Wednesday, April 3, 2013

Woman of Statistics, Trailblazer: Gertrude Mary Cox

By Helena Hoen, Apr 3, 2013

Gertrude Mary Cox was an influential statistician in the 20th century, instrumental in developing the field of statistics.   Cox was my maternal grandmother’s sister and me and my family used to spend holidays with her.  She died of leukemia when I was young, though, and I became curious about her professional and personal life.  Here is some of what I learned after many hours in the North Carolina University Library Archives and interviewing willing friends and colleagues of hers in the Raleigh area.

SAS Jmp, a statistical software company in Cary, North Carolina, is doing a tribute to statisticians through history in recognition of the Year of Statistics, 2013.  That prompted me to share some information about Gertrude, including a shorter version of this post.  Please see the SAS Jmp link at the bottom for more posts on statisticians.  

Impact of Gertrude Cox’s Work
(Jan 13, 1900 - Oct 17, 1978)  Cox was a statistician and consultant with strong administrative skills and the ability to inspire. She used these strengths to build a statistical empire and promote the use of statistics, locally in North Carolina and then nationally and internationally.  Prior to R.A. Fisher’s ideas published in the 1920s, researchers relied on their individual instincts to design and analyze their studies. It was difficult to compare results across studies and draw conclusions.  With researchers using Fisher’s objective ideas, the nature of scientific research was about to change, to bring about more reliable conclusions, rigor, and repeatability.

A little help spreading the word gave momentum to the revolution. “I think that [Gertrude Cox] realized earlier and more keenly than almost all of us how useful statistics and the statistical point of view can be in human affairs,” wrote William Cochran in Biometrics in 1979. In an article in the International Statistical Review, Cochran said, “I doubt if anyone contributed more than Gertrude Cox to building up the profession of statistics as we know it today.”

Cox founded the Department of Statistics at North Carolina State College in 1940, served as first director of Research Triangle Institute’s Statistics Research Division, and consulted worldwide on research projects and to help others develop statistics programs.  She co-authored the text Experimental Design with William Cochran, and was the founding editor of Biometrics, a journal of the International Biometric Society.  She served as president of the American Statistical Association, Treasurer of the International Statistical Institute and on convention committees, and was recognized with numerous honors including an honorary doctorate from Iowa State University and buildings in her name at both North Carolina State University and the Research Triangle Institute.

Stumbling into Statistics
Gertrude Cox initially pursued a career as housemother of an orphanage for young boys in the Midwest.  She had become a Methodist Deaconess and then worked at the orphanage. To become housemother, she needed a bachelor’s degree, so she went to Iowa State College and chose to major in math, because, she said in her diary, it was easy.

At Iowa State, George Snedecor was a professor in the Mathematics Department, but recognizing the utility of statistics, had taken the step of offering a statistics course.   He managed to bring Fisher to Iowa State for a visit, which rallied excitement for his idea of providing training in statistics in the United States.  Cox stayed at Iowa State after earning a Bachelor’s in Math and became the first person to earn an MS in Statistics there. Then in 1933, after a shortened stint at University of California, Berkeley, to begin a PhD program in psychology, Cox answered Snedecor’s call to work as a consultant and manage his Iowa State Statistics Lab of “computers” for the statistical consulting projects.  Computers, as they referred to them, were the women hired to compute sums of squares and standard deviations on big, cumbersome Merchant and Monroe desk calculators.  Because of their attention to detail, women were often sought for this task.

Cox simultaneously worked on a PhD in statistics, managed the computing group for Snedecor’s Stats Lab, served as a statistical consultant, and taught statistics courses, including graduate-level courses in experimental design at Iowa State in the 1930s.

It was there at Iowa State where she first became a colleague of William Cochran. Cochran and Gertrude Cox later wrote the textbook Experimental Designs together, first published in 1950.  It was a seminal book that enabled many researchers to apply design theory to their studies, and it is still in print today. Cochran also later served on Cox’s statistics faculty at the Statistical Institute, a combined program across three North Carolina universities encompassing both applied and theoretical focuses.

