William H. Kruskal, Mentor and Friend

Statistic al Scienc e 2007, V ol. 22, No. 2, 277– 278 DOI: 10.1214 /0883423 06000000411 Main article DO I: 10.1214/0883 42306000000420 c  Institute of Mathematical Statisti cs , 2007 William H. Krusk al, Mento r and F riend Judith M. T anur T ow ard the end of 1963 I w as in terview ed b y Da vid Sills who was Editor in Ch ief of the International Encyclop e dia of the So cial Scienc es , then in prepa- ration. The job op ening w as for a staff editor to assist Bill Krusk al in his w ork as asso ciate editor for statistics for the Encyclop e dia . I later learned (by judicious sno oping in to p ers on n el files) that after the interview Da vid informed Bill that “Mrs. T a- n ur is yo ung and pliable.” It is alw a ys interesting to explicate the exact stimulus that giv es rise to a resp onse—in this case I strongly susp ect that it w as m y reply to a particular question p osed during the in terview that caused Da vid to mak e that judg- men t (though I w as indeed quite yo ung at the time). Da vid, having had to referee ent irely to o many in- tellectual battles b et we en asso ciate editors and staff editors, had ask ed me if I w ould b e willing to tak e guidance from Bill. My genuine sho c k at the th ough t that I migh t not b e willing to—I thin k I stammered something ab out his b eing William K rusk al, one of the originators of the f amous Krusk al–W allis test and me b eing a recen t MA in mathematical statis- tics and so of course I’d b e willing to tak e guidance— probably con tribu ted h ea v ily to m y b eing h ired. Ho w luc ky I w as to b e h ired! The job in tro duced me to man y of the greats in conte mp orary statistics and laun c h ed me into a career of editing and expli- cating statistics. Most imp ortantly , it ga ve me my first c hance to w ork with Bill. What a role m o del! In his wo rk on the E ncyclop e dia , Bill cared ab out ev erything—and I mean eve rything—fr om w eigh t y issues of con tent and exp osition to tin y issu es of t y- p ography: was there really a t w o-p oint space b e- t w een a symb ol and its su b script? I took a self- taugh t crash course in the printing of m athematics, Judith M. T anur is Distinguishe d T e aching Pr ofessor Emerita, Dep artment of So ciolo gy, State University of New Y ork, Stony Br o ok, New Y ork 11794-4356, USA e-mail: jtanur@notes.c c.sun ysb.e du . This is an electronic reprint of the orig ina l article published b y the Institute of Mathematical Statistics in Statistic al Scienc e , 2007, V ol. 22, No. 2 , 277– 2 78 . This reprint differs fr om the o riginal in pagina tion and t yp ogr aphic detail. but learned wh atev er I know of s u bstant iv e statis- tics by w orking with Bill and the contributors as w e edited, aske d for r ewriting and edited again. Wh en w e ask ed for a rewrite, often Bill w ould do a sample to show the con tribu tor what he had in mind , but h e w as alw a ys v ery careful not to tak e o ver the article, including in eac h suc h mailing the disclaimer th at he didn’t wan t to put words in to the contributor’s p en. Nev ertheless, many of the c ont ribu tors found it most conv enien t to just adopt Bill’s sample rewrit- ing. Th us , m uc h of the material b oth in the Int erna- tional Encyclop e dia of the So cial Scienc es and in its offsho ot, the Internationa l Encyclop e dia of Statistics that Bill and I edited, came from Bill’s p en. Indeed, m uc h more than has ev er b een ac knowledged. More broadly , the co ve rage of statistics in the en- cyclop edias represen ts Bill’s conceptualization of the field and its ramifications. W ell b efore I joined the staff he had b een the leader of the p ro ject that de- cided wh at article s to include, and the decisions we re indeed encyclop edic. Right next to articles on Esti- mation (separate articles for Poin t Estimation and Confidence In terv als and R egions) there are articles on Err ors (separate articles on Nonsampling Errors and on The Effects of Errors in Statistical Assump - tions). As well as sev eral articles on