Development


10:00 a.m. - 10:15 a.m.     Break


10:15 a.m. - 12:15 p.m.     CRITERIA FOR STATISTICAL EVALUATION OF DATA
                            (Lynnhaven Inlet)

     Session Chair:   Bertram Litt,  Retired, Biostatistics Group, Office of
     Pesticide Programs (OPP), Hazard Evaluation Division

     Importance of Realistic Criteria for Statistical Evaluation — Bertram
     Litt, OPP, Hazard Evaluation Division, Retired

     Criteria for Tolerance Limits — Richard Levy, Biostatistics Group,
     Office of Pesticide Programs,  Hazard Evaluation Division

     Criteria for Combining Data Sets — Herbert LaCayo, Office of Pesticide
     Programs, Hazard Evaluation Division

* Please Note:  Sessions will begin and end PROMPTLY at the specified times.

                                      1

-------
                                                   AGENDA FOR THE THIRD ANNUAL
                                                  EPA CONFERENCE ON STATISTICS
                                                             March 17-20, 1987
Wednesday (Continued)

     Criteria for Appropriate Statistical Models — R. Clifton Bailey, Office
     of Water Regulations and Standards,  Analysis and Evaluation Division


12:15 p.m. - 1:15 p.m.      Lunch
                            (Chatney's)


 1:15 p.m. - 3:00 p.m.      CRITERIA FOR HEALTH AND EXPOSURE ASSESSMENTS
                            (Lynnhaven Inlet)

     Session Chair:  John Creason,  Chief, Biostatistics Branch, Office of
     Research and Development,  Health Effects  Research Laboratory

     Environmental Health Models — Allan H. Marcus, Visiting Scientist,
     Office of Air and Radiation, Office of Air Quality Planning and Stan-
     dards,  Research Triangle Park

     A Dose-Response Model for Teratology Data — Dorothy G. Wellington,
     Office of Policy, Planning and Evaluation, Chemical and Statistical
     Policy Division

     Empirical Bayes Rates for Rare Events in  Small Areas — Max Woodbury,
     Professor,  Center for Demographic Studies, Duke University

     Uses of Statistical Analysis in Setting the Gasoline Lead Standard —
     Joel D. Schwartz, Office of Policy,  Planning and Evaluation, Economic and
     Regulatory Analysis Division


 3:00 p.m. - 3:15 p.m.      Break


 3:15 p.m. - 4:00 p.m.      DATA PRESENTATION  DEVICES
                            (Portside Ballroom)

     Session Chair:  Mark Goldstein,  Information Resources Management, Region
     III, Policy and Management Division

     Producing Visuals ~ Bruce Mitchell, Manager,  Mutual Information Center,
     Research Triangle Park

-------
                                                   AGENDA FOR THE THIRD ANNUAL
                                                  EPA CONFERENCE ON STATISTICS
                                                             March 17-20, 1987
Wednesday (Continued)

 4:15 p.m. - 6:00 p.m.      POSTER SESSION
                            (Crystal Lake and Redwing Lake)

     Session Organizers:  R. Clifton Bailey, Statistics Section, Office of
     Water Regulations and Standards, Analysis and Evaluation Division; and
     Henry D. Kahn, Chief, Statistics Section, Office of Water Regulations and
     Standards, Analysis and Evaluation Division
                             THURSDAY, MARCH 19

 8:30 a.m. - 10:15 a.m.     SIMS:  STATISTICAL METHODOLOGY FOR ENVIRONMENTAL
                            ISSUES
                            (Lynnhaven Inlet)

     Session Chair:  Wayne Ott, Chief, Air Toxics and Radiation Staff, Office
     of Research and Development, Environmental Monitoring Systems Division

     SIMS Studies:  Development of Environmetrics — Donald L. Thornsen,
     President, SIMS

     Role of Statistics in Environmental Monitoring and Regulation — Paul
     Switzer, Professor,  Department of Statistics, Stanford University

     Indirect Exposure Assessment:  Implication for Regulation Programs and
     Public Health Risks  — Naihua Duan, Member SAB Review Committee, Econo-
     mics Department,  The Rand Corporation

     Epidemiology in Risk Assessment for Regulatory Policy ~ Alice
     Whittemore, Professor, Department of Family, Community, and Preventive
     Medicine,  Stanford University


10:15 a.m. - 10:30 a.m.     Break
10:30 a.m.  - 12:00 p.m.      SIMS:   STATISTICAL METHODOLOGY FOR ENVIRONMENTAL
                            ISSUES (Session Continued)

     Community Studies:   Toxic Wastes and Disease Clusters — James H. Ware,
     Professor,  Department of  Biostatistics,  Harvard University

-------
                                                   AGENDA FOR THE THIRD ANNUAL
                                                  EPA CONFERENCE ON STATISTICS
                                                             March 17-20, 1987
Thursday (Continued)

     Geostatistics for the Environment:  Probability Tool for Decisionmaking -
     Andre G. Journal, Professor,  Applied Earth Sciences, Stanford University

     Approaches to Space-Time Acid Deposition Modeling and Implications for
     Impact Detection — James V.  Zidek, Professor, Statistics Department,
     University of British Columbia


12:00 p.m. - 1:00 p.m.      Lunch
                            (Chatney's)
 1:00 p.m. - 3:00 p.m.      SESSION I — WORKSHOPS

   i/l:00 - 2:45
        Environmental Statistics — Workshop la (Linkhorn Bay)
        Paul Switzer, Statistics Department, Stanford University

     1:00 - 2:00
        Power of Persuasion Using Your PC — Workshop 2a (Chesapeake Bay)
        Linda A. DeLuise,  Office of Pesticide Programs, Program Management and
        Support Division

        The Artificial Statistician — Workshop 3a (Broad Bay)
        William P.  Smith,  Office of Policy,  Planning and Evaluation, Chemical
        and Statistical Policy Division

     2:00 - 3:00
        Using STATGRAPHICS To Analyze Data on the PC — Workshop 4a (Broad
        Bay)
        David Svendsgaard, Office of Research and Development, Health Effects
        Research Laboratory

        Power of Persuasion Using Your PC — Workshop 2b (Chesapeake Bay)
        Linda DeLuise (Repeated)
 3:00 p.m.  - 3:15 p.m.       Break

-------
                                                   AGENDA FOR THE THIRD ANNUAL
                                                  EPA CONFERENCE ON STATISTICS
                                                             March 17-20, 1987
Thursday (Continued)

3:15 p.m. - 5:15 p.m.       SESSION II — WORKSHOPS

     3:15 p.m. - 4:45 p.m.
        Environmental Statistics - Workshop Ib (Linkhorn Bay)
        Paul Switzer

     3:15 p.m. - 4:15 p.m.
        Power of Persuasion Using Your PC — Workshop 2c (Chesapeake Bay)
        Linda DeLuise

        Using STATGRAPHICS To Analyze Data on the PC — Workshop 4b (Broad
        Bay)
        David Svendsgaard

     4:15 p.m. - 5:15 p.m.
        The Artificial Statistician — Workshop 3b (Broad Bay)
        William P. Smith

        Power of Persuasion Using Your PC — Workshop 2d (Chesapeake Bay)
        Linda DeLuise (Repeated)


 8:00 p.m.  - 10:00 p.m.      Open Workshop (Computers will be available for
                            individual use)
                            (Broad Bay and Chesapeake Bay)
                              FRIDAY,  MARCH 20

 8:30 a.m.  - 10:15 a.m.     COMMUNICATIONS WITHIN EPA:  PANEL PRESENTATION
                            (Lynnhaven Inlet)

     Session Organizer:  William Hunt, Chief, Data Analysis Section, Office of
     Air Quality Planning and Standards,  Monitoring and Reports Division

     Session Chair:  Marcia E. Williams,  Director, Office of Solid Waste

-------
                                                   AGENDA FOR THE THIRD ANNUAL
                                                  EPA CONFERENCE ON STATISTICS
                                                             March 17-20, 1987
Friday (Continued)

     Panel Members:
        Don R. Clay, Deputy Assistant Administrator, Office of Air and
          Radiation
        Rebecca Hanmer,  Deputy Assistant Administrator, Office of Water
        Victor J. Kimm,  Deputy Assistant Administrator, Office of Pesticides
          and Toxic Substances
        Joseph Carra, Acting Director, Waste Management Division, Office of
          Solid Waste
        Barry Nussbaum,  Chief, Field Operations and Compliance Policy Branch,
          Office of Mobile Sources, Office of Air and Radiation


10:15 a.m. - 10:30 a.m.      Break


10:30 a.m. - 12:15 p.m.      COMMUNICATIONS WITHIN EPA:  Discussion Session


12:30 p.m. - 1:30 p.m.      LUNCHEON SESSION
                            (Starboard Ballroom)

     Session Chair:  Mel Kollander, Statistical Policy Branch, Office of
     Policy, Planning and Evaluation

     Overview of COMMUNICATIONS WITHIN EPA Session — Marcia Williams,
     Director, Office of Solid Waste

     Closing Remarks --   N. Phillip Ross,  Chief,  Statistical Policy Branch,
     Office of Policy, Planning and Evaluation, Chemical and Statistical
     Policy Division

     Conference Wrapup — Mel Kollander, SPB, OPPE, Chemical and Statistical
     Policy Division


 2:00 p.m.                  Buses Leave

-------
ATTENDEE LIST

-------
                      1987 EPA CONFERENCE ON STATISTICS
                       MARCH 17-20, VIRGINIA BEACH, VA
*Harv Allen
   Director, Office of Standards and Regulations, Office of  Policy, Planning,
   and Evaluation, (PM-223),  401 M Street, S.W., Washington, D.C.  20460,
   (202) 382-4001, (FTS)  382-4001.

Ruth H. Allen
   Office of Acid Rain, Environmental Monitoring, and Quality  Assurance,
   Office of Research and Development, (RD-680), 401 M  Street,  S.W.,
   Washington, D.C. 20460, (202) 382-5945, (FTS) 382-5948.

*R. Clifton Bailey
   Analysis and Evaluation Division, Office of  Water Regulations  and
   Standards, Office of Water,  (WH-586), 401 M  Street,  S.W., Washington, D.C.
   20460, (202) 382-5411, (FTS) 382-5411.

Janes C. Baker
   Region 3  (HV'.'.-SR) , Suite 500, 999 18th Street, Denver,  CO 80202-2405,
   (303) 293-1524, (FTS)  564-1524.

Steven Bayard
   Carcinogen Assessnent Group, Office of Health and Environmental Assessreat,
   Office of Research and Development  (RD-689) , 401 M Street,  S.IJ.,
   Washington, D.C. 20460, (202) 382-2680, (FTS) 382-2630.

Jerome Blondell
   Office of Pesticide Programs,  (TS-769C) , 401 M Street,  S.W.,  Washington,
   D.C. 20460, (703)  557-2564,  (FTS) 557-2564.

Hark C. Blossar
   Water Pollution Branch, Delaware Department  of Natural  Resources and
   Environmental  Control, 89 Kings Highway, P.O. Box 1401,  Dover,  DE  19903
   (302) 736-4590.

*Paul W. Britton
   Environmental  Monitoring Systems Laboratory, Office  of  Research and
   Development, 26 W. St. Clair Street, Cincinnati, OH  45268,  (513) 569-7325,
   (FTS) 684-7325.

"Joseph Carra
   Waste Management Division,  Office of Solid  Waste,  ('JH-565), 401 M  Street,
   S.W., Washington D.C. 20460, (202)  382-7919, (FTS)  382-7919.

Chao W. Chen
   Carcinogen  Assessment Group, Office of Health and Environental Assessment,
   Office of Research and Development,  (RD-689), 401 M  Street, S.W.,
   Washington, D.C. 20460,  (202)  382-5719,  (FTS) 382-5719).

Leo J. Clark
   Central  Regional Laboratory, 839 Bestgate  Rd., Annapolis, MD 21401,
   (301) 224-2740, (FTS) 922-3752.

•Participants

-------
*Don Clay
   Deputy Assistant Administrator for Air and Radiaton,  (ANR-443), 401 H
   Street, S.W., Washington, D.C. 20460, (202) 382-7403,  (FTS) 382-7403.

Jim Coqliano
   Cancer Assessment Group, Office of Health and Environmental Assessment,
   Office of Research and Development,  (RD-689), 401 M Street, S.V.,
   Washington, D.C. 20460,  (202) 386-2575, (FTS) 382-2575.

Margaret G. Conomos
   Exposure Evaluation Division, Office of Toxic Substances,  (TS-798), 401 M
   Street, S.W., Washington, D.C. 20460, (202) 382-3958,  (FTS) 382-3958.

James Craig
   Office of Policy, Planning, and Information, Office of Water,  (WH-563),
   401 M Street, S.W., Washington, D.C. 20460, (202) 382-2791, (FTS)  382-2791.

*John P. Creason
   Health Effects Research  Laboratory, Office of Research and Development,
   (MD-55), Research Triangle Park, MC 27711, (919) 541-2598, (FTS)  629-2598.

Thomas C. Curran
   Office of Air Quality Planning and Standards, Office  of  Air and Radiation,
   (MD-14), Research Triangle Park, NC 27711, (919) 541-5558, (FTS)  629-5558.

James M. Daley
   Information and Regulatory Systems Division, Office of Standards  and
   Regulations, Office of Policy, Planning, and Evaluation,  (PH-223), 401 M
   Street, S.W., Washington, D.C. 20460,  (202) 382-2743,  (FTS) 382-2743.

John Davidson
   Office of Policy Analysis, Office of Policy, Planning, and Evaluation,
   (PM-221), 401 K Street,  S.W., Washington, D.C.  20460,  (202) 382-5484,
   (FTS) 382-5484.

*Linda A. DeLuise
   Registrations Division,  Office of Pesticide Programs,  (TS-767C),  401  M
   Street, S.W., Washington, D.C.  20460,  (703) 557-8542,  (FTS)  557-8542.

*Miahua  Duan
   Economics Division, The  Rand  Corporation,  1700  Main  Street, Santa Monica,
   CA 90406.

George H. Duggan
   Office of Air Quality  Planning and Standards, Office  of  Air and  Radiation,
    (MD-12), Research Triangle Park, NC  27711,  (919) 541-5645, (FTS)  629-5645.

Richard  G. Eilers
   Drinking Water  Research  Division,  Water Engineering  Research  Laboratory,
   Office of Research and Development,  26 W.  St. Clair  Street, Cincinnati,  OH
   45268,  (513) 569-7809,  (FTS)  684-7809.

*Avenew  Eliqou
   NRC Fellow,  Health Effects Research  Laboratory, Research Triangle Park,  NC
   27711.

"Participants

-------
Dan Engleberg
   U.S. General Accounting Office, Rra. 5844, 441 G  Street,  M.W.,  Washington,
   D.C. 20548, (202) 275-3760.

Evan J. Englund
   Environmental Monitoring Systens Laboratory, Office  of  Research  and
   Development, P.O. Box 15027, Las Vegas, NV 89114,  (702)  798-2248,
   (FTS) 545-2248.

*Linda Erdreich
   Environmental Criteria and Assessment Office, Office of Research and
   Development, 26 West St. Clair Street, Cincinnati, OH 45268,
   (513) 569-7531,  (FTS) 684-7531.

Bob Faoro
   Office of Air Quality Planning and Standards, Office of Air and  Radiation,
   (MD-14), Research Triangle Park, NC 27711,  (919)  541-5558,  (FTS)  629-5558.

Jerzv  Filar

Bernice Fisher
   Hazard Evaluation Division, Office of Pesticide  Programs,  (TS-796C),  401 M
   Street, S.W., Washington, D.C. 20460,  (703)  557-0959, (FTS)  557-0959.

*George T. Flatman
   Environmental Monitoring Systems Laboratory, Office  of Research  and
   Development, P.O. Box 15027, Las Vegas, NV  89114,  (702) 798-2628,
   (FTS) 545-2628.

Ruth Foster
   Office of  Standards  and Regulations,  Office  of  Policy, Planning, and
   Evaluation,  (PM-223), 401 M Street, S.W., Washington, D.C.  20460,
   (202) 382-2468,  (FTS) 382-2468.

Neil H. Frank
   Office of  Air Quality Planning and Standards, Office of Air and  Radiation,
   (MD-14), Research Triangle  Park, NC 27711  (919)  541-5558,  (FTS)  629-5558.

Mary J. Frankenberrv
   Office of  Toxic  Substances,  (TS-798) ,  401 M Street,  S.W.,  Washington,  D.C.
   20460,  (202) 382-3890,  (FTS) -382-3890.

Warren P.G. Freas  III
   Office of  Air Quality Planning  and Standards, Office of Air and Radiation,
    (MD-14), Research Triangle  Park, NC 27711,  (919) 541-5558,  (FTS) 629-5558.

George Garland
   Characterization and Assessment  Division, Office of  Program Management and
   Support, Office  of  Solid Waste,  (WH-562B),  401  M Street, S.W., Washington,
   D.C.  20460,  (202)  382-4761,  (FTS)  382-4761.

J. Shermer  Garrison
   Technical  Analysis  Division,  Maryland Office of Environmental Programs,
   P.O.  Box 13387,  Baltimore,  MD  21203,  (301)  225-6285.

 "Participants

-------
*Mark Goldstein
   Information Resources Management, Policy and Management  Division, Region
   III, (3PM50), 841 Chestnut Street, Philadelphia, PA 19107,
   (215) 597-3530,  (FTS) 597-3530.

Stephen Goranson
   Environmental Monitoring Branch, BSD, Region V  (MD-5SEM),  536 S. Clark
   Street, Chicago, IL 60605, (312) 353-2306,  (FTS) 353-2306.