Arriving in North Carolina
In her own recollections of the North Carolina State statistics programs, Cox credits Frank Graham, President of the North Carolina University system with the idea of developing a statistics program, originating from a chance meeting on a Seaboard train in early 1940.  Graham ran into WF Callander of the U.S. Department of Agriculture, who said he wished to establish a statistics program in the Southeast similar to the one in Ames, Iowa.  Graham said he would do it in a North Carolina university and proceeded to gain the approvals and financial support needed for this venture, in part to help with the “war need”, World War II.

Snedecor at Iowa State was asked for recommendations.  Gertrude saw the letter where he had recommended five men and asked him why he didn’t recommend her.  Then, in a letter dated Sep 7, 1940, Snedecor “half-heartedly added, ‘if you would consider a woman, I know of no one better qualified than Gertrude M. Cox.’ “  Against the advice that they would never listen to a woman in the South, as an understatement, it was surprising and exciting when she later told everyone that she had accepted the position.  On Nov 1, 1940, she reported for work to develop the Department of Experimental Statistics in the School of Agriculture at North Carolina State College, probably the first independent statistics department at a university.  The assigned mission was to provide courses, statistical consultation, and computing services for the research staff at NC Agricultural Experiment Station and the School of Agriculture.  She did so and then went beyond the scope of the original mission to develop a shared university statistical institute with graduate degrees in applied statistics, theoretical statistics, and biostatistics at NC State in Raleigh, Chapel Hill, and Greensboro.

Building the Department of Experimental Statistics at North Carolina State College
Cox spoke and wrote plainly, had large visions, and went about them in a practical, steadfast manner.
She had a talent for knowing when she had the right person, sizing up the candidate and the job that needed to be done and matching them.   About 90% of the time, she was right.  Pretty good hit ratio, agreement is shown by the fact that she and four of her faculty were elected into the National Academy of Sciences:  Raj Chandra Bose, William Cochran, Gertrude Cox, Wassily Hoeffding, and Harold Hotelling.  Cox would say, “Get good people and give them what they need.”   Cochran received a phone call from her when she was making plans for the Statistical Institute asking who he thought the two best theoretical statisticians were.  He gave his opinion but stated he didn’t see how it was relevant.  She replied, “That part is my business.” 

For the experimental statistics group at NC State, she hired people with expertise in the areas of application so they could better work with the scientists they were providing statistical consultation for.  The statistics degree program required a strong emphasis in applications such as biology, physics, or chemistry. 

Cox was successful at procuring funds.  To support teaching all the statistics courses for the different applied disciplines in the Statistics Department, she arranged committed monetary support from the different departments on campus, a rarity.  She obtained funding through the Rockefeller and Ford foundations.  It was rumored that one foundation told her that they would go ahead and fund her or else they’d keep receiving these letters from her asking for still more.  She pulled in funds through outside consulting jobs with government agencies and private businesses, but paid all the faculty on the same scale.  Equitable pay lead to a cohesiveness in the faculty rather than resentment that could have resulted if those who did outside consulting had been paid more because those services brought in more funds.   

Cox had a strong sense of mentoring with faulty and with students.  She would tell people, “if I were you, I would …”   She encouraged her faculty to get PhDs, and made faculty and students aware of opportunities.  She herself had benefitted from Snedecor’s mentoring.  Additionally, she promoted professional development by arranging statistical short courses and seminars.  These statistical gatherings also served to network statisticians, where they shared and debated ideas.

Cox was flexible.  When she approached NCSU about adding theoretical statistics to the program, they said it would be a duplication of effort and thus it needed to be at a separate NC university.  It was not how she had wanted the program at NCSU to evolve, but she accommodated by forming the Statistical Institute, and procured funding needed to offer Harold Hotelling a worthwhile salary to bring him in to head the theoretical program of mathematical statistics which was to be at NC Chapel Hill, a role he filled from 1946 until he died in 1973.

In his congratulatory letter to Cox when the 1970 dedication of Cox Hall at NC State was in the planning stage, Snedecor joked about Cox Hall being bigger than the Snedecor building on the Iowa campus named after him, with hand drawings of the two buildings.  Indeed, Snedecor brought statistics to the United States, his building was first, but Gertrude made the field of statistics bigger, and her building was bigger.

Retirement, RTI, and International Consulting
When Cox retired from North Carolina State University in 1960, the university was only beginning to consider providing employee pensions.  She mentioned to a friend that they only offered a small one. However, she had contacts outside of NC State she could find work with, or rather they found her.  She helped organize and served as the first director of the Statistics Research Division at Research Triangle Institute.  She retired from RTI in 1965.  In her five years there, the statistics program became a strong research organization. In a tribute to her contributions, RTI named a building after her dedicated on 9/9/99.  They picked 9/9/99 because it’s easy to remember and numerically aesthetic, as statisticians would think, numbers are beautiful.

Cox’s influence was truly global.  In a publication in Science in Context in 2009, Patti Hunter detailed how Cox lived in Cairo for a year’s time in 1964 and ’65 to help the Egyptians develop the Institute of Statistical Studies and Research.  Throughout her career including post RTI retirement, Cox consulted internationally.  She traveled to Thailand to provide technical and administrative assistance to the Kasetsart (Agricultural) University and to serve as a statistical education adviser to the government of Thailand. She worked in Hawaii with the sugar institute and pineapple growers, in Guatemala for the World Health Organization on childhood nutrition, in El Salvador and Costa Rica for the US Government, in Honduras for United Fruit, and in South Africa on sampling gold.

She especially loved working in developing countries. She liked being around the people that lived and worked there, and they liked and admired her.  Cox attended international statistical societies’ meetings in Asia, Europe, Australia, North America and South America.  She was in the field of statistics at a very exciting time and added her own infectious enthusiasm.  And she stayed active throughout her life.  In one of her last Christmas letters, she wrote, “All is fine, I’m just a little slower.”

Commentary on Cox as a Female Professional
Much of what Gertrude Cox achieved was a first, first for anyone, a man or a woman.  Furthermore, the fact that she was a woman, it was the 1930s-1970s, and mostly in the South made her accomplishments that much more impressive.  So how did she do that?  My impression was, whether anyone else did or not, she wasn’t going to see herself being a woman as a problem.  She was still aware, though, that influencing administrators, funders, and fellow statisticians sometimes took extra persuasiveness for a woman.  One colleague said there were times that he was uncomfortable because Gertrude had requested he go to a meeting for her to present her point of view, which differed from his, because the group would be more likely to listen to a man.  

An often repeating theme when I interviewed some of her friends and colleagues was that she was not a “women’s libber.”  I think more accurate statements would be that she did not see men as the enemy, and she did not think complaining would help.  What she did was recognize everyone’s strengths and encourage all, including women, to pursue knowledge and also choose whatever roles suited them.  For instance, she would defend her choice of who to award graduate stipends to in the statistics department based on ability while disregarding gender.  Additionally, her mere existence was a role model for career minded women.  At the same time, she valued the role that women played as wives and mothers.  This was evident in the way that she discussed shared interests and spent time with female co-workers and wives of faculty members.  Regarding women balancing both work and family, she spoke at a local church about a child care program they were developing.  She recognized in her speech that not every one’s situations and choices are the same, and thus there was a legitimate need for childcare for working mothers and single parents.

When I look at old news articles, in some of them I sense an irrelevant slant.  Local articles showed Cox arranging flowers or standing next to her doll collection, wrote about her hobbies a little too much, and commented on her physique.  I believe her work as a statistician to be far more worthy of highlighting.

Romance in Cox’s life turned out to be an elusive topic, but some of her friends and colleagues shared what little they knew.  Sometimes they said she didn’t think it would be fair to a man if she were to marry.  If she did, she would quit her job to tend to the husband and raising a family.  My favorite answer comes from a friend of hers during her second retirement, when she was still a statistical consultant keeping a pretty aggressive travel schedule.  Men had their own agenda, and Gertrude had hers, and she probably thought that theirs was pretty boring.  It would be hard to keep up with her.  Cox didn’t write about any romantic interests in her diaries.  Her diaries were mostly recordings of what she did and not thoughts or feelings.  Her thoughts were, however, in letters.  She was a prolific letter writer, often writing several a week.  From these I get the impression that male companionship would have been welcomed if there were a way, and the right person.   She was very social, liked entertaining at her house, and enjoyed having house guests.   Still, she seemed to express, it’s nice to have good company to share your life with.

The Author
I have gotten to know Aunt Gertrude much better through her own records and from speaking with those who knew her best.  It gives me a great appreciation for her and her friends, colleagues, and family.  I am a biostatistician, myself, applying statistics to medical research with investigators at Providence Health in Portland, Oregon.  

Thanks for support and contributions on the Cox history project:  Nancy Hall, University of Delaware;  Patti Hunter, Westmont College;  Catherine Moyers, Chicago, Illinois;  Josh Horstman, Nested Loop Consulting;  George Eckert, Indiana University;  Annie Zangi and Arati Mejdal, SAS Jmp, Carey, North Carolina; Edith Sylla, North Carolina State University;  Morice Toler, North Carolina State University; Bradley Wells, Mark Espeland, and Robert Byington, Wake Forest University;  Robert Monroe, Sarah and Phillip Carroll, Julie McVay, Jack and Vi Rigney, Al and Betty Finkner, and Larry Nelson, Raleigh North Carolina;  Klaus Hinkleman, Virginia Polytechnic Institute and State University; Jan Hoen, Margaret Harrach, Myra and Klaas Hoen; last but not least, Patrick and Liam Hoen.  Thank you, Annie Zangi, Arati Mejdal, and Michael Crotty at SAS Jmp for the opportunity to post a few words on Gertrude.

1.  Anderson RL, Monroe R, Nelson L.  Gertrude Cox – a modern pioneer in statistics.  Biometrics.  35(Mar 1979): 3-7.
2.  Cochran WG.  Some reflections.  Biometrics.  35(Mar 1979): 1-2.
Cochran WG.  Gertrude Mary Cox, 1900-1978.  International Statistical Review.  47(1) (Apr  1979), pp.97-98.
3.  Hunter P.  Gertrude Cox in Egypt: A Case Study in Science Patronage and International Statistics Education during the Cold War.  Science in Context.  March 2009.  22(1):47-83.
4.  Monroe RJ, McVay FE.  Gertrude Mary Cox, 1900-1978.  The American Statistician.  Feb 1980.  34(1): 48.
5.  North Carolina State University Library Archives, Gertrude Cox collection.  Including:
a.  Cox GM.  The Institute of Statistics, Raleigh, NC, October 1940-December 1960, Chronology and reminiscence.  June 19, 1976, Rollins College, Florida.
                b.  Cox GM.  Diaries.
                c.  Cox GM.  Letters.
                d.  Cox GM.  Log of Foreign Countries Visited.
6.  Interviews with  Larry Nelson, Robert Monroe, and Jack Rigney conducted by Helena Hoen, 1995-1996.
7.  Cox GM.  Correspondence with family members.
8.  Brook EB.  Tales of statisticians, George W Snedecor.  2001.   

Other links on Gertrude
1.  SAS Jmp blog “A Tribute to Statisticians”:
2.  NC State University Archives:
4.  A chronology of mathematicians: