Sept 08 Am Stat News

8/14/2019 Sept 08 Am Stat News

1/84


2/84


3/84

Executive Director

Ron Wasserstein: [email protected]

Associate Executive Director and Director of Operations

Stephen Porzio: [email protected]

Director of Programs

Martha Aliaga: [email protected] of Science Policy

Steve Pierson:[email protected]

Managing Editor

Megan Murphy: [email protected]

Production Coordinators/Graphic Designers

Melissa Muko: [email protected] Vigyz: [email protected]

Publications Coordinator

Val Snider: [email protected]

Advertising Manager

Claudine Donovan: [email protected]

Special Contributors

Martha Aliaga Keith Crank Rosanne DesmoneRebecca Nichols Rick Peterson

Gladys Reynolds Fritz Scheuren

September 2008 Issu

DEPARTMENTS

1 President's Invited Column

61 Professional Opportunities

o be a world leader in promoting statisticalpractice, applications, and research; publishingstatistical journals; improving statistical education;and advancing the statistics profession.

MISSION STATEMENTSupport excellence in statistical practice, research,

journals, and meetings. Work for the improvementof statistical education at all levels. Promote theproper application of statistics. Anticipate andmeet the needs of our members. Use our disciplineto enhance human welfare. Seek opportunities toadvance the statistics profession.

VISION STATEMENT

Amstat News welcomes news items and let-ters from readers on matters of interest to the asso-ciation and the profession. Address correspondence toManaging Editor,Amstat News, American Statistical

Association, 732 North Washington Street, Alexandria VA 22314-1943 USA, or email [email protected] must be received by the first day of the preced-ing month to ensure appearance in the next issue (forexample, June 1 for the July issue). Material can be sentas a Microsoft Word document, PDF, or within anemail. Articles will be edited for space. Accompanyingartwork will be accepted in graphics file formatsonly (.jpg, etc.), minimum 300 dpi. No material in

WordPerfect will be accepted. Amstat News(ISSN 0163-9617) is published

monthly by the American Statistical Association, 732North Washington Street, Alexandria VA 22314-1943USA. Periodicals postage paidat Alexandria, Virginia,

and additional mailing offices. POSTMASTER: Sendaddress changes toAmstat News, 732 North WashingtonStreet, Alexandria VA 22314-1943 USA. Send Canadianaddress changes to Station A, P.O. Box 54, Windsor ONN9A 6J5; [email protected]. Annual subscriptionsare $50 per year for nonmembers. Amstat Newsis themember publication of the ASA. For annual membershiprates, see www.amstat.org/join or contact ASA MemberServices at (888) 231-3473.

American Statistical Association732 North Washington Street

Alexandria, VA 223141943 USA(703) 6841221 FAX: (703) 684-2036

ASA GENERAL: [email protected]

ADDRESS CHANGES: [email protected]

AMSTAT EDITORIAL: [email protected]

ADVERTISING: [email protected]

WEB SITE: www.amstat.org

Printed in USA 2008American Statistical Association

STATISTICIANS IN HIST

DAY IN THE LIFE

CAREER GUIDE

SPECIAL MATERIAL

4 Kimiko Bowman Never Say Never

7 Irene HessSampling Is in the Details

11 Jack Chao-sheng Lee

13 Paul Dixon MintonBuilding a Department

15 Mollie OrshanskyAuthor of the Poverty Thresholds

Read about the careers andlives of statisticians whose

names are recognizableas both published

authors and scientists

Statisticians and other expertsoffer career advice for

those looking to sharpentheir quantitative careers

21 Life as a Stochastic ModelerAlan Gelfand, Duke University

25 On Becoming a TeacherDaren Starnes, The Lawrenceville School

30 What Does a Teaching Associate Professor Do?Pam Arroway, North Carolina State University

34 Snakes and Ladders: Building Career in StatisticsDavid L. Banks, Duke University

43 Are You Media Shyor Media Rosanne Desmone,ASA Public Relations Specialist

46 Making a Name:Up-and-Coming Statisticians oVerge of Great Things

28 Where Are Your Colleagues?

38 Top Five Favorites

54 Math Is Music; Statistics Is Literature

(Or, Why Are There No Six-Year-Old Novelists?)

Cover design by Megan Murphy

HISTORY

CAREER


4/84

2 AMSTAT NEWS SEPTEMBER 2008

PRESIDENTSINVITEDCOLUMN

Statistics in Monitoring the Nations HealtNathaniel Schenker, Senior Scientis

Research and Methodology, NCHS/

Ive spent the past nine years at the

National Center for HealthStatistics (NCHS), having previ-

ously taught biostatistics at UCLA for11 years and worked at the U.S.Census Bureau for three years beforethat. NCHS, which is part of theCenters for Disease Control andPrevention, is the nations principalhealth statistics agency. Its motto isMonitoring the Nations Health,and its mission is to provide statisticalinformation that will guide actionsand policies to improve the health of

the American people.What Is Done at NCHS?

When I used to take my infant sonfor a medical check-up, the pedia-trician would plot his height and

weight on growth charts. Little didI know the charts were produced byNCHS and were based primarily ondata from NCHS National Healthand Nutrition Examination Survey(NHANES). Similarly, when carsbegan to use unleaded gasoline, Ididnt realize the formulation of poli-cy to move away from leaded fuel wasbased on analyses of data on blood-lead levels from the NHANES. (Formore on this policy, see the PresidentsInvited Column in the May issue of

Amstat News).Indeed, NCHS collects and ana-

lyzes data and produces informationfor the public on a wide range of

health topics, including vital events

and demography, health conditionsand behaviors, access to and qual-ity of health care, and injuries anddisabilities. Sources of NCHS datainclude birth and death certificates,as well as various sample surveys thatcollect information via review ofmedical records, personal interviews,physical and dental examinations,and laboratory tests.

S Stands or StatisticsGiven that the s in NCHS stands for

statistics, statisticians obviously havea major role in the work of the agency. As do many agencies in the federalstatistical system, NCHS has separatedivisions that focus on specific data col-lection systems and topics. Statisticians

within these divisions work on thedesign and implementation of thedata collections, analyses of the data,and preparation of the data for releaseto the public. Other statisticians (suchas I) work in offices within the agencythat focus more on cross-cutting analy-ses and research and on the develop-ment of statistical methodology, whilealso providing consultation to the datadivisions. Examples of methodologi-cal research areas at NCHS includeoptimal strategies for sample design,questionnaire design, modeling andestimation, data mining, developingindexes of health, confidentiality, andhandling missing data.

An NCHS statistician often

as part of a team, together withstatisticians as well as specin areas such as computer scdemography, economics, epiology, medicine, public healthsociology. here are ample opnities for individual research pras well. Moreover, collaborationquently occur with researchers atfederal and local government agein industry, and in academia.

Mr. Academic Goes (Bac

WashingtonWhen I decided to leave my teacademic job in UCLA's Deparof Biostatistics in sunny Califorreturn to the East Coast and wothe government, I was often Are you crazy? Why would ythat? Well, there were a numreasons, some of them personalas my growing up in Washingtothe East Coast having a lot to

As a statistician, however, I wased by the prospect of working source of health data that are imtant and of interest to many othe agency. Indeed, at UCLA, Iran across NCHS data and reseaanalyzing them. Also, NCHS slike a relatively small, friendly agso I didnt feel as though Id g

working there. And, it seemed toa culture of conducting researI thought Id be able to contin

NathanielSchenker

This month, I have invited Nathaniel Schenker of the National Center for Health Statistics to discuss how thatagency provides information about the nations health. Since before 1960, NCHS has monitored health statistics,initially under the stewardship of Forrest Linder. It is a wonderful resource for data and information.


5/84

SEPTEMBER 2008AMSTA

research career. (After coming to NCHS,I also became an adjunct professor in the

Joint Program in Survey Methodology atthe University of Maryland, so I didnt haveto abandon academia completely.) Finally,the Washington area has an exceptionallyactive statistical community. For example,it is home to the largest chapter of the ASA:the Washington Statistical Society.

Do I miss being a full-time professor at afine university? Yes. Am I glad that I changed

jobs? Absolutely. Besides the fact thatchanging jobs every once in a while helps to'keep ones blood flowing,' Ive enjoyed theatmosphere and work at NCHS. Im less ofa free agent in some ways now than I wasin academia (although I also dont have tohustle for grant support the way I did), andthere are some bureaucratic hassles involvedin working for the government; but every

job has its own administrative headaches,and any annoyances due to governmental

red tape have been far outweighed by theopportunities Ive had for interesting work.

A Small Sample o My Projectso illustrate some of the statistical workthat goes on at NCHS, Ill briefly outlinea couple of major projects in which Ivebeen involved.

In 1997, the Office of Managementand Budget revised the standards for clas-sifying federal data on race and ethnicity.

A key provision was that respondents tofederal data collections be given the optionof choosing more than one race group todescribe the person in question. Because theprevious (1977) standards called for only asingle race group to be reported, data col-lected under the 1977 standards are notcomparable with data collected under therevised standards. his can cause problemsin trend analysis, the calculation of a vitalratefor which the number of events inthe numerator and the population size inthe denominator can come from differentsources using different standardsand inother types of studies that combine dataclassified under the two standards.

he decennial census, a widely usedsource of denominators for vital rates, beganallowing multiple-race reporting in 2000. omake the 2000 census data (and intercensaland postcensal estimates) comparable

with data classified according to the 1977standards, NCHS, with assistance from theU.S. Census Bureau, produced bridgedcensus counts by county, age, sex, and

raceestimates of the counts that wouldhave been obtained had the prior standardsbeen in effect (www.cdc.gov/nchs/about/major/dvs/popbridge/popbridge.htm). hebridging process used models predictingresponses under the prior standards fromresponses under the current standards andcovariates. hese bridging models weredeveloped using data from NCHS National

Health Interview Survey (NHIS), which hasallowed multiple-race responses for manyyears, but which also asks multiple-racereporters for a single race group that bestdescribes the person in question.

I worked on many aspects of this project,but perhaps my biggest contributions wereto the formulation of the overall approachused to solve the problem, the developmentof the bridging models, and the derivationof methods for assessing uncertainty in anal-yses that use the bridged data.

A second project involved multiple

imputation of missing income data. heNHIS provides a rich source of data forstudying relationships between incomeand health and for monitoring healthand health care for persons at differentincome levels. However, the nonresponserates are high (roughly 30%) for two keyitems: total family income in the previouscalendar year and personal earnings fromemployment in the previous calendar year.o handle this missing-data problem andallow analysts of the data to assess theuncertainty due to missing data, mul-

tiple imputation of these items has beenperformed for several years (www.cdc.gov/nchs/about/major/nhis/2006imputedincome.htm) in collaboration with researchers atthe University of Michigan.

Several features of the data made theproject particularly challenging and inter-esting. First, the data are hierarchicalin nature, with one of the key variablesreported at the family level and the otherreported at the person level. Second, there

were cases in which the value of one vari-able (e.g., personal earnings) could berestricted by the value of another variable

(e.g., whether the person was employed),but the values of both variables were miss-ing simultaneously. hird, in some cases,family income and personal earnings need-ed to be imputed within bounds becausepartial information was available aboutthem (e.g., when a range was provided forfamily income, rather than an exact dollarvalue). Finally, several variables of various

types (continuous, categorical, countused as predictors, but they sometimsmall amounts of missingness as wel

I was involved in many aspects oproject (as with the bridging projectespecially in developing and evaluatimethods for imputation, writing tecdocumentation, and consulting withlysts using the multiply imputed dat

An Exciting, yet SometimesScary, Jobhe two projects just described have al general characteristics in commonthey required teamwork, both amonat NCHS and between NCHS andorganizations. Second, they requiretistical research. In fact, both projecto publications in JASA [98(464101(475)]. hird, the problemscomplicated, and they required massumptions and implementing apmations for their solution. Finally,

was great interest in the data, and pboth inside and outside the agencyclamoring for release of the data whi

work was going on.Projects with the second through f

characteristics create the need to puenvelope with regard to methodunder time constraints while maintthe integrity of the data. his result

job that is exciting, yet sometimes s

Challenges and Opportunit

Statistical work at NCHS, and i

government in general, is challefor administrative and technical reFor example, budgets are tight; cosincreasing for data collection, proceand dissemination; there is a desire opart of policymakers, researchers, aners for more information; and the ption of confidentiality is becomingdifficult. Oh, yes, and lets not forgfamous graying of the federal work f

with many employees nearing retirage.

Of course, with challenges

opportunities. Difficulties due to resources and conflicting desires andstraints with regard to data will needaddressed through the creative devment of efficient methods for collnew data and methods for getting theout of analyses of existing data. Angraying work force will result in opefor new generations of statisticians toout such work.


6/84


STATISTICIANS IN HISTORY

Kimiko Bowman

She could have been a horticulturist. Shehad, after all, thought it would be excitingto create new plants. But Kimiko Bowman,

better known by her byline as K. O. Bowman,chose to be a statistician and help develop a for-mula for approximating the distribution of max-imum likelihood estimators.

It was by chance, really, that Bowman chosestatistics. Some might even say statistics choseher. She began her college education majoring inhome economics at Radford College. I wanted

to make sure I would graduate, and I thoughthome economics would be relatively easy, saidBowman. In my second year, the president ofthe college called me into his office and advisedme to change my major to the science field. Hethought my future would be much better in sci-ence. I [had] liked mathematics ever since I was asmall child, so I immediately changed my majorto math and chemistry.

Never Say NeverBowman Helps Provide Formula for Approximating the

Distribution of Maximum Likelihood Estimators

Val SniderASA Publications Coordinator


7/84

SEPTEMBER 2008AMSTA

About a year later, Bowman graduated with abachelors degree and won a National Institutesof Health fellowship in mathematical statisticsfrom Virginia ech that sealed her interest in thestatistics profession. hree years after that, shehad in hand a masters degree and PhD in sta-

tistics. She completed her education with a doc-torate in mathematical engineering from okyoUniversity, but it was during her time at Virginiaech that her research really took hold.

Bowman met L. R. Shenton, her thesisadvisor, at Virginia ech andfor the next45 yearsworked with him to provide a for-mula for approximating the distribution ofmaximum likelihood estimators. hey suc-ceeded by developing formulae for skewnessand kurtosis statistics of maximum likelihoodestimators with respect to sample size and bydeveloping an approximation formula for the

percentage points of Pearson distributions.By combining the two results, Bowman andShenton were able to approximate the distri-bution of maximum likelihood estimators.

Another important accomplishment wassuccessfully implementing a divergent seriesalgorithm for large computers. In sampling,statistics series like this often occur, and thealgorithm for the coefficients is extremelycomplicated and reaches four-dimensional

One must nlose sight what we wato accomplisSo, I trienot to woabout slighor insignificathings.

space, said Shenton. Bowman not only imple-mented the algorithm, but it works, thanksto Dr. Bowman, said Shenton.

Bowmans life wasnt all work, however. She also

found time to be a strong advocate for those withdisabilities. She served on the National ScienceFoundation Equal Opportunities in Science andEngineering advisory committee and Committeeon People with Disabilities. She wrote a report

while chairing the latter that resulted in grantsbeing set aside to provide accommodation orspecial equipment for people with disabilities

which helped promote their participation in sci-ence and engineering at NSF.

Bowman also chaired the Statistical racking ofEmployment of People with Disabilities task forcefor the Presidents Committee on Employment of

People with Disabilities. Her main duty was tomonitor the questionnaire for Census 2000 sorelevant questions were asked about people withdisabilities and accurate results were obtained.

Perhaps her interest in being a champion forpeople with disabilities stems from the physicalhardships she, herself, has overcome. In her youth,Bowman suffered from polio that paralyzedher from the neck down. Doctors told her she

would never walk again, but after two years of

At Virginia Tech, Kimiko Bowman worked with L. R. Shenton, her thesis advisor, to provformula for approximating the distribution of maximum likelihood estimators.

Kimiko Bowman, who became a U.S. citizen in 1958,was the first foreigner to receive a doctorate from theUniversity of Tokyo.


8/84


rehabilitation, she was walking. Gladys Reynolds,a statistician and good friend of Bowmans fromher Virginia ech years, said, I want to highlighther determination and perseverance. In 1980,she was having polio-like symptoms and wasdiagnosed with Post Polio Syndrome. In spiteof this, she has continued to be as active andproliferate as always.

Bowman considers herself a triple minority,being an Asian and a woman, as well as someone

who suffers the effects of polio. here weremany struggles I had to overcome, Bowmansaid. However, one must not lose sight of what

we want to accomplish. So, I tried not to worryabout slights or insignificant things.

Instead, she became one of the first womento be elected Fellow of the American Statistical

Association in 1976, worked hard to advance sci-entific research using statistics, advised a number ofgraduate students, and published three books andapproximately 200 papers. One of those papers,ables for Determining Statistical Significanceof Mutation Frequencies, coauthored with M. A.Kastenbaum, received a Citation Classic in 1989for being the fourth most-cited paper in the his-tory ofMutation Research, an international jour-nal. Additionally, Bowman is an elected fellow ofthe American Association for the Advancementof Science (1970), an elected member of theInternational Statistical Institute (1978), and anelected fellow of the Institute of MathematicalStatistics (1987).

In an effort to do more for minoritthe statistics profession, Bowman also ba contributing editor to the Current IndStatistics in 1977 and continued in this pofor more than 10 years. During that timtranslated articles from Japanese journalEnglish, added keywords and abstractsincluded them in the index. She also was to go to Japan as a liaison scientist for the of Naval Research. She initially went formonths, but continued to visit throughocareer. In 1987, she was invited to attenInternational Statistical Institute meetiokyo and to give a keynote address at the sameeting at Mt. Fujiyama. During the meeshe received an audience with the Crown and Princess of Japanan unforgettable mfor her.

In 1994, Bowman retired from Oak National Laboratory as a senior researcentist. Retirement has not slowed her dhowever. She continues to be a guest scin the Computational Sciences and EnginDivision at Oak Ridge, working on resea

distributional properties of estimators anstatistics under non-normal sampling. Shconsults with colleagues within Oak Ridaspects of statistics such as procedures for ving computer models in economics and sttic models in epidemiology and biology.

words of Reynolds, As a woman, minorita person with a disability, she has certainlyan inspiration and mentor to many of us scientific community.

Kimiko Bowman,at Virginia Tech,advised a num-ber of gradu-ate studentsand publishedthree books andapproximately200 papers.


9/84

SEPTEMBER 2008AMSTA

Sampling Is inthe Details

[Scheuren] Irene, it is so good to see you well and still active pro-fessionally at 97. hanks for letting me interview you for all yourmany friends and colleagues who want to catch up with all youhave done. Can we start with something about your family andearly life before you came to the University of Michigan?

[Hess] I grew up in Kentucky. In addition to my parents, I hadone sister, Beulah Marie. I was at Evansville College three yearsand then went to Indiana University for one year for a bachelorsdegree, but thats the only degree I have. I guess you want me totalk about my start as a sampler?

[Scheuren]Well, of course, but a little more please about whatyou did before that. Didnt you teach in high school?

[Hess] No, I taught in junior high school in Central City,Kentucky. I was really not happy doing something like that,though. I was always interested in mathematics, and I would haveliked teaching mathematics at a higher level. And, when you areteaching in public school, you first know that youre to teach chil-dren. And I really didnt enjoy that. I couldnt enjoy grade-school

Interview withIrene Hess

Fritz Scheuren

Ida Irene Hess was born in Muhlenberg County, Kentucky. Her father was a mining engineer in CentralCity. Her mother, who graduated from Valparaiso College with her father, worked at home. Aftergraduating from Indiana University, Hess returned to Central City to teach math and English at the

local junior high school from 1932 to 1942. In 1940, she came to work with Leslie Kish at the SurveyResearch Center. Under Kish, she trained many graduate students in the details of sampling. he following

is an interview with Hess, conducted in February 2008 by Fritz Scheuren, an ASA past-president.

STATISTICIANS IN HIST


10/84


mathematics. I did not want to be responsible for teaching chil-dren and encouraging them in what I would consider to be theright way.

So, I decided to try for statistical employment in the federalgovernment. But to pass the civil service exam, you had to havea minimum of six hours in statistics and I did not have that, so I

went about getting it. I considered Indiana first, but there wasntanything there that interested me at that time. I was aware of Iowa.Leslie [Kish] asked me once why I didnt go to Iowa. I didnt tell him

why, but my mothers parents lived in Iowa. It was always so hotin Iowa in the summer, so I just wouldnt consider going to schoolin Iowa because I just couldnt take that heat. I ordered cataloguesfrom several universitiesIndiana, Michigan, and Kentucky, per-haps others. I came to Michigan two summers (1940, 1941). hatis how I got my six hours in statistics, and that was what I neededfor the civil service exam, which I passed.

[Scheuren]And you came to Washington after that?

[Hess]Yes, it was right after the start of World War II in Decemberand there was a civil service exam in May. he Bureau of LaborStatistics sent out invitations, and thats what I applied for and

went first to the Bureau of Labor Statistics. hen, I was at theBureau of Standards for a short time when they were developing

the proximity fuse for bombs in the war. When the war was over, Idecided I better get away from the Bureau of Standards. Of course,I had already been here, at Michigan, for two years because I wasreally interested in statistics. So, I went out to the [U.S.] CensusBureau and talked to somebody out there. hey were just organiz-ing the first sampling section at the [U.S.] Census Bureau.

Anyway, I joined the sampling section at the [U.S.] CensusBureau around 1944 or something like that. I was there untilI came here in 1954. Kish was in charge of sampling here in

Michigan, but his assistant was going to get married and maway and Kish needed somebody else. I was recommendKish and he wrote to me. In the meantime, at the [U.S.] Ce

Bureau, I think at least twice, when I came to work in the ming, they told me I should leave. I had no retention poinjustify my continuing employment at the [U.S.] Census Bu(When a person returned from military service, that personto be re-employed.)

[Scheuren] hey were laying off people?

[Hess]Yes, that was because of the return of veterans frowar, and every time someone came back and had to have they selected me to lay off. hat happened twice. Joe Stei

worked hard to get me back each time. Anyway, on one Friday evening, I walked into my apartment and picked umail and there was this communication from Leslie Kish ouin Michigan.

When I got the letter, I felt it was going to be a job offI knew that I was going to take it, but I didnt want to. I

Washington and I always liked the [U.S.] Census Burresponded to Leslies letter, and he invited me here to look arIve been here ever since, and that was 1954. In 1981, I retirebeen retired for 27 years.

[Scheuren] But youve been working here as a retired peevery weekday ever since? hats a real compliment tand to the center.

[Hess]Well it has been a long time. he way it began wasstarted this book [on sampling]. We had some very complexple designs for various projects and it bothered me that nobod

around here knew anything about the sampling activities. Sowritten or was working on that book, Sampling for Social Research Surveys, 19471980. I decided to stay and finish it.

[Scheuren] So, that is how you are continuing to work after ment? What came next that kept you coming here every day

[Hess] Roe Goodman. I dont know if you were ever acquwith Goodman.

[Scheuren] No, I wasnt, but I know of his work.

Irene Hess (left and above) worked at the U.S. Census Bureau during the late 50s and sthere until her retirement in 1981.



11/84

SEPTEMBER 2008AMSTA

[Hess] Goodman was sold on controlled selection.

[Scheuren] Well, I am too. In fact, one of my former students,Yan Liu, did her dissertation on balanced sampling with mearelated idea.

[Hess] Goodmans family was here in Michigan, so he came in one

day concerned about how well-controlled selection by computeris compared with hand-drawn controlled selection. So, thats howwe happened to have the bookControlled Selection Continued. AsI remember it, I think we were already well into the book before Ihad to retire.

[Scheuren]You had to retire because of age? What was the ageyou had to retire?

[Hess] Yes, I had to retire at 70. hat was in 1981.Anyway, Goodman was very interested in controlled selection andthe material we set up and worked on in the book. he last timeI saw Goodman was one afternoon before Easter. Goodmans wife

was already in Kansas and he was going to go take the train andjoin her. Anyway, he put in his final edits to the book. We didntknow they were final edits, of course, at the time. Well, he wenthome to Kansas. He had a heart attack and died that night. hatmade me determined to finish the book.

When my time came to retire, I probably still had somethingto do there. I dont know if it was all doneprobably not. I dontremember how long it took me to get these first two books finishedafter I retired.

[Scheuren] But there were more books after that werent there?

[Hess] Yes, several. he last one of these was publishSeptember 2007, I think.

[Hess, changing the subject]As you know, the center has donof work on telephone surveys. But, I never got very deeply intelephone field. All of these things I worked on were face-tarea probability surveys. I believed wholeheartedly in this app

And, hence, I never got involved in the telephone business.

How it started at the center was Charlie Cannell and Bobgot a project to work on telephone sampling. One day, I know if it was Cannell or Kahn, asked me if there was somebthe sampling section who could work with them on this teleproject they had and I told them Bob Groves. And, I also tolGroves that Cannell and Kahn would be talking to him. Andid, and they worked on various projects, and Groves has

with telephone surveys ever since. I was no admirer or interesthe telephone. Anyway, I was getting to the point where I h

Irene Hess worked with Leslie Kish, but

also worked with Mildred Jean Harter,seen here with Hess (above and right) in

Hess office in the late 50s.

Leslie Kish, ASA president in 1977;he passed away in 2000.


12/84


retire, so it was not something for me to get into and I had enoughof these other things [talked about earlier] to finish up.

[Scheuren]You sure did. Look at these accomplishments, all doneafter you retired, too.

[Hess] Now, Im completely out of any project and I dont haveanything that I really want to write about. So, I figure I ought toclean up everything that I have around here and get out.

[Scheuren] Dont do that. You do have more to do. At least helpme edit this interview, please?

[Hess] Bob Groves says he just likes to have me around.

[Scheuren, changing the subject] Let me ask you about your workwith Kish, and then we will finish up with your stint as the firstchair of the Section on Survey Research Methods.

Books by Irene Hess

Probability Sampling of Hospitals and Patients(1961)Irene Hess, Donald C. Reidel, and homas Fitzpatrick

Sampling for Social Research Surveys(1995) Irene Hess

Controlled Selection Continued(2002) Irene Hess andSteven G. Heeringa

The Practice of Survey Research at the Survey Research Center,19471980(1985) Irene Hess

[Hess] Early on, I did publish a number of papers with Kremember a paper in The American Statistician on nonresphat paper was something I guess Kish shared with me, achat was just shortly after I came here.

[Scheuren] hat nonresponse paper that you and Kish d

wonderful piece of applied work. I have cited it many times. WI used to do the History Corner in The American Statisticrepublished it.

[Hess]After about 1960, Kish got out of the day-to-day samactivities. He was going to educate the world. So, we didnttogether much after that. For example, he did not have anythdo with the projects that I was working on and what we were

with Roe Goodman. Kish was not involved with any of thwas just in a separate area of the center completely.

[Scheuren] One last item? Can you talk about your involvwith the Section on Survey Research Methods?

[Hess] Do you know how the section started? Within the Statistics Section, there was a subsection, and I guess it existmaybe two or three years. But anyway, at one time, I was nnated for chair of the subsection. here was an election withSocial Statistics Section and I happened to have been nomand then elected for chair of the subsection. In the meanthe powers that be decided to have a separate section for sresearch, and they saidbecause I had just won the electthat I would automatically be chair of the new Section on SResearch Methods. So, I was the first chair of the Section on SResearch Methods, 1977.

Do you remember this [showing the engraved silver plateto her by the Section on Survey Research Methods in 1998]

[Scheuren] You still have this? I remember it well. I think even have given it to you.

[Hess] hat was my last annual meeting. I remember it veryI was surprised, overwhelmed.

[Scheuren] Irene, you have done such wonderful things for Im going to read this inscription. he inscription says, Irenfor distinguished service and unstinting efforts in the furtheof survey research methods.

[Hess] In July 1999, I had extensive surgery on my right legI have never been free of some kind of health struggle sinctime. And I have never been to another meeting of the AmStatistical Association, and I guess Ill never get there again

anyway, 1998 was very special. hat was a real graduation fo

[Scheuren] Oh my, youve done so many things, Irene, andof your major publications were finished after you retired. Eyou were so busy day-to-day that you didnt have time tothese books. But, you did write them, eventually. How can thank you?

Irene Hess (left) celebrates her 97th birthday with Rhea Kish onAugust 26, 2007.


13/84

SEPTEMBER 2008AMSTAT

Chao-sheng Lee can be described as one of the most out-standing people in aiwans present statistical community.His 40-year career of teaching and research has resulted in

nearly 100 academic papers in well-known international periodi-cals. he National Chiao ung University (NCU) GraduateInstitute of Statistics, where Lee established himself when hereturned from the United States, has become a cornerstone for thedevelopment of aiwans growing statistical academic circles. heNCU Graduate Institute of Finance, which Lee helped estab-lish, has provided aiwans financial sector with the talent neededfor growth and innovation.

Lee graduated from National aiwan Universitys Department

of Business in 1964 and received his masters in economics fromthe University of Rochester in 1969. He studied under SeymourGeisser at he State University of New York, Buffalo, and receivedhis PhD in statistics in 1972. Upon completing his education, hebegan his teaching career at the University of Minnesota, and thentaught at Wright State University, as well as other universities, formany years. He then went to Bell Labs and devoted himself toresearch in economics and statistics.

Jack Chao-sheng Lee

Prepared by theInstitute of Statistics,National Chiao Tung University

he following excerpt is based on the 2007 article In Memoryof Professor Jack Chao-sheng Lee (19412007), whichappeared in the Journal of Data Science, Vol. 5, 143150. Fora complete version, visit www.sinica.edu.tw/~jds/In-memory-of-

Jack-Lee.pdf.

STATISTICIANS IN HISTO


14/84


In 1992, Lee was invited by Chi-Fu Den, the president ofNCU, and sang-Hai Kuo, the dean of the College of Science,to return to aiwan and take up the post of chair of the newlyestablished Institute of Statistics. Until stepping down from thispost in 1999, he persuaded many top statistical researchers whohad graduated from well-known schools to return to aiwanand teach. He also encouraged the now common practice ofresearch and acquiring software, as well as hardware. His efforts

enabled the Institute of Statistics at NCU to quickly riseto the level of other established institutes of statistics, name-ly those of National sing-Hua University and NationalCentral University.

In August 1995, Lee was the first academic in aiwan toreceive the ASA Fellow honor from the American Statistical

Association. Lee had already been selected as a member of theInternational Statistical Institute in 1990 and elected as thepresident of the International Chinese Statistical Association in1992. He was invited on multiple occasions to make speech-es around the world in many renowned schools, includingHarvard, Columbia, Cornell, and [the] University of Minnesotain the United States; Oxford and [the] University of London in

England; Hitotsubashi University and Kobe University in Japan;[the] University of British Columbia in Canada; and INSEADin France, to name a few.

In order to adapt to the trend of global financial developLee participated in the establishment of the Institute of Fiat NCU in 2002 and took on the responsibility of head institute for two years. During his tenure in charge, he was eUniversity Chair Professor of NCU in recognition of his dtion and accomplishments, as well as his reputation withidemic circles.

Lee had published many papers, and his research su

included many fields in statistics, including multivariate antime series analysis, growth curves, Bayesian inference, and fication and pattern recognition. He also published in othersuch as portfolio management and option evaluation, with palar application to finance.

Lee conducted his academic studies conscientiously. Sinreturn to aiwan, he has supervised numerous exceptional mand PhD graduates. He has four PhD graduates who are culy teaching in well-known universities and colleges. His mgraduates have also held pivotal positions in numerous arbusiness and academia. His pupils owe much to him, and hacapability to build on his success and to further glorify whhas started.

hroughout his life, Lee lived plainly and simply, and hsion for learning never diminished. He, now and always, wigreat role model for scholars to admire and learn from.


15/84


Paul Dixon Minton was born in Dallas,exas, the third of four sons in the familyof William M. and Addie Evelyn Croft

Minton. Having grown up in the economicallydifficult times of the Great Depression, he wasable to attend college only because he received anemergency scholarship to attend SouthernMethodist University (SMU). he university wasexperiencing its own difficulties, struggling to fillclasses, and decided to offer such scholarships toyoung Dallasites who had done well in highschool, were recommended by their principals,and could not otherwise afford to go to school.

Minton earned a bachelors and mastersdegree from SMU, a school to which he wouldreturn later to found and direct a department ofstatistics. His studies were interrupted by World

War II, during which time he worked as a crypta-nalyst for the FBI. Following the war, he returnedto Dallas as an instructor and graduate student

at SMU in the Department of Mathemawas at that point that Minton was introduprobability and statistics by Edwin Mouzonhad, himself, written a dissertation on staat the University of Illinois. After comphis masters degree, Minton was encouragMouzon to continue his graduate studies new program in statistics at he UniverNorth Carolina.

Minton earned a PhD in statisticsNorth Carolina State University undetutelage of Gertrude Cox, during the timInstitute of Statistics of the Greater Uniof North Carolina consisted of the Deparof Mathematical Statistics at UNC Chapand the Department of Experimental Staat NC State in Raleigh. Many famous scians were either faculty members or v

when Minton was a student there, incR. A. Fisher, Harold Hotelling, William Co

Building aDepartment

Paul Dixon Minton

Dwight B. Brock,Westat



16/84


Jacob Wolfowitz, Herbert Ellis Robbins, RichardL. Anderson, Raj Chandra Bose, and others toonumerous to mention.

Minton once related a story in which Fisher wassitting in the front row of the audience when he(Minton) gave his first paper as a graduate student.

As if that fact alone was not enough to frightenan already nervous graduate student, the situa-tion became infinitely worse (Mintons words)

when Fisher left the room in the middle of the talk.Minton told another story about his final PhD oralexamination, in which he had a bad case of laryn-gitis that he attributed to nerves. o his knowledge,he had the worlds only silent orals.

Minton returned once again to SMU andbegan to build a set of courses in mathematical andapplied statistics for students from a wide range ofsubject-matter departments in the university. heseofferings gradually evolved into the formation ofthe Department of Statistics at SMU, now knownas the Department of Statistical Science. At thesame time, Minton began to recognize the impor-

tance of computing in statistics, and, as a result ofhis expressed opinions, he was assigned to direct thefirst computer center at SMU, which housed theUnivac 1103, one of the few large scientific com-puters available at the time.

Funded primarily by a training grant from theNational Institutes of Health for training biostat-isticians, Minton established a new Departmentof Statistics at SMU in 1962. It was significantin obtaining faculty approval of the new depart-ment that he had provided research consultation ineither statistics or computingor bothto everydepartment in the university. he new department

offered masters and doctoral degrees following theNorth Carolina model and received consultationand assistance from Cox. he department subse-quently expanded to offer degrees at all levels, toprovide consultative assistance to faculty researchand outside clients, and to conduct research in sta-tistical theory and methods.

After 10 years of developing and administeringthe SMU department, Minton moved toRichmond, Virginia, to take the position ofdean of the School of Arts and Sciences at

Virginia Commonwealth University in 1972.here, in addition to his duties as dean, Mintonformed an Institute of Statisticsa form ofliaison office between the Department ofMathematical Sciences in the School of Arts andSciences and the Department of Biostatistics inthe Medical College of Virginia. He also wasactive in statistical consulting in local industry,including paper manufacturing, polymer plasticsprocessors, pharmaceuticals, a national bakingcompany, and a federal agency. He retired from

VCU in 1988. hat year, a special symposium

was organized in his honor, attended by former students and colleagues.

Minton was the recipient of numerous acaand professional awards, including being naFellow of the American Statistical Associatiobeing an early recipient of the ASA Founders for service to the association and the professihis honor, the SMU Statistics Department cthe Paul Minton Award for the student whohighest on the basic qualifying examinationthe ASA created the Paul Minton Service A

which has been given annually since 1992. Helongtime member of the ASA, IMS, and BioSociety. He founded the North exas Chapthe ASA and was active in a variety of roleing the years he was in Dallas. Later, he waactive with the Virginia Academy of Science

Virginia Chapter of the ASA) and in organiziSouthern Regional Conferences on Statistiparticipated in numerous committees, task councils, and boards, and he served one tevice president of the ASA.

Minton had an active life outside the tics profession that included a love of clamusic and the ability to create terrible puncombined those interests with his knowlestatistics in the following way. As part

ASAs 150 celebration in 1989 in WashinDC, he and some friends created a stcal songbook, from which they performa banquet held on that occasion. In the of these statistical punsters, the tune Hai from the musical South Pacific bthe results for a test of statistical significBarely High. hese and other rewritten

lyrics can be found in the songbook, whicsubsequently published in the August issue ofThe American Statistician.

In a sense, entertaining audiences with hous statistical lyrics to well-known songs banother way for Minton to be an advocastatistics and promote the profession to a

world. His life and work as a statisticiansummarized at his memorial service in the of his son, Roland: He recruited studethe newly developing field of statistics, he money to support them during their stdays, and he helped them find jobs aftegraduated. Others have said of him that ha gentle person, with a gentle temperamenhe changed the lives of hundreds, if not sands of statistical students, just as he chthe lives of many others who knew him othe profession.

A memorial session for Minton was held2008 Joint Statistical Meetings in Denver. from that session will be included in the

Archives Collection at Iowa State Universit

R. A. Fisher

Harold Hotelling

Jacob Wolfowitz

William Cochran


17/84


Mollie Orshansky was born on January 9,1915, in the Bronx in New York City.She was the daughter of Ukrainian-

Jewish immigrants who spoke limited English.Although her father worked hard at a number ofjobs, Orshansky and her sisters grew up poorinher words, the family could barely make endsmeet. he girls slept two to a bed and wore hand-me-down clothing. Orshansky remembered going

with her mother to stand in relief lines for surplusfood. As she was to say later, If I write about thepoor, I dont need a good imaginationI have agood memory.

Orshansky was both the first high-schoolgraduate and first college graduate in her family.

She graduated from Hunter College High Schoolin Manhattan (then a school for gifted young

women) in 1931. Because she received two schol-arships from the college, she was able to attendHunter College (at that time a womens college).She graduated from Hunter in 1935 with an ABin mathematics and statistics. She was a statisticianby training and profession, although she has some-times been referred to as an economist.

Mollie Orshansky:

Author of

the PovertyThresholds

Gordon M. Fisher

Editors Note:Views expressed in this article are those of the author, and should not be construed as representinpolicy of the U.S. Department of Health and Human Services. his article is condensed from an unpublished dpaper with references. hat paper is based on extensive research involving published articles and documents, unplished documents, and conversations with Mollie Orshansky.

A daughter from a poor family graduatingcollege during the Great Depression, Orshans

not have the luxury of attending graduate before she started working. Instead, she foundas a statistical clerk in the New York DepartmHealths Bureau of Nursing, where she workinfant mortality and other subjects for a year.

In 1936, the U.S. Childrens Bureau (then the U.S. Department of Labor) offered Orsha

job as a junior statistical clerk. She accepted thmoving from New York to Washington, DCfirst job assignment involved logarithmic equfor 600 infants who had been in a study. I1939, the bureau promoted her to research c

job in which she stayed until January 1942,

ing on biometric studies of child health, grand nutrition.

It was while she was working at theChildrens Bureau that Orshansky began graduate courses. At various times between 1937 and 1960, she took courses in economistatistics at the U.S. Department of AgricGraduate School and American University.

January 1942 to March 1943, Orshansky too



18/84


as a statistician for the New York City Department of Health, won a survey of the incidence and therapy of pneumonia.

alking about her government career, Orshansky once comed, I basically always worked with women, except when I was

war agencies [the National War Labor Board during World Wand the Wage Stabilization Board during the Korean War]. D

World War II, in particular, large numbers of men left civilian jserve in the military. As a result, a number of female workers we

to get jobs that would not have been open to them under ncircumstances. Describing the situation for female federal woDorothy Rice, a colleague of Orshanskys, later said, Any femahad anything on the ball really did very well during the war. Amen went to the war and we had to carry on.

In March 1943, Orshansky secured a job at the U.S. NationLabor Board as chief of the Program Statistics Division of the Reand Statistics Branch. She planned and executed the collectioanalysis of data required for the boards decisions on wage adjustand the effects of the boards stabilization activities. She stayed

job until September 1945, when the operations of the boardbeing terminated.

From September 1945 to March 1951, Orshansky work

a family economist at the Bureau of Human Nutrition and HEconomics of the U.S. Department of Agriculture (USDA). Shducted research in family consumption and levels of living, caout a variety of assignments. She was in charge of preparing dthe estimated value of household furnishings and equipment onfor the departments balance sheets of agriculture for 1947195

In 1948, Orshansky and a colleague were responsible for resing to letters from members of the public asking how they couldends meet on their existing income in the face of the severe (bstandards) postWorld War II inflation. Orshansky and her col

would send the letter writers pamphlets about preparing a familyget and planning low-cost and moderate-cost meals using Ufood planshypothetical food budgets providing nutritious ddifferent cost levels. (his shows Orshansky was working with Ufood plans at least 15 years before she was to use them to developoverty thresholds.)

About 1949, Orshansky carried out an assignment to update dard budget (an estimate of necessary living costs) for a single ing woman in New York. In June 1950, she presented a paperEquivalent Levels of Living: Farm and City at the annual mof the Conference on Research in Income and Wealth (CRIWpaper and the other papers presented at the meeting (includinby Milton Friedman) were published in volume 15 of CRIWs a(and still continuing) series, Studies in Income and Wealth.

In March 1951, during the Korean War, Orshansky sea job at the U.S. Wage Stabilization Board as the director Program Statistics Division of the Office of Economic Analys

planned and directed the boards statistical program until A1953, when the operations of the board were being terminateto the end of the war.

In 1952, while Orshansky was working at the U.S. Stabilization Board, she was elected a member of the EconomSociety (an international society for the advancement of econtheory in its relation to statistics and mathematics). Of the 7didates for election to membership in 1952, Orshansky wonly woman.

When Orshansky developed her poverty thresholds, shemade use of information she had worked with at the U.S.Department of Agriculture (USDA). She based her thresholds

on the economy food plan, which was the cheapest of fourfood plans (hypothetical food budgets providing nutritiousdiets at different cost levels) developed by USDA.

From a finding of USDAs 1955 Household Food ConsumptionSurvey (the latest such survey available during the early1960s), she knew families of three or more persons had spentapproximately one-third of their after-tax money income onfood in 1955. The way in which she used this survey findingwas by considering a hypothetical average family that wasspending one-third of its income on food and by assumingthe family had to cut back on its expenditures sharply. Sheassumed expenditures for food and non-food would be cut

back at the same rate so the family would continue to spenda third of its income for food.

When the food expenditures of the hypothetical familyreached the cost of the economy food plan, she assumedthe amount the family would spend on non-food itemswould also be minimal, but adequate. (Her procedure didnot assume specific dollar amounts for any budget categorybesides food.) Following this logic, she calculated povertythresholds for families of various sizes by taking the dollarcosts of the economy food plan for families of those sizes andmultiplying the costs by a factor of threethe multiplier.(She followed somewhat different procedures to develop

thresholds for two-person and one-person units.)

She differentiated her thresholds not only by family size, butby farm/nonfarm status, by the gender of the family head, bythe number of family members who were children, and (forone- and two-person units only) by aged/non-aged status.

The result was a detailed matrix of 124 poverty thresholds, laterreduced to 48.

To avoid confusion, the preceding explanation has beenphrased in terms of the economy food plan. However,Orshansky actually developed and discussed two sets ofpoverty thresholds, one derived from the economy food plan

and one derived from the somewhat less stringent low-costfood plan. (The latter set was the one she preferred.) It was thelower of the two sets of poverty thresholdsthe set derivedfrom the economy food planthat the Office of EconomicOpportunity adopted as a working definition of poverty inMay 1965. One probable reason for the adoption of the lowerset of thresholds was that the lower set yielded approximatelythe same number of persons in poverty as the Council ofEconomic Advisers $3,000/$1,500 poverty line.


How Mollie OrshanskyDeveloped the Poverty Thresholds


19/84


From August 1953 to February 1958, Orshansky worked as a foodeconomist for USDA in a successor office to the Bureau of HumanNutrition and Home Economics, where she had worked before. Sheplanned and directed the collection and analysis of data on food con-sumption and expenditures of American households.

In 1956 and 1957, Orshansky was the senior coauthor of tworeports on family food expenditures and food consumption based ona food consumption survey of rural families in the North Central

[Midwest] region. She was one of a number of people who gave tech-nical assistance in the preparation of a series of reports on USDAs1955 Household Food Consumption Survey, and she wrote amajor section of a summary report on the same survey. his 1955Household Food Consumption Survey was the source from whichOrshansky would calculate the multiplier she later used to developher poverty thresholds.

In February 1958, Orshansky went to work for the Social SecurityAdministration (SSA) in an office that later became the Office ofResearch and Statistics (ORS). ORS seems to have been one of asmall number of federal officesthat provided significant work oppor-tunities for women professionals at this time; during Orshanskys firstdecade there, both the director and the deputy director were women.Orshansky had several titles at SSA, but can best be described as asocial science research analyst.

Orshansky performed a number of assignments during her earlyyears at SSA. Her first was to prepare an article on standard budgets(family budgets) and practices in setting fee scales in 21 large cities.She also prepared several annual updates of an analysis of the incomesources of young survivors (widows under age 65, particularly those

with minor children). She prepared a medical care standard for theBudget for an Elderly Couple, of which the Bureau of Labor Statistics(BLS) was preparing an interim revision. She also prepared data for16 charts on financial resources of the aged for the Chart Book for the1961 White House Conference on Aging.

While Orshanskys development of the poverty thresholds was amajor milestone in both social policy history and statistical history, it

grew out of ordinary work activitiesan answer for the record fora congressional hearing and an in-house research project. During a1960 congressional hearing, a senator asked HEW Secretary ArthurFlemming if he had figures on how much it costs a retired couple tolive. Flemming said HEW would provide an answer for the record,and Orshansky was the civil servant who prepared an unattributedsubmission for the record. She mentioned the Budget for an ElderlyCouple, which BLS was then revising, and a similar budget preparedby a group in New York. In addition, she provided two rough mea-sures of income inadequacy for an elderly couple that she developedby applying multipliers derived from USDAs 1955 Household FoodConsumption Survey to the cost of USDAs low-cost food plan (at thattime, the cheapest of USDAs three food plans)almost exactly as she

was to do several years later in her 1963 and 1965 poverty articles.In early 1963, Orshansky was assigned to do an in-house research

project on poverty as it affects children. At that time (the year beforethe War on Poverty was declared), there was no generally acceptedmeasure of poverty, so to carry out this research project, Orshanskydeveloped one (see How Mollie Orshansky Developed the Povertyhresholds on the previous page for her methodology). In July 1963,she published results of her research project in a Social Security Bulletinarticle, Children of the Poor, in which she also described the initialversion of her poverty thresholds.

Orshansky told several interesting stories aboutevents during the development of her thresholds

As noted in the sidebar, How Mollie Orshansky Developed thePoverty Thresholds,Orshansky based her thresholds on USDAseconomy food plan. She had been working with the USDA fooplans at least as early as 1948, and so was familiar with them. Inthe context of developing poverty thresholds for families, she

became concerned about the economy food plan not allowinfor purchases of food away from home, either at work or schooFor the purpose of developing poverty thresholds, she wantedto modify the cost of the food plan by adding $0.15 a day perperson to it to allow for the husband in a family to buy coffeeat work or for children to buy snacks. However, her supervisorwould not allow her to do so.

One major source for Orshanskys July 1963 article was aspecial tabulation of Current Population Survey (CPS) data ththe Social Security Administration (SSA) purchased from theU.S. Census Bureau for $2,500. The results showed the medianannual income of nonfarm, female-headed families with

children was $2,340. Orshansky was horrified when she realizhalf of these families lived for a year on less than SSA paid forone statistical tabulation. She later commented, I determinedwas going to get my $2,500 worth.

In the CPS, thousands of sample cases represent millionsof families and persons in the general population. PublishedU.S. Census Bureau reports based on survey data alwaysshow figures that relate to the national totals of families andunrelated individuals. However, when Orshansky got the resuof her $2,500 tabulation of CPS data, they gave her only theunweighted counts of sample households in various povertyand nonpoverty categories; SSAs payment to the bureau had

not been enough to pay for computing weighted nationaltotals for her, so she had to do the work herself, by hand, tocalculate weighted national totals from the unweighted samcount. She also calculated the poverty gap by hand for herJanuary 1965 article. She didnt even use a calculator.

In January 1964only six months after the publicatiOrshanskys obscure articlePresident Lyndon Johnson decl

war on poverty. In a chapter on the problem of poverty in itsannual report, the presidents Council of Economic Advisers put forward its own rough measure of poverty: $3,000 for faof all sizes and $1,500 for unrelated individuals. he CEAs $

figure was not derived in any way from Orshanskys work; hothe CEA report did cite two dollar figures from Orshanskys Julyarticle (without giving her name as the author) to show that the$3,000 figure was a reasonable level for a poverty line.

When Orshansky saw the January 1964 CEA report (includireference to her dollar figures), she was disturbed by the CEAs to vary its $3,000 family poverty line by family size, as this rein understating the number of children in poverty relative topersons. he CEA figure led to the odd result that an elderly c


20/84


with $2,900 income would be considered poor, but a family witha husband, wife, and four little children with $3,100 income wouldnot be. In addition, the presidents declaration of a war on povertyevidently led SSA to give a higher priority to Orshanskys poverty

work. As a result, Orshanskys supervisors asked her to do an analysisextending her families-with-children poverty thresholds to the wholepopulation. She completed this analysis in late 1964 and it was pub-

lished in the Social Security Bulletin in January 1965 as Counting thePoor: Another Look at the Poverty Profile.

he publication of Orshanskys January 1965 article came whenthe Office of Economic Opportunity (OEO)the lead agency forthe War on Povertywas being set up. OEO officials were enthusi-astic about Orshanskys poverty thresholds, considering them to bean advance over the CEAs $3,000-for-all-family-sizes figure. OEOresearch chief Joseph Kershaw commented, Mollie Orshansky saysthat when you have more people in the family, you need more money.Isnt that sensible? In May 1965, OEO adopted Orshanskys thresh-olds as a working definition of poverty for statistical, planning, andbudget purposes, and, in August 1969, her thresholds were made thefederal governments official statistical definition of poverty.

When she developed the poverty thresholds, Orshansky was anobscure civil servant who worked [d]own a dimly lit hall, amongstacks of computer printouts [at] a paper-covered desk However,after her thresholds were adopted as the federal governments povertyline, she became much more well known. Because of frequent citationsof her work in academic articles and books, someone once referred toher as the ubiquitous footnote. Besides presenting papers at a num-ber of professional meetings and publishing a number of articles, shetestified and/or provided written documents to congressional commit-tees on 10 occasions between December 1967 and 1990.

On five occasions between 1968 and 1980, Orshansky participatedin federal interagency committees that reviewed the poverty thresh-olds. he 19681969 committee made two modest revisions inthe thresholds, and it was the thresholds with these revisions that

were made the official federal statistical definition of poverty. In1981, several minor changes recommended by the 19791980committee were made.

Following up on a 1973 subcommittees recommendation for anew income survey vehicle, the HEW echnical Working Group onincome data and models proposed that a new survey be developedto provide better information on the income and related characteris-tics of the population and on participation in government programs.

Orshansky was a member of this group. he technical workingreviewed and contributed to the plan for what became the InSurvey Development Programthe research and developmentfor the Survey of Income and Program Participation.

In 1982, Orshansky retired from SSA after a government that lasted for more than 40 years. She died on December 18, in New York City.

Orshansky received a number of honors for her achievemShe received a Commissioners Citation from the Social Se

Administration in 1965 for her creative research and analyticaand the Distinguished Service Award (the departments highest nition of civilian employees) from HEW in 1976 for her leadin creating the first nationally accepted measures of income adcy and applying them diligently and skillfully to public polic1974, she was elected a Fellow of the ASA for her leadership development of statistics for the measurement of poverty. Aftretirement, she received a national award from the Childrens DFund in 1989 and an Award for Distinguished Contribution fro

American Political Science Association in 1993.Orshanskys achievements also were recognized in a very di

setting. She may be the only statistician to have been discussedmajor television show. One subplot of he Indians in the Loan episode of he West Wing originally broadcast in Nov2001, involved the adoption of a new poverty measure, and oneacter alluded to how Orshansky developed the current povertysure. While the discussion of issues relating to a new poverty msounded plausible, the episode grossly mischaracterized the ratfor Orshanskys methodology for developing the poverty thresh

Of the contributions to American public policy that Orshmade during her career, the greatest by far was her developmentpoverty thresholds. he poverty line has become a major featthe architecture of American social policy. Although the measurhave its shortcomings, the poverty line gives us a means of ident

and analyzing the makeup of the groups in our society with thresources. Orshanskys thoughtful analyses of the poverty popubegan a tradition, and there are now numerous researchers andcates who conduct such analyses and draw policy implicationthem. Even though there may not be consensus on answers, thetion How does it affect the poor? has become a test for propolicies and programs. And a simplified version of the povertyused to determine eligibility not only for certain federal programfor a number of state, local, and private programs, as well.


Mollie Orshansky (third from left) at the first Conference on Women in the War on Poverty on May 8, 1967. With her are Mary Keyserling, JosephWeiner, and Hyman Bookbinder. Near the end of her speech, Orshansky said, our statistics, imperfect though they may be, show us where prlems are even if they cannot always reveal exact dimensions[C]alculations relating to poverty exist only to help make them disappear, aif we can think bold solutions and dream the big dream, we can wipe out the scourge of poverty before we all agree on how to measure it.


21/84

Did you know your ASA membership includes online access to

Journal ofBusiness & Economic Statistics

Te American Statistician?

and

Journal of the American Statistical Association

Log in at MEMBERS ONLY todayto access your journals!www.amstat.org/membersonly

Te ASA

Brings the Libraryto Your Desktop


22/84


Day in the

LIFE


23/84


24/84


How Did I Get Here?So, that is the story of what I do. But, how did I get here? I fol-lowed a fairly typical course for my time, starting as an under-graduate major in mathematics. In my senior year, I took anintroduction to mathematical statistics course from the book bythe same name written by Robert Hogg and Allen Craig (embar-rassingly, using the first edition, spanning all of 240 pages). hematerial was so elegant, I was smitten, and I applied to graduate

programs in statistics, moving to Stanfordas far from NewYork City as I could go.

I thoroughly enjoyed my four years at Stanford. (he late1960s was a remarkable time for a statistics department in theSan Francisco Bay areaand for the country.) I emerged withtraining as a mathematical statistician, really the only path avail-able at that time. As I have told many people, I had a wastedyouth in some ways, not discovering until the late 1980s that I

was born to be a Bayesian (and, with it, to be a modeler).People who influenced me include my adviser from Stanford,

Herbert Solomon, a wonderful, generous man who was particu-larly good at recognizing how people could best contribute toprojects. He was one of the earliest team builders in our field.

Charles Stein, who was also from Stanford, influenced me ina distant way. I believe I hold the record for the most coursesever taken from Stein by a single student. His beautiful, deepthoughts in decision theory led me down that research path,

which led me to empirical Bayes and, eventually, research wthe Bayesian paradigm. Finally, Adrian Smith was most intial to me. Following conversations with him in the mid- t80s, I arranged a sabbatical to work with him at NottinUniversity. I arrived seeking to use his numerical integsoftware for certain empirical Bayes problems and left wiGibbs sampler!

In this regard, I was fortunate to be involved in a seminal c

bution to our field. Indeed, working with Smith, it was a reable feeling to come upon the Gibbs sampler from the 1984of Stuart Geman and Donald Geman, Stochastic RelaxGibbs Distributions, and the Bayesian Restoration of Imagesrecognize it was better suited for handling Bayesian computhan it was for its original purposesampling Markov rafields. In fact, for the first half of the 1990s, I spent almomy time writing as much about the use of Gibbs samplinMarkov chain Monte Carlo algorithms for Bayesian model and model determination as I could. hese were heady timthose of us involved, as we realized we were elaborating on anology to successfully address the problem of Bayesian cotationto replace high-dimensional integration with simu

from high-dimensional distributions.In the mid 1990s, I started moving from research confincomputational matters to the analysis of spatial and spatial-temdata. Again, I was fortunate. Spatial statistics had strugg

MacKichanS O F T W A R E , I N C .

LATEX

LATEX

LATEX


25/84

a field, functioning at the periphery of mainstream statisticalwork, and was particularly in need of inexpensive, high-speedcomputation to enable tools vital for visualization. Fortunately,this computing power arrived and, with it, the advent of geo-graphic information systems (GIS) software to create effectivemaps that tell a broad range of stories. However, GIS softwarehad an obvious limitation from a statisticians perspectiveit wasdescriptive, but not formally inferential. A wonderful opportu-

nity for stochastic modeling revealed itself, an opportunity withfew players at the time and a growing need in the applied com-munity as more data with spatial content were being collected.

Here, again, there was more good fortune for me, becauseadvantages to hierarchical modeling for analyzing spatial databecame evidentthe advantages we attach to working in gen-eral in the Bayesian framework: fully model-based inference withaccurate assessment of uncertainty. Additionally, the customaryasymptotics, which are employed in time series, so-called expand-ing domain asymptotics, may be inappropriate for spatial data

where infill asymptotics are perhaps more relevant. However,that the latter asymptotics typically reveal information aboutunknowns is bounded in terms of inference. In this regard, theBayesian framework provides exact inference, avoiding possiblyinappropriate asymptotics. Of course, the implied caveat is thatthe data never overwhelm the prior; we must be more attentiveto prior sensitivity than in other areas.

oday, my good fortune persists through continuing researchrelationships with talented former students. In particular,Sudipto Banerjee and Brad Carlin, both at the University ofMinnesota, invaluably helped to shape my spatial data researchagenda. During my career, I also found wonderful interdisciplin-ary collaborators. When I was at the University of Connecticut,I worked seriously at building bridges with ecology and evolu-tionary biology, yielding productive research relationships with

John Silander and Kent Holsinger. Moving to Duke Universitymade things even better, as Duke is a naturally collaborative

institution; research teams develop over the campus with strongencouragement from the university. What program on campusis more naturally interdisciplinary than statistics? Now at Duke,I have productive relationships with researchers in the NicholasSchool of the Environment, particularly Jim Clark and MarieLynn Miranda.

In summary, it seems my career has been one of ongoing goodluck. To some extent, I believe this is true, but I also believeyou can make your own luck. Scientific curiosity, receptivity toreading in other fields, availability for collaboration, willingnessto listen, and attention to effective communication all facilitatecreating opportunities. In addition, developing the full toolkitstrong theoretical training, stochastic modeling expertise, mod-

ern computing skills, and data analysis experienceenables oneto take advantage of these opportunities. Some of us work better

with certain parts of the toolkit and enjoy certain parts morethan others; however, as statisticians in the 21st century, we caneach find our own way to contribute and, in that sense, we areall fortunate.

Make the most of your

ASA membership.Visit the ASAMembers Onlysite!

-Personalize your login ID

-Access all your ASA journal, CIS, andJSTOR subscriptions with one easy login

-Partner discounts from Wiley, CRC Press,and others

-ASA member forums-Enews archive

Welcome to ASA


26/84


27/84

SEPTEMBER 2008 AMSTATSEPTEMBER 2008 AMS

DAY IN THE

As a fledgling mathematician, I was blessed to learn from sev-eral outstanding teachers. First and foremost, there wasSammi Yopp, my second-grade teacher at the North Carolina

Charlotte Country Day School. Yopp sparked my mathematicalcuriosity and cajoled me into playing chess as a means of honing myanalytic thinking skills. Several years later, my mathematical abilities

were challenged by Sue Sams, my ninth-grade honors algebra II/trigonometry teacher at East Mecklenburg High School. Sams was,to put it mildly, a disciple of Mary P. Dolciani. he following year,I landed in the classroom of another Dolciani disciple, RogerBruwhel at West Charlotte High School. Bruwhel was the one whoencouraged me to leave my comfort zone by applying to the NorthCarolina School of Science and Mathematics (NCSSM). Off toNCSSM I went, where I was surrounded by a host of talented

mathletes, budding scientists, and computer programmers. As ajunior sitting in Dan eagues Calculus with opics class one day,I made the decision to become a high-school math teacher. Why?Because I witnessed firsthand the tremendous passion, thirst forunderstanding, and knack for asking just the right question at theright moment that allows great teachers such as eague to inspirestudents to do extraordinary things.

From NCSSM, I went on to pursue a bachelors of science inmathematics with secondary teaching certification at the Universityof North Carolina, Charlotte (UNCC). At the time, UNCC wasa teaching university, which enabled me to take all my classes withprofessors from the outset. At the urging of my professors, I par-ticipated in Budapest Semesters in Mathematics, where I fell in

love with combinatorics and probability. Having never left thesouthern United States until that moment, I admit to experiencinga healthy dose of culture shock as a westerner in what was thensocialist Budapest.

From Eastern Europe, I accepted an offer to join a graduateprogram in pure mathematics at the University of Michigan. As I

was nearing the end of my masters degree requirements, it became

On Becoming a Teacher

Daren Starnes, The Lawrenceville School


28/84


clear to me that my interests had shifted dramatically away fromclassical pure mathematics in the direction of research methods,combinatorics and graph theory, and statistics. Ironically, my onlystatistical coursework to that point was a calculus-based probabilityand statistics course as a sophomore at UNCC. Having reached animportant fork in the road, I opted to end my relationship withpure math and begin my high-school teaching career.

Charlotte Country Day School (CCDS) was my provingground, the place where I learned the craft of teaching from sup-portive colleagues such as im imson, Sheila McGrail, and SueSchwartz. My interest in curriculum development blossomed.CCDS launched the International Baccalaureate (IB) program, andI volunteered to take on the IB Higher Level Math course. In myfifth year there, McGrail and I designed a course called Probability,Statistics, and Finite Math to better serve the needs of students

who were not bound for AP Calculus. When AP Statistics koff in 19961997, I happily agreed to play a leadership role. still, CCDS science colleague Wright Robinson convinced team teach AP Statistics with AP Environmental Science, tlaborate on a field and lab manual, and to co-present at sconferences. Robinson got me started as a writer, having hauthored several childrens books about ecological topics. than anything, Robinson convinced me to get out there as

tributing member of my profession.When our youngest son graduated from CCDS, my wi

I headed west to he Webb Schools in California to begnext chapter in our lives. I assumed leadership of the mathedepartment, and with the support of some exceptional collebegan to tweak the existing curriculum. We dissolved our exgeometry course and created an integrated geometry with acourse that continues to evolve today. In addition, we added stone course for our most talented math students titled CalBased Probability and Statistics.

Externally, my role with the College Board expanded dracally. I led four or five one-day workshops for AP Statistics teevery year and a similar number of week-long summer inst

hese opportunities to help coach other teachers gave me a cto share some of the lessons I learned in the trenches. At thetime, interacting with other highly motivated professionals enaged me to further refine my own teaching.

In the summer of 1998, I began reading AP Statistics eEvery year since, I have attended this annual gathering of school and college/university statistics teachers, which somerefer to fondly as stats summer camp. For the past eight I have been privileged to serve as part of the reading leadteam, first as a table leader and, more recently, as a question lNeedless to say, reading hundreds of exam papers every daythe highlight of the experience for most of us. Instead, its thmendous opportunity to network with like-minded profess

who have amassed an abundance of content knowledge and of resources for teaching introductory statistics. here are abuopportunities for discourse during breaks, at lunch, in the evsocial lounge, and at various nightly professional events. rutold, some of my best conversations about statistics teachingtaken place at AAA baseball games.

Shortly after my wife and I moved west, Dan Yates conme about assisting him with some of the calculator features second edition of The Practice of Statistics, the successful school AP Statistics textbook he coauthored with David MHaving spent three years co-editing the echnology ips coluthe National Council of eachers of Mathematics (NCM) zine,Mathematics Teacher, I felt comfortable accepting this ra minor contributor in the writing process. Somewhere alon

way, I mentioned to Yates that I had a few suggested enhments to the text based on my experiences as a user. o my suhe called my bluff and asked me to compose a revised versChapter 1 that incorporated my recommendations. Doing sofar longer than I had expected, with many false starts alon

way. he net result of my efforts was a 120-page tome disas a single textbook chapter. With a typical dose of his soucharm, Yates quipped, You know, if chapters were this longhave to package the book with a wheelbarrow. In the end,

If I have become a teacher leader, it isbecause I have followed in the footstepsof the many great mentors and role mod-els who have shaped my thinking.

Environmental Section

Upcoming ENVR Workshop!What: Statistical Issues in Monitoring the Environment

When: October 2224, 2008

Where: National Center or Atmospheric Research(NCAR), Boulder, Colorado

This workshop covers state-o-the-art applications andstatistical methods in environmental monitoring. Sessionson applications include monitoring in ecology, monitor-ing in air quality, monitoring o aquatic resources, andmonitoring o climate change and its impacts. The spa-tio-temporal data collected in environmental monitoringpresent interesting and challenging statistical problems,such as modeling o space-time correlation, analysiso the huge amount o correlated data, and analysis ohigh-requency monitoring data. Technical sessions willcover recent developments in statistical methods or envi-

ronmental data. A one-day short course on the analysis ospatial and spatio-temporal data will be oered and taughtby Doug Nychka. There also will be a poster session. Posterabstracts need to be submitted by September 23, 2008.For more inormation, contact Hao Zhang, Department oStatistics, Purdue University, West Laayette, IN 47906; (765)496-9548;[email protected].


29/84


instructed to trim the chapter to 60 pages. hat was probably themost painful editing task I have ever undertaken! (here was oneconsolation: Nearly all the trimmed material ended up in a supple-ment somewhere.)

So, my second career as a writer was launched. I teamed with Yatesand Moore on a second ventureStatistics Through Applications, ahigh-school textbook for nonAP Statistics coursesthat was com-pleted in 2004. With both of my coauthors now retired, Ill have my

hands full with future revisions.In 2004, I was appointed to the ASA/NCM Joint Committee

on Curriculum in Statistics and Probability by then NCM presidentJohnny Lott. Over the years, the joint committee has been respon-sible for several significant projects to develop materials that promoteK12 statistics education, including the Quantitative Literacy Series,the Elementary Quantitative Literacy Series, the Science Educationand Quantitative Literacy Series, and Data-Driven Mathematics.he joint committee also has assumed responsibility for designingteacher workshops to support effective classroom use of these materi-als. With the ASAs recent publication ofGuidelines for Assessment andInstruction in Statistics Education (GAISE), the committee has begundeveloping materials and training to help teachers implement the cur-

ricular framework described therein. Roxy Peck and I just completeda capstone module for high-school students, titled Making Sense ofStatistical Studies, which began as an ASA Member Initiative sev-eral years ago when Peck, Henry Kranendonk, and June Morita weremembers of the joint committee.

After a brief stint in independent school administration, I aed an endowed chair position as Master eacher in Mathemahe Lawrenceville School this past fall. his unique position me to team teach with a different member of the departmentrimester. By planning, instructing, and assessing collabora

we are able to improve our individual teaching practices. Oudents also benefit from the distinct approaches and perspecttheir co-teachers. Other responsibilities as master teacher incurriculum design, technology training, and professional dement. Much of my time outside the classroom is spent infessional growth activities involving other teachers, either hLawrenceville or across the country at meetings, conference

workshops. Distilling that collective wisdom into better leafor students is at the heart of what I do on a daily basis.

If I have become a teacher leader, it is because I have followthe footsteps of the many great mentors and role models whoshaped my thinking. As my career has evolved, I have driftedmy origins as a pure mathematician into the realms of data anand statistics. All the while, my primary focus has remained sly on how to help students learn mathematics (including stain ways that will ensure their ability to use what they have le

to solve important problems today, tomorrow, and in the dfuture. After all, if students see no potential to do anything prtive with what theyre learning, why should they bother?

DAY IN THE


30/84

KA

SOUTH DAKOTA

NEBRASK

WASHINGTON

OREGON

MONTANA

IDAHO

CALIFORNIA

NEVADA

UTAH

ARIZONA

NEW MEXICO

TEXA

COLORADO

WYOMING

NORTH DAKOTA

Where Are YouColleagues

Graphic by Val Snider

Geofrey BohlingKansas Geological Survey

Lawrence, Kansas

My work is aimed at usinginormation to reduce the

non-uniqueness in subsuracecharacterization problems.

Stephanie Lopez CanoThe University o Texas

San Antonio, Texas

Aside rom teaching statistics inthe classroom, I am director o the

Statistical Consulting Center.

Harold NelsonOf ce o Financial Management

Olympia, Washington

My group does a lot o purelydescriptive work, a lot o it

centering on the lack o healthcare insurance.

Willis JensenW.L. Gore & Associates

Flagsta, Arizona

Im the primary statistician or aportion o a medical devicemanuacturing company.

Nancy WangMDS Pharma Services

Lincoln, Nebraska

[I] use statistics in thedesign and analysis o phase I

and IIa clinical trials.

Kristi EpkeSan Domenico School

San Anselmo, Caliornia

I am a high-schoolmathematics teacher, not a

statistician, but I teachAP Statistics.


31/84

IOWA

MISSOURI

ILLINOIS

WISCONSIN

MICHIGAN

INDIANA

WEST

VIRGINIA

NEW YORK

ARKANSAS

LOUISIANA

TENNESSEE

MISSISSIPPI ALABAMA

FLORIDA

GEORGIA

SOUTH

CAROLINA

NORTHCAROLINA

VIRGINIA

OHIO

PENNSYLVANIANEW JER

DE

MARYL

M

VERMONT

WA

MINNESOTA

KENTUCKY

Cindy AmmonsChapel Hill-Carrboro SchoolsChapel Hill, North Carolina

I teach AP Statistics tohigh-school students.

And What AreThey Doing?

Luke DavulisMaine Department o LaborAugusta, Maine

I work at gathering the datathat orm estimates o

employment and wages byoccupation.

Liam OBrienColby College

Waterville, Maine

I teach statistics toundergraduates and research

statistical applications inpublic health.

James HardinUniversity o South CarolinaColumbia, South Carolina

I use statistics to analyzedata collected in various

health studies. Malay GhoshUniversity o Florida

Gainesville, Florida

I teach statistical theory and doresearch in both statistical theo

Documents

Sept 08 Am Stat News