Paul D. Graves
   Budget Division, Office of the Comptroller,  (PM-225), 401  M Street, S.W.,
   Washington, D.C. 20460, (202) 382-4166, (FTS) 382-4166.

*Daniel Greathouse
   Hazardous Waste Engineering Research Laboratory, 26 W. St. Clair Street,
   Cincinnati, OH 45268, (513) 569-7859, (FTS) 684-7859.

*Rebecca W. Hanger
   Director, Office of Water Enforcement and Permits, Office  of Water,
   (EH-335), 401 M Street, S.W., Washington,  D.C.  20460, (202) 475-8488,
   (FTS) 475-8488.

Jeffrey A. Hassen
   Pennsylvania Department of Environmental Resources, Bureau of Waste
   Management, P.O. Box 2063, 7th floor, Harrisburg, PA 17120, (717) 737-6239.

TOP Heiderscheit
   Health Effects Research Laboratory, Office  of Research and Development,
   (MD-55), Research Triangle Park, NC 27711,  (919) 541-2281, (FTS) 629-2281.

*Richard C. Hertzberg
   Environmental Criteria and Assessment Office, Office of  Research and
   Development, 26 West St. Clair Street, Cincinnati, OH 45268,
   (513) 569-7582,  (FTS) 684-7582.

Matthew y. Hnatov
   Office of Water Regulations and Standards,  Office of Water, (VH-586), 401 M
   Street, S.W., Washington, D.C. 20460,, (202) 382-5412, (FTS) 382-5412).

John Hollev
   Office of Mobile Sources, Office of Air and Radiation, (EN-397F), 401 M
   Street, S.W., Washington, D.C. 20460, (202) 382-2635, (FTS) 382-2635.

*Thoras T. Hollowav
   Region VII, Environmental Services Division, 25 Funston  Road, Kansas City,
   KS 66115,  (913) 236-3884,  (FTS) 757-3884.

Daniel N. Hopkins
   Industrial Permits, Waste Management Division,  Office of Water,  Region VI,
   (6WPI), Renaissance Tower, 1201 Eln Street, Dallas, TX 75270,
   (214) 767-4378.

Kimberly Hummel
   Region III, 841 Chestnut Street, Philadelphia,  PA 19107,  (215)  597-3362,
   (FTS) 597-3362.

"Participants
                                      10

-------
'William F. Hunt, Jr.
   Office of Air Quality Planning and Standards, Office of Air and Radiation,
   (MD-14), Research Triangle Park.  NC 27711, (919) 541-3334,  (FTS) 629-5558.

*Robert Jernigan
   Statistical Policy Branch, Office of Policy, Planning, and  Evaluation,  (PM-
   223); Department of Mathematics and Statistics, The Anerican University,
   Washington, D.C. 20016, (202) 885-3120.

Barnes Johnson
   Statistical Policy Branch, Office of Policy, Planning, and  Evaluation,
   (PH-223), 401 M Street, S.W., Washington, D.C. 20460,  (202) 382-2684,
   (FTS) 382-6226.

*Andre G. Journel
   Professor, Applied Earth Sciences, Stanford University, Mitchell Building,
   Room 313, Stanford, CA  94305.

*Henrv D. Kahn
   Office of Water Regulations and Standards,  (WH-586), 401  M  Street,  S.W.,
   Washington, D.C. 20460, (202) 382-5406,  (FTS)  382-5406.

*Victor J. Kinrn
   Deputy Assistant Administrator for Pesticides  and Toxic Substances,  {TS-
   788), 401 M Street, S.W., Washington, D.C. 20460,  (202) 382-2910,  (FTS)
   382-2910.

Kathleen D. Knox
   Statistical Policy Branch, Office of Policy, Planning  and Evaluation,
   (PM-223), 401 M Street, S.W., Washington, D.C.  20460,  (202) 382-2714,
   (FTS) 382-2714.

*Mel Kollander
   Statistical Policy Branch, Office of Policy, Planning,  and  Evaluation,
   (PM-223), 401 M Street, S.W., Washington, D.C.  20460,  (202) 382-2734,
   (FTS) 382-2734.

*Herbert LaCayo, Jr.
   Statistical Policy Branch, Office of Policy, Planning,  and  Evaluation,
   (PM-223), 401 M Street, S.W., Washington, D.C.  20460,  (202) 475-9659,
   (FTS) 475-9659.

*Richard A. Levy
   Office  of Pesticide Programs, Office of  Pesticides  and Toxic  Substances,
   (TS-769C), 401 M  Street,  S.W., Washington,  D.C.  20460,  (202)  382-7398,
   (FTS) 382-7398.

Lloyd Lininqer
   Statistical Policy Branch, Office of Standards and  Regulations,  Office  of
   Policy,  Planning, and  Evaluation,  (PM-223),  401 M  Street, S.W.,  Washington,
   D.C.  20460,  (202)  382-2680,  (FTS) 382-2680.

*Bertram Litt
   Biostatistics Group, Hazard  Evaluation  Division, Office of  Pesticide
   Programs,  (TS-769C), 401  M Street,  S.W.,  Washington,  D.C. 20460 (retired).

 •Participants
                                       11

-------
Arthur N. Lubin
   Region V, 536 S. Clark Street, Chicago, IL 60605,  (312)  886-6226,
   (FTS) 886-6226.

*Allan H. Marcus
   Office of Air Quality Planning and Standards, Office  of  Air  and  Radiation,
   (MD-12), Research Triangle Park, NC 21711, (919) 541-5217,  (FTS) 629-5217.

Elizabeth H. Marqosches
   Office of Toxic Substances, (TS-798), 401 M  Street,  S.W.,  Washington,  D.C.
   20460,  (202) 382-3511, (FTS) 382-3511.

Thomas Matthew
   Department of Mathematics, University of Maryland,  Baltimore County,
   Catonsville, MD 21228, (301) 455-2347.

Peter Matthews
   Department of Mathematics, University of Maryland,  Baltimore County,
   Catonsville, MD 21228, (301) 455-2347.

Tom McCurdy
   Ambient  Standards Branch,  (MD-12), Research  Triangle Park,  HC 27711,
   (919) 541-5665, (FTS) 629-5665.

*Bruce Mitchell
   Manager, Mutual Information Center, Research Triangle Park,  NC 27709.

Deirdre  L.  Murphy
   Department  of  Health  and  Mental  Hygiene, Office of Environmental Programs,
   201 W.  Preston  St., P.O.  Box  13387, Baltimore,  MD  21201, (301) 225-6293.

Cornelius  J. Nelson
   Office  of Pesticide Programs,  (TS-766),  401  M Street, S.W.,  Washington,
   D.C.  20460,  (202) 475-9565,  (FTS)  475-9565.

William Nelson
   Environmental  Monitoring  Systems Laboratory, Office of Research and
   Development,  (MD-75), Research Triangle  Park,  NC 27711,  (919) 541-2301,
    (FTS) 629-2301.

*Vaun Mewill
   Assistant Administrator  for  Research  and Development (RD-672), 401 M
   Street,  S.W.,  Washington,  D.C. 20460,  (202)  382-7676, (FTS)  382-7676.

*Barry  Nussbaum
   Field Operations  and  Support  Division,  Office of Mobile Sources, Office of
   Air  and Radiation,  (EH-397F),  401 M  Street,  S.W.,  Washington, D.C. 20460,
    (202) 475-9659,  (FTS)  475-9559.

*Havne  R.  Ott
   Office  of Acid Deposition, Environmental Monitoring, and Quality Assurance,
   Office  of Research  and Development,  (RD-680), 401 M Street, S.W.,
   Washington,  D.C.  20460  (202)  382-5793,  (FTS) 382-5792.

"Participants
                                       12

-------
*David A. Parrish
   Region VI, Environmental Services Division, 1201 Elm  Street, Dallas, TX
   75270, (214) 767-9093, (FTS) 729-9093.

*Lorenz R. Rhomherq
   Office of Toxic Substances, (TS-798) , 401 M Street, S.V., Washington, D.C.
   20460, (202) 382-3895, (FTS) 382-3895.

*Vfilson B. Rigqan
   Health Effects Research Laboratory, Office of Research  and  Development,
   (MD-55),  Research Triangle Park, HC 27711,  (919) 541-7540,  (FTS)  629-7540.

Alan C. Rogers
   Statistical Policy Branch, Office of  Policy, Planning,  and  Evaluation,
   (PM-223), 401 M Strset, S.W., Washington, D.C.  20460,  (202)  382-2680,
   (FTS)  332-2680.

Belinda Ronca-Battista
   Office or Radiation Programs, Office  of  Air and Radiation,  UHR-460) ,  401  M
   Street, S.V.. Washington, D.C.  20460,  (202) 475-9605,  (FTS)  475-9605.

*H. ..Phillip  Ross
   Statistical Policy Branch, Office of  Policy, Planning,  and  Evaluation,
   (P:>223) , 401 M Street, S.«., Washington, D.C.  20460,  (202)  332-2680,
   (FTS)  382-2680.

John Ruggero
   Environmental  Ser"icis  Division, Region  III,  (3ES12), 841 Chestnut Street,
   Philadelphia,  PA  19107,  (FTS> 597-9857.

*Jcel  Schwartz
   Office of Policy  Analysis,  Office  of  Policy,  Planning,  and Evaluation,
   (PM-221) , 401  M Street,  S.U., Washington,  D.C.  20460, (202) 382-2732,
   (FTS)  382-23T82.

Bir.al  Sinha
   Department  of  Mathematics.  University of Maryland, Baltimore County,
   Catonsville,  MD 21228,  (301)  455-2347.

Roy  L.  S?ith
   Environmental  Services Division,  Region III,  (3ES12)  , 841 Chestnut Street,
   Philadelphia,  PA  19107,  (FTS)  597-9857.

      ia7. P.  Smith
    Office of Policy, Planning,  and Evaluation, (PM-223), 401 M Street, S.'J.
    Washington,  D.C. 20450,  (202)  382-2697, (FTS)  332-2697.

 Andrew G. Stead
    Office of Research and Development, (MD-55), Research Triangle Park, HC
    27711. (919) 541-3143, 
-------
Cynthia R. Stroup
   Office of Toxic Substances, (TS-798) , 401 M  Street,  S.V.,  Washington   D  C.
   20460, (202) 382-3891, (FTS) 382-3891.

Clayton L. Stunkard
   Statistical Policy Branch, Office of Policy, Planning,  and Evaluation,
   (PM-223), 401 M Street, S.W.,  Washington, D.C.  20460,  (202)  475-9659,
   (FTS) 475-9659.

*David J. Svendsgaard
   Biometry Division, Office of Research and Development,  (MD-55) ,  Research
   Triangle Park, NC 27111,   (919) 541-2468,  (FTS)  629-2468.

*Paul Switzer
   Professor, Statistics Department, Stanford,  University,  Stanford, CA  94305,

*Donald L. Thcrssen
   President, SIMS, 97 Parish Road, South, Hew  Canaan,  CT  06840.

Harit Trivedi
   Management Sciences and Systems Planning Division,  Pennsylvania  Departrent
   cf Environmental Resources, 3600 Vartan Way, Harrisburg, PA 17120,  (717)
   657-4634.

*Janes
   Professor, Department of Biostatistics , Harvard  School  of  Public  Health,
   677 Huntington Ave., Boston, HA 02115.

John Warren
   Office of Standards and Regulations, Office of Policy,  Planning,  and
   Evaluation, (PM-223), 401 M Street, S.W., Washington, D.C.  20460,
   (202) 382-2633, (FTS) 382-2683.

Dorothy Wellington
   Statistical Policy Branch, Office of Policy, Planning,  and E"aluation,
   (PM-223), 401 11 Street, S.V., Washington, D.C. 20460,  (202)  475-8204,
   (FTS) 475-8204.

*Alice Whittemore
   Professor, Department of Family, Community, and  Preventive Medicine,
   Stanford University, HRP Building, Room 110, Stanford,  CA  94305.

*Karcia E. Williams
   Director, Office of Solid Waste, Office of Solid Waste  and E.-.ergency
   Response, (WH-562) , 401 M Street, S.W., Washington, D.C. 20460,
   (202) 382-4627, (FTS) 382-4627.

*Hax Wcodburv
   Professor, Center for Demographic Studies, Duke  University,  2117  Canpus
   Drive, Durham, NC 27706.

"Participants
                                      14

-------
•James V.  Zidsk
   Professor,  Statistics Department,  University of British Columbia, Ponderosa
   Annex C,  ROOIP 101-D,  2021 West mall, Vancouver, B.C., Canada V6T1W5.

Robert Zisa
   Office of Compliance  Monitoring,  office of Pesticides and Toxic
   Substances, (EN-342), 401 M Street,  S.W., Washington, D.C. 20460,
   (202) 382-7835,  (FTS) 382-7835.
 "Participants
                                       15

-------
ABSTRACTS

-------
SESSION:  Criteria for Statistical Evaluation of Data
TITLE:    Realistic Criteria for Statistical Evaluation
AUTHOR:   Bertram D. Litt, Leader (Retired), Biostatistics Group, Office of
          Pesticide Programs, Hazard Evaluation Division
                                  Abstract

     1.   What are realistic criteria?

     2.   Who benefits from these criteria and how?  Conversely, what are the
      *
          implications of  the. lack .of suitable criteria?
     3.    A team approach to the design of realistic criteria.
          a.   The administrative decisionmaker will:
              1.  Identify the issue that requires criteria.
              2.  Appoint an interdisciplinary panel,  consisting of subject
                  matter scientists, a statistician, and a manager.
          b.   The panel  will:
              1.  Identify the critical marker and its competitors.
              2.  Set  a  working goal of operational criteria.
              3.  Determine the level of protection offered by the operational
                  criteria.
              4.  Estimate cost and feasibility of meeting the operational
                  criteria.
              5.  Provide recommendations for  criteria,  together with esti-
                  mated  costs  and benefits to  the  decisionmaker.
          c.   The administrative decisionmaker will:
              1.  Accept the recommended criteria  or recycle the process  with
                  a  recommendation for more (or less)  stringent criteria.

-------
SESSION:  Criteria for Statistical Evaluation of Data
TITLE:    Criteria for Tolerance Limits
AUTHOR:   Richard A. Levy, Biostatistics Group, Office of Pesticide Programs,
          Office of Pesticides and Toxic Substances
                                  Abstract

     The criteria for statistical tolerance limits will be illustrated by the
use of two examples taken from the author's application of these techniques to
environmental issues.  In the first example, "Aldicarb:  Calculation of Upper
10, 5, 1 Percentile of Residue Data for Potatoes, Citrus, and Drinking Water,"
nonparametric statistical tolerance interval methods are applied to the
problem, and an SAS program to calculate these intervals is provided.  For the
second example, "Guidance on Sampling for Verification of Cleanup of PCB Fire
Incidents," a parametric statistical tolerance interval technique is applied
to the problem, and an approximate calculation method is derived.
                                     17

-------
SESSION:  Criteria for Statistical Evaluation of Data
TITLE:    Criteria for Combining Data Sets
AUTHOR:   Herbert LaCayo, statistical Policy Branch, Office of Policy,
          Planning and Evaluation
                                  Abstract

     This talk considers the question:  "When should data sets not be combined
in significance testing?"  Several examples are given to illustrate some of
the problems that may arise if data sets are joined indiscriminately.  These
examples suggest rules that can be used to avoid such problems.
                                     18

-------
SESSION:  Criteria for Statistical Evaluation of Data
TITLE:    Criteria for Selecting a Statistical Model
AUTHOR:   R. Clifton Bailey, Analysis and Evaluation Division, Office of Water
          Regulations and Standards, Office of Water
                                  Abstract

     Statistical models are the basis for communication in data analysis.
Statistical models are somewhat like an artfully designed container that must
be correctly sized to hold the data without extreme distortion.  The mathe-
matical structure of a statistical model provides a powerful aid to amplify
our thoughts and ask fundamental questions, plan new studies, and evaluate the
outcomes.  Although statisticians have developed a plethora of evaluation aids
to the modeling process — goodness of fit, influential observations, robust
statistics, residual plots, outlier tests, etc. — there are other considera-
tions to statistical modeling.  This presentation will examine the more
familiar ideas of statistical modeling and appeal to users of statistical
models (including statisticians) to become familiar with important considera-
tions of model development such as symmetry, canonical forms, invariance,
dimensional analysis, identifiability, characterization theorems, limiting
forms, forbidden outcomes, and numerical methods for estimation as tools to
improve data collection and analysis.  Examples from toxicity testing,
survival models, ecological sampling, and environmental surveys illustrate key
features and emphasize the point that no meaningful data collection can take
place unless there is at least some rudimentary theory.
                                     19

-------
SESSION:  Criteria for Health and Exposure Assessments

TITLE:    Environmental Health Models

AUTHOR:   Allan H. Marcus, Visiting Scientist, Office of Air and Radiation,
          Office of Air Quality Planning and Standards
                                  Abstract


     Dose-response or dose-effects curves are often derived from cross-

sectional observational studies by multiple linear regression using ordinary

least squares.  More sophisticated analyses may be genuinely necessary,

including time series regression models and intrinsically nonlinear regression

models.  We present three case studies:


     1.   Daily mortality in London winters 1958-72, attributed to high
          concentrations of particulate matter, shows a complicated time-lag
          structure;

     2.   Hanlinaar. relationships between blood lead and lead intake from
          various sources can be derived using nonlinear lead absorption and
          blood plasma-red cell partition models; and
                                       _
     3.    Sensitivity of (ferythrocyte protoporphyrinl (an indicator of lead
          toxicity)  to iron nutritional status as well as lead exposure can be
          derived from nonlinear models of heme biosynthesis.  SAS implemen-
          tation will be discussed.

-------
SESSION:  Criteria for Health and Exposure Assessments
TITLE:    A Dose-Response Model for Teratology Data
AUTHOR:   Dorothy Wellington, Office of Policy, Planning and Evaluation,
          Chemical and Statistical Policy Division


                                  Abstract

     While a large body of statistical methodology has been developed for the
assessment of carcinogenic risks, there is relatively little for noncarcino-
genic risks, in particular, the risks of teratologic and other adverse
developmental effects resulting from parental exposure to toxic materials.
Risk models for carcinogenesis are not directly applicable to the laboratory
data from teratology studies because of the nonindependence of observations
within the same litter.
     Among the few approaches that have been suggested for low-dose modeling
of teratologic data,  a two-stage model proposed by Rai and Van Ryzin, which
combines the probabilities of litter effect and individual pup effect, appears
to be the most promising.  The applicability of this model has been investi-
gated with several data sets from the National Toxicology Program covering six
toxicants, two species, and both sexes.  It has been found to represent
reasonably the dose-response relationship between exposure and several
developmental toxicity end-points.
                                     21

-------
SESSION:  Criteria for Health and Exposure Assessments
TITLE:    Empirical Bayes Rates for Rare Events in Small Areas
AUTHORS:  Max A. Woodbury, Professor, Center for Demographic Studies, Duke
          University
                                  Abstract

     In assessing the risks of specific disease outcomes around specific
pollution sources, such as toxic waste dumps, one must deal with the problems
that (1) frequently the disease outcomes are rare and  (2) the exposed popula-
tion is small.  A consequence of these two conditions  is that one must
frequently make decisions (e.g., prioritize the clean-up of sites) based on
rates that are very unstable.  In this paper, we present a methodology that
provides estimates of rates,  that will improve, in aggregate, the decisions to
be made from the rates for small populations.  Specifically, we propose to use
a negative binomial regression strategy to model the small area rates.  From
these modeled rates we can calculate "posterior" rate  estimates, which are
weighted combinations of the observed rates and the rates predicted from the
regression.  The weights are selected by examining the amount of super-Poisson
variability in the distribution of rates over all small areas.
     If this super-Poisson variation is large, it suggests that the weight
assigned to the observed rates should be large.  If it is small, then more
weight should be assigned to the regression estimate — which is produced from
the total information in the sample.  Thus, the composite rate estimate has
the properties of being the best compromise estimates  in that the bias of the
regression rate estimate is reduced by combining it with the observed rate,
and the instability of the observed rate is reduced by combining it with the
regression estimates.
                                     22

-------
SESSION:  Criteria for Health and Exposure Assessments
TITLE:    The Use of Statistics in Setting the Gasoline Lead  Standard
AUTHOR:   Joel 0. Schwartz, Economic and Regulatory Analysis  Division, Office
          of Policy, Planning and Evaluation


                                  Abstract

     Carrying out a risk assessment requires estimating a dose-response
relationship and modeling the changes that a regulation  will cause.  Once a
relationship between exposure and an outcome has been found in data, addition-
al analyses are often helpful in elucidating the causal nature of the rela-
tionship and the shape of the relationship.  In addition, for many toxic.
contaminants, a distribution of body burdens exists in the population due to
differences in exposure and uptake from multiple sources.  In modeling the
results of changing a single source, it is important to examine the impacts on
the shape, as well as the mean of the population distribution.
     In the Regulatory Impact Analysis for reducing lead in gasoline, exten-
sive statistical tests were used to deal with both issues.  We found convinc-
ing statistical evidence of a causal relationship between gasoline lead and
blood lead to buttress the experimental findings.  By formulating specific
alternative hypotheses that could possibly confound the results and testing
them with our data,  we were able to strengthen our confidence about the
results.  We also found evidence that the shape of the blood lead distribution
changed as gasoline  lead levels changed, and that failure to account for this
would have lead to substantial overestimates of the benefits.

-------
SESSION:  Data Presentation Devices
TITLE:    Producing Visuals
AUTHOR:   Bruce Mitchell, Manager, Mutual Information Center, Office of Air
          Quality Planning and Standards
                                  Abstract

     This presentation will focus on the production of computer-generated
slides and transparencies.  The review of general frame composition includes
the use of color, graphics, and information density.  This discussion will
also include some possible answers to these commonly asked questions:
     •  Should I make slides or transparencies?
     •  What software should I use?
     •  Who will physically make my slides or transparencies?
     •  How long will it take?
     •  What quality can I expect?
     •  Where can I get help?
Acknowledgments:  The author would like to thank Warren Freas (EPA OAQPS) and
Tim Gunter {University of North Carolina)  for their generous expenditure of
time and talent.
                                     24

-------
SESSION:  Poster Session
TITLE:    How Good Are Analytical Results From Your Water Laboratory?
AUTHOR:   Paul V. Britton, Quality Assurance Branch, Office of Research and
          Development, Environmental Monitoring Systems Laboratory


                                  Abstract

     EPA has conducted three major series of Water Laboratory Performance
Evaluation Studies over the last ten years:  one for general wastewater
laboratories, one for drinking water laboratories, and one for major National
Pollutant Discharge Elimination System (NPDES) permittees.  Experience from
these studies has taught us how to characterize competent analytical perform-
ance and how often to expect the average laboratory in our studies to produce
an analytical result that suggests competent performance.
     Analytical results from a competent laboratory may be quite different
from the known true sample concentration, even when that laboratory is
operating properly.  And, the best of laboratories may periodically develop
analytical or reporting problems.
                                     25

-------
SESSION:  Poster Session
TITLE:    Evaluation and Presentation of Water Quality Data  for the Missouri
          River Near Omaha, Nebraska
AUTHORS:  Norman H. Crisp, Environmental Services Division,  Region VII
          Joseph E. Joslin, Environmental Services Division, Region VII
          Thomas T. Holloway, Environmental Services Division, Region VII
                                  Abstract

     River water quality is subject to numerous factors including:  the
physical processes of solution and sedimentation; chemical processes of
oxidation and reduction; temporal variation due to changes in flow; and impact
from tributaries and point source discharges.  During March 1984, a study was
conducted on the Missouri River near Omaha, Nebraska, to determine river water
quality and to evaluate the effect of the Omaha metropolitan area on the
river.  The results of the study indicated that river quality was generally
good.
     When the data were subjected to analysis of variance, the two principal
factors found to affect river water quality were flow and point source
discharges.  During periods of elevated flow, most pollutant concentrations
increased.  These increases were highly correlated with suspended solids
concentration.  Waste sources increased the concentrations of phosphorus and
ammonia and caused the development of a distinct and measurable plume.  This
presentation illustrates how the data evaluation results were presented.
                                     26

-------
SESSION:  Poster Session
TITLE:    Small Sample Properties of Estimators of the Odds Ratio Under
          Constant Multiple Hatching
AUTHOR:   Ayenew Ejigou, NRC Fellow, Health Effects Research Laboratory
                                  Abstract

     The bias in estimators of the odds ratio is investigated for a range of
sample sizes, matched controls, and odds ratios.  Comparison with the likeli-
hood estimator shows that when sample sizes are small, other estimators will
have considerably lower mean squared error.  The extent of the bias strongly
depends on the odds ratio to be estimated, the number of matched controls, and
the sample size; a new estimator is also recommended.
                                     26a

-------
SESSION:  Poster Session
TITLE:    Applications of a New NOEL/RfD Statistical Procedures  for  Small
          Samples with Oichotomous  (Incidence) Response Data
AUTHORS:  Linda Erdreich, Office of Research and Development, Environmental
          Criteria and Assessment Office
          Kenneth Brown, Chapel Hill, NC
          Lisa LaVange, Research Triangle Institute
                                  Abstract

     In noncancer risk assessment, the no-observed-effect level  (NOEL) is the
highest experimental dose level at which one does not reject the hypothesis
that the expected response rate is the same as in the control group.  The NOEL
is then scaled downward by uncertainty factors to obtain the reference dose
(RfD).
     The variability of the NOEL and the false negative rate need to be esti-
mated.  The expected value of the NOEL/RfD, its variability, and the relation-
ship between Type I errors (false positive rates) and Type II errors (false
negative rates) are all interrelated and depend on sample sizes, the expected
(but unknown) response rates, dose values, and the statistical methodology
employed.  A statistical method has been developed that treats these inter-
relationships.  This poster describes and illustrates how this procedure can
contribute toward evaluating and regulating noncarcinogenic risks by measuring
some uncertainties in the process due to sampling variability.
     The new NOEL procedure developed can be applied to:  (1) estimation of
the expected value of the NOEL and its variability; (2) calculation of the
power to detect specified alternatives; (3) evaluation of the effects of
sample size and dose placements; and (4)  setting the false negative rates
(Type II errors)  against a specified alternative, instead of controlling the
false positive rates (Type I errors).  The statistical procedure tests the
pair-wise equality of treatment and control response rates,  beginning with the
                                     27

-------
highest dose.  It conditions on the total number of responses and the rejec-
tion of any preceding comparisons.
     Proceeding in this way and counting all extreme permutations gives an
exact sequential test ("exact" in the same sense as the well-known Fisher's
exact test) and eliminates the need to introduce approximations from large
sample theory (which may be very inaccurate in small samples) or multiple
comparison procedures.
                                     28

-------
SESSION:  Poster Session
TITLE:    Geostatistical Analysis
AUTHOR:   George T. Flatman, Mathematical Statistician, Exposure Assessment
          Research Division
                                  Abstract

     This poster session display will use real-world examples to present the
two steps in spatial analysis.  The first step is semivariogram modeling.  The
linear and spherical models of semivariograms will be illustrated by the
Dallas Lead Study, and the need for directional semivariograms to correct for
anisotropy will be illustrated by the Missouri Dioxin Study.
     Second, the potential outputs of spatial analysis will be displayed as
isopleth and iso-area contour maps and contour maps of estimates of pollution
levels or estimates of probability of pollution levels.  This technique of
spatial statistics has promise for exposure assessment, monitoring, and
remediation.
                                     29

-------
SESSION:  Poster Session
TITLE:    Expert Systems to Assist in Decisions Concerning Land Disposal of
          Hazardous Vastes
AUTHOR:   Daniel G. Greathouse, Land Pollution Control Division, Hazardous
          Waste Engineering Research Laboratory
                                  Abstract

     Review of permits for land disposal of hazardous wastes requires numerous
decisions concerning technical and policy issues.  Some require interpretation
and application of information in research reports, others involve interpreta-
tion and evaluation of specialized test data, and others involve assessment of
compliance with latest regulatory policies.  Specialized knowledge concerning
a number of technical areas and a broad base of environmental regulatory
experience are necessary to perform these reviews adequately.  This need for
current knowledge and background in addition to the concern that reviews be
consistent (i.e., permits are judged the same by all reviewers) prompted our
interest in expert systems.
     Expert systems are computer programs (software) designed to provide
advice concerning specialized, narrow areas.  The programs are designed to
emulate the advice of subject specialists by incorporating the rules of
criteria that they use in terms of "IF/THEN" statements.  For example, IF the
carbon black content of an HDPE liner is less than 2 percent, THEN the liner
is inappropriate.  Two characteristics of these programs differentiate them
from traditional programs.  They are essentially large pattern matching
routines that seek a solution (advice) that corresponds to the pattern of
input data.  The program logic is separated from the rules (IF/THEN state-
ments) to facilitate ease of rule modification or refinement.  As with
traditional programs, these systems can perform calculations but these
calculations are not the primary features of these systems.
                                     30

-------
     The system being developed to assist in assessing the chemical resistance
of flexible membrane liners based on short-term immersion test data will be
demonstrated.  The rule base was developed by Arthur D. Little (ADD based on
the input of nine subject specialists, representing a test laboratory, liner
manufacturers and distributors, and resin suppliers.  The ADL system was
written in Arity Prolog linked to Lotus 1-2-3.  In order to improve flexibil-
ity and increase ease of usage, it has been rewritten in Arity Prolog by Jay
Clements of CSC assigned to the EPA Hazardous Waste Engineering Research
Laboratory.  The latter system will be demonstrated.
                                    31

-------
SESSION:  Poster Session
TITLE:    Statistical Methods for the Analysis of Toxic Interactions
AUTHOR:   Richard C. Hertzberg, Methods Evaluation and Development Staff,
          Environmental Criteria and Assessment Office


                                  Abstract

     The analysis of toxicity data on chemical mixtures is complicated by
several factors, not the least of which is the presence of two or more
chemicals.  ECAO recently developed a data base representing all available
published studies on toxic interactions.  The data base includes more than
2,400 observed interactions taken from more than 300 research articles.  A
coding scheme was developed to describe the nature of each interaction, with
"additivity" representing no interaction between two toxicants.  The statisti-
cal procedures that were reported were tabulated and evaluated for appro-
priateness.  The largest group included no statistics, followed closely by
studies that presented significance levels but did not identify the statisti-
cal method used.  The Student t test was the most commonly identified method
and was the most common method used to test dose additivity.
                                     32

-------
SESSION:  Poster Session
TITLE:    Data Quality of the FY 1986 Regional Ambient Fish Tissue Monitoring
          Program
AUTHORS:  Thomas T. Holloway, Environmental Services Division, Region VII
          Bruce Littell, Environmental Services Division, Region VII


                                  Abstract

     Region VII, in conjunction with its State agencies, is active in the
collection and analysis of fish tissue samples for toxic substances.  This
presentation shows the procedures and results for an evaluation of data
quality for the FY 1986 activity.  Topics to be presented include:  data
completeness, sample representativeness, data comparability (including the
effects of fish size and sampling location on measured concentrations),
detection limits, overall (sampling and analytical) precision, and bias.  The
evaluation can serve as a model for overall assessment of data quality for
other environmental monitoring activities.
                                     33

-------
SESSION:  Poster Session
TITLE:    Rubberband Regression
AUTHOR:   Robert W. Jernigan, American University and Statistical Policy
          Branch, Office of Policy, Planning and Evaluation
                                  Abstract

     The concept of least squares is sometimes perplexing to people who are
new to statistical techniques.  Why squares, rather than absolute values, or
some other power?  A common answer is:  because of ease of derivation — not
very intuitive.  I will present a simple, intuitive, noncalculus-based device
for demonstrating least squares regression.  The ideas are based on an equi-
librium of rubberband forces and torques on a rod, representing the least
squares regression line.  It is shown, using only simple physical principles,
that the standard equations for the slope and intercept can be derived without
the use of calculus.  Extensions to least absolute value regressions, weighted
regressions,  and principal components are easily obtained.
                                     34

-------
SESSION:  Poster Session
TITLE:    Niagara River Study
AUTHOR:   Henry D. Kahn, Office of Water Regulations and Standards


                                  Abstract

     This paper considers some statistical issues in the analysis of a set of
monitoring data collected in the Niagara River during a recent 2-year period.
The monitoring data were collected as part of a joint Canadian/United States
program to assess the extent of pollution problems in the river.  Data were
collected bimonthly by the Canadians at monitoring stations located at the
head and mouth of the river.  Samples of water and suspended sediment were
analyzed for 81 substances (68 synthetic organic compounds and 13 metals).
The sampling program was initiated because of the concern that discharges from
sources along the river may affect water quality in both the Niagara River and
Lake Ontario.  Statistical analyses of the data to determine whether signi-
ficant changes in pollutant levels have occurred will be discussed.
                                     35

-------
SESSION:  Poster Session
TITLE:    An Evaluation of Pesticide Residues in Fish Tissue
AUTHOR:   David A. Parrish, Environmental Services Division, Region VI


                                  Abstract

     Pesticide and heavy metal levels in fish tissues were extracted from
STORET and analyzed using SAS on the EPA's IBM mainframe at NCC.  Data
selected was limited to that collected by the Oklahoma State Health Depart-
ment.  This data set contained approximately 900 samples from about 90 sites.
An index was developed to estimate the potential risk of a mixture of pollut-
ants to human health.  This index was based on standard risk assessment
techniques used by EPA scientists.  This index was used to compare the risk
between consumption of different species and between water bodies.  Most of
the risk was associated with the occurrence of three compounds, PCBs, chlor-
dane, and total DDTs.  One major purpose of this study will be to demonstrate
the use of analytical procedures on existing data for the States in Region VI
at our annual water quality data workshop.
                                     36

-------
SESSION:  Poster Session
TITLE:    Using Pharnacokinetic Models to Improve Dosimetry for Noncontinuous
          Inhalation Exposures to Vapors
AUTHOR:   Lorenz Rhomberg, Exposure Evaluation Division, Office of Toxic
          Substances
                                  Abstract

     In estimating health risks from inhalation of toxic gases, exposures are
often quantified according to "Haber's Law," that is, the product of vapor
concentration and the time spent breathing that concentration are taken as the
exposure measure (e.g., ppm-hours).  A corollary that is used in cancer risk
assessment is that intermittent exposures are equivalent in risk to life-long
exposures at lower concentrations, so long as the lifetime average daily
exposure is equal.
     However, short but intense exposures may introduce complications and non-
linearities (such as changes in absorption and disposition of the compound, as
well as saturation of its metabolism) that can affect the equivalence of short
and long exposures that are equal in terms of ppm-hours.  Pharmacokinetic
models can be used to investigate differences in tissue-level exposures under
various exposure regimes.  Their usefulness is illustrated by applying a
physiologically based pharmacokinetic model for perchloroethylene inhalation
in humans to such questions as:
     •    Does body burden accumulate during repeated daily workplace
          exposure;
     •    Do 8-hour, time-weighted averages provide good characterization of
          varying workplace vapor concentrations; and
     •    Can occasional consumer exposures be adequately measured as a
          concentration-time product.
                                     37

-------
SESSION:  Poster Session
TITLE:    Modeling Rare Events in Small Populations
AUTHORS:  Wilson B. Riggan, Office of Research and Development, Health Effects
          Research Laboratory
          John P. Creason, Chief, Biostatistics Branch, Office of Research and
          Development, Health Effects Research Laboratory
                                  Abstract

     In health risk assessment,  environmental statisticians and epidemiolo-
gists face the problem of rare events in small areas and populations.  These
rates are highly variable due to the discreteness of the events.  We present a
system for generating more stable rates and patterns over space and time.  The
system uses a two-stage Empirical Bayes procedure to produce composite
(smooth)  rates, which result in more stable patterns.  The results are pre-
sented graphically and in tables.
                                    38

-------
SESSION:  Poster Session
TITLE:    Numerical Method of Estimation of the Elemental Concentration of
          Mitochondrian Cells in Biological Tissue Samples
AUTHORS:  Judy A. Stober, Toxicology and Microbiology Division, Office of
          Research and Development, Health Effects Research Laboratory
          Dennis Black, Computer Science Corporation
          Tammy Mills, Computer Science Corporation
                                  Abstract

     A project was recently undertaken to develop a short-term bioassay using
x-ray microanalysis to measure elemental concentrations in liver cells.
Shifts in elemental concentrations are known to be early indicators of the
effects on the liver caused by organic solvents.  A new way to study elemental
concentrations in cells is one which uses an electron microscope to analyze
small sections of frozen tissue.  Using a technique called "x-ray micro-
analysis," information on elements in a small area of a cell excited by a
focused beam of electrons is collected as "x-ray spectra."  These spectra are
a series of overlapping peaks, each representing a different element.
     One approach to determining the elemental concentrations of biological
samples is the continuum-normalization method.  Information needed to deter-
mine elemental concentration levels must be abstracted from overlapping peaks
and from the continuum, which includes a noisy background contributed by other
factors in the analysis system.  The sequential simplex is the basis of the
technique that has been adapted to deconvolute raw spectral data into individ-
ual peak values.  Regression analysis is then used to estimate the continuum
value.  As a result, the spectral peak data can then be separated from the
continuum and elemental concentration values can be obtained, which can be
interpreted as toxicological responses of cells to the insult of organic
chemicals.
                                    39

-------
SESSION:  Poster Session
TITLE:    GRAPE:  Graphic Representations of Activity Patterns and Exposure
AUTHORS:  Jacob Thomas, General Sciences Corporation, Laurel, Maryland
          Perng-Fei Vang, General Sciences Corporation, Laurel, Maryland
          Herb Hunt, General Sciences Corporation, Laurel, Maryland
          Wayne Ott, Office of Information Resources Management


                                  Abstract

     Two carbon monoxide (CO) exposure studies, one in Denver, Colorado, and
the other in Washington, DC, gathered a great deal of data on individual
activities and CO exposure in various microenvironments.  Answering the need
to examine these data in both tabular and graphical form, EPA has developed
the Graphic Representation of Activity Patterns and Exposure  (GRAPE) computer
system.  Operating on a personal computer, GRAPE allows the user to develop
search criteria against which the data base is searched.  These search
criteria can include any combination of respondent's age, sex, occupation,
date of study, day of week of study, and maximum 8-hour exposure level to CO.
Retrieved records can be viewed graphically on the IBM PC/XT computer.  GRAPE
also allows the user to view a histogram of all records meeting the search
criteria.  GRAPE is intended as a practical tool for examination of the
Denver-Washington data base, the relationship of activity to exposure, and the
relationship of demographic variables to exposure.  GRAPE also can be adapted
to study other data bases on other pollutants.
                                     40

-------
SESSION:  SIMS:  Statistical Methodology for Environmental Issues
TITLE:    SIMS Studies:  Development of Environmetrics
AUTHOR:   Donald L. Thomson, Jr., President, SIMS
                                  Abstract

     For more than 12 years, SIMS has been active in the development of
mathematical and statistical methodologies applicable to environmental issues,
These same methodologies have collectively become recognized as "environ-
metrics" much as one considers other disciplines such as biometrics (biology)
and econometrics (economics).  Advances in this fast-developing field will be
highlighted and some of the principal contributions from the professional
statistical community will be presented.  EPA has had a unique role in this
connection; EPA's contributions as well as some of the direct benefits to EPA
will be noted.
                                     41

-------
SESSION:  SIMS:  Statistical Methodology for Environmental Issues
TITLE:    Role of Statistics in Environmental Monitoring and Regulation
AUTHOR:   Paul Svitzer, Professor, Statistics Department, Stanford University


                                  Abstract

     Highlights of the research activity of the Stanford University/SIMS
program in environmental statistics will be reviewed.  The program made a
number of contributions to spatial statistical analysis with particular
application to environmental problems.  Spatial-temporal models were developed
for ambient air quality, as well as rainfall acidity, using seasonal com-
ponents, long-term trends, and spatial auto-correlation.  These were applied
to problems of spatial contouring, year-to-year variability estimation,
monitoring design, and relationships among pollutant species.  Estimates of
spatial variograms were developed for these applications.  Problems associated
with regulatory standards were investigated from a statistical point of view,
requiring application of new developments in the distribution theory of
extreme values.
                                     42

-------

-------
SESSION:  SIMS Methods Development and Applications
TITLE:    Indirect Exposure Assessment:  Implications for Regulatory Programs
          and Public Health Risks
AUTHOR:   Naihua Duan, Economics Division, The Rand Corporation
                                  Abstract

     With the recent advances in exposure assessment, it has become evident
that environmental exposure must be considered in conjunction with micro-
environments in which exposure occurs.  A human subject's exposure to carbon
monoxide is not only affected by the ambient concentration of carbon monoxide,
but is also affected by the amount of time spent in transit, the type of
transit, and the concentrations associated with specific types of transit.
Other types of human activities, such as indoor combustion, could also have
major impacts on the subject's total exposure.  Therefore, human activity
patterns and nonambient concentrations are important determinants for the
total exposure.
     In light of this research, it is important to consider the regulatory
implications of indirect exposure assessment.  The goal of reducing public
health risk can be achieved in a variety of ways.  (The reduction of ambient
concentrations, for which EPA is mandated to achieve, is not the only way, and
might not be the most effective way.)  Alternative regulatory vehicles, which
either reduce the nonambient concentrations or reduce the amount of time spent
in high-concentration microenvironments, could be more effective.  Most of
those vehicles might fall outside of EPA's jurisdiction, although EPA can
still pay an important role in providing the expertise to guide the government
agencies having the appropriate jurisdiction.
     Most of the existing research on indirect exposure assessment is con-
cerned with air pollution.  The concept can also be applied to alternative
media such as water or food.  The analogue of microenvironment for those media
would be various sources of intake,  such as tap water, bottled water, car-
bonated beverages,  etc.  It is again important to consider the amount of
intake from the various sources in order to assess the total exposure.
                                     43

-------

-------
SESSION:  SIMS: Statistical Methodology for Environmental  Issues
TITLE:    Epidemiology in Risk Assessment for Regulatory Policy
AUTHOR:   Alice S. Whittemore, Professor, Department of Family, Community, and
          Preventive Medicine, Stanford University


                                  Abstract

     The Twentieth century has seen the rapid evolution of many new fields
concerned with protecting public health.  Epidemiology and risk assessment
have several of the features common to these new fields, and important
differences.  Both are needed to make the difficult decisions required in
setting standards for levels of toxic agents in the workplace and environment.
They differ in their arms, orientation, and time scale.
     Epidemiology has been defined as "the study of the distribution of a
disease or a physiological condition in human populations and of the factors
that influence this distribution" (emphasis added).  By contrast, health risk
assessment denotes research and evaluation to characterize the probability of
physical harm to humans attributable to a particular agent or group of agents.
While the distribution of disease provides the focus for epidemiologic
research, concern for adverse effects of specific toxicants drives risk
assessment.  Moreover, while epidemiologic studies proceed at the glacier-
like pace needed to mobilize large staffs of support personnel and to monitor
large populations over long periods of time, risk assessment activities
acquire the urgency felt by regulators, who must make decisions (including
decisions to postpone decisions)  today.  Most important, while epidemiology is
a scientific field that draws upon medicine, demography, and statistics, risk
assessment is a hybrid of science and policy that draws not only upon fields
such as epidemiology, toxicology, chemistry, and engineering, but also upon
psychology, politics, economics,  law,  and social justice.
     These inherent differences in emphasis, timing, and nature complicate the
role played by epidemiology in risk assessment for regulatory policy.  In
1985, this role is still largely one of epidemiology's uncharted galaxies.  I
will review the role's history,  and the reasons why it will continue to play

                                     44

-------
an essential part in regulatory decisionmaking.  The role has placed epidemic-
logic findings and epidemiologists at the center of political controversies,
and I will discuss the positive and negative side effects of this new visibi-
lity.  Finally,  I will explore ways to prevent the negative side effects and
ways to increase the utility of epidemiologic data for regulatory risk
assessment.

-------
SESSION:  SIMS:  Statistical Methodology of Environmental  Issues
TITLE:    Investigating a Cancer Cluster in a Massachusetts Community
AUTHORS:  Roger Day, Harvard School of Public Health
          James H. Ware, Harvard School of Public Health
          Daniel Vartenberg, Harvard School of Public Health
          Marvin Zelen, Harvard School of Public Health


                                  Abstract

     Growing public awareness of cancer risk from hazardous waste sites and
other local environmental problems has led to numerous reports of excess
cancer in communities believed to be at risk.  The number  of such cases makes
it difficult for public health agencies to investigate each report
exhaustively.  Thus, methods are needed for investigating  possible cancer
clusters using relatively simple data-collection and analysis strategies.
This paper reports an effort to investigate a cancer cluster in Randolph, MA,
by interviewing only households that had reported a history of cancer through
a local community network.  These data were supplemented by information on
cancer incidence between 1982 and 1984 from the Massachusetts Tumor Registry,
and by data on cancer deaths between 1964 and 1984 obtained from the
Massachusetts Department of Public Health.
     The investigation used newly developed spatial clustering techniques to
confirm the existence of an unexpected number of cancer cases in a region
originally identified by the residents of the study neighborhood.  The cancer
cases occurring near the cluster center were investigated  further for unusual
patterns of spatial-temporal clustering and disease site distribution.  Using
a 20-year residence history constructed from the annual town census, we
determined that risk of cancer was not associated with duration of residence
in the community.   The community had experienced an excess number of lung
cancer cases relative to the expected number derived from  the Connecticut
Tumor Registry; the frequency of cancer at other disease sites was consistent
                                     46

-------
with expectation.  The paper discusses the relative utility of death and
incidence data,  and also comments on the utility of "restricted" investi-
gations relative to comprehensive studies of neighborhoods at risk.
                                     47

-------
SESSION:  SIMS:  Statistical Methodology for Environmental Issues
TITLE:    Geostatistics for the Environment:  Probability Tools for
          Decisionmaking
AUTHOR:   Andre G. Journel, Professor, Applied Earth Sciences Department,
          Stanford University
                                  Abstract

     Evaluation of the risk of making a wrong decision is of a paramount
importance in environmental studies.  The challenge of statistics is not much
to provide a better estimate of an unknown decision parameter, but to charac-
terize the distribution of that unknown parameter whatever the "best" estimate
retained.
     The indicator/probability kriging allows a nonparametric, nongaussian-
related modeling of conditional (posterior) distributions.  These distribu-
tions are conditional not only on the data size and configuration but also on
the data vlaues, thus accounting for the heteroscedasticity often found in
environmental data.
     Availability of such conditional distributions of unknowns, say pollutant
concentrations over a site, allows mapping of isopleth maps of probability of
exceedance, risks and of false positive and false negative, and the assessment
of the need for additional sampling.
     A case-study related to a Dallas lead smelter site is presented.

-------
SESSION:  SIMS:  Statistical Methodology for Environmental  Issues
TITLE:    Acid Deposition:  Models and Data Analysis
AUTHOR:   James V. Zidek, Head, Department of Statistics, University of
          British Columbia
                                  Abstract

     This talk will survey some recent results obtained by the group of
investigators in the Pacific northwest sponsored by the EPA through a coopera-
tive agreement with SIMS.  These results will concern the spatial and temporal
structure of the acidic precipitation field and the associated analysis of
data from the MAP3S monitoring network.
     An attempt to fit a model proposed by Eynon and Switzer will be des-
cribed.  This very plausible, additive model, which includes a scavenging term
to account for precipitation volume, fails to account for more than a small
percentage of the variation in observed pH levels.  So a more basic, general
linear space-time model was fitted to explore the structure of the deposition
series.  The results will be described.  In particular, it will be argued that
space-time interaction cannot be ignored.  The spatial correlation patterns
will be shown to be exceedingly complex and a novel, very general approach to
modeling these patterns, which are of fundamental importance in network
design, will be described.
     If the space-time process is assumed to be smooth (continuous, say, or
differentiable),  a very general, new approach to modeling obtains.  The
surprisingly strong implications of this assumption with respect to the
spatial covariance structure when the spatial process is assumed to be
stationary (in the wide sense)  will be described.  And the nonparametric
alternative to Kriging that derives from these assumptions will be described
and applied to the analysis of  the MAP3S data.
                                     49

-------

-------
SESSION:  Workshop
TITLE:    Environmental Statistics
AUTHOR:   Paul Switzer, Professor, Statistics Department, Stanford University


                                  Abstract

     1.  Relating Environmental Quality Standards to Monitoring.  Standards
may be articulated in terms of averages of pollutant concentrations that vary
continously in space and time.  In practically all situations, the monitoring
data do not have the same spatial or temporal "support."  For example,
monitoring is always spatially discrete and, for nongaseous pollutants, is
usually temporally discrete.  How does one relate the monitoring schedule to
the articulated standards?

     2.  Indices of Environmental Quality.  Because we measure multiple
pollutant concentrations in most monitoring programs, some form of reduction
is sometimes attempted.  Methods related to principal components are sometimes
used but without a specific goal.  One possible goal is to choose an index
that most clearly exhibits temporal or spatial trends.  This goal may be posed
in a formal way that allows the index to be chosen in an optimal fashion.

     3.  Model Uncertainty.  Calculations of potential environmental impacts,
pollutant source allocations, as well as straightfoward spatial interpolation
all depend on modeling assumptions.  Row should one incorporate the inevitable
infidelity of assumed models into expressions for uncertainty of the kind of
calculations described above?
                                     50

-------
SESSION:  Workshop
TITLE:    Power of Persuasion Using Your PC
AUTHOR:   Linda A. DeLuise, Resource Management and Evaluation Branch, Office
          of Pesticide Programs
                                  Abstract

     Even great ideas must be sold.  Statisticians often fail to remember
this.  They feel that their ideas will sell themselves and that the data they
are presenting will be as clear to others as it is to them.  Many computer
programs and tools are available to help enhance their presentation and help
both the statistician and nonstatistician understand, accept, and implement
the statistician's idea.
     This workshop will deal with this issue.  It will show you how to use
your PC to present your data more clearly.  We will discuss PCs and some of
the software packages available at EPA.  The main emphasis will be on the
LOTUS package and how this can be used to present data more effectively.  The
specifics will include how to set up a LOTUS file, input data, use the
regression feature, produce graphs from the data, conduct the present sen-
sitivity analysis, and write reports — in general, how to produce profes-
sional-looking presentations using your PC.  We will emphasize presentation,
not content.  At least one real-life problem will be presented and alternative
solutions will be analyzed.
                                     51

-------
SESSION:  workshop
TITLE:    Expert System Advisor for Survey Design
AUTHOR:   William P. Smith, Statistical Policy Branch, Office of Policy,
          Planning and Evaluation
                                  Abstract

     A simple expert system capable of acting as an "artificial statistician"
will be demonstrated.  Many areas of statistics can be snythesized into a set
of logical rules to form a knowledge base by which an expert system could be
created to advise nonstatisticians.  The role as advisor is stressed because
artificial intelligence is a long way from replacing a "competent"
statistician.
                                      52

-------
SESSION:  Workshop
TITLE:    Using Statgraphics to Analyze Data on the PC
AUTHOR:   David J. Svendsgaard, Biometry Division, Office of Research and
          Development, Health Effects Research Laboratory
                                  Abstract

     This statistical system is written in APL, which is a language that
permits the quick development of systems.  Consequently, STATGRAPHICS can be
modified to provide applications that are tailored to your needs.  This
workshop will teach you how to plot data and run various routines in
STATGRAPHICS.  Also, some of the capabilities of this system will be demon-
strated using EPA data and applications.
                                     53

-------
SESSION:  Communications Within EPA
CHAIR:    Marcia E. Williams, Director, Office of Solid Waste


                                  Abstract

     The panel will consist of three EPA senior managers and two EPA statis-
ticians.  The EPA managers will give a 30-minute presentation that focuses on
regulatory decisions which provide good and/or bad examples of activities in
which statistical design and analyses played an important role.  The topics
presented should provide insights into answering the following questions:

     1.   What is the role and use of data collection and statistical analysis
          in EPA decisionmaking?
     2.   Are the trends in the Agency toward more or less dependency on data?
          What are the reasons for the trends?
     3.   To what extent do you rely on existing data rather than generating
          new data?  Why?

     After the managers make their presentations, the statistician members
will use these questions as a basis for a panel discussion on ways to improve
the analytical support behind regulatory decisions.
                                    54

-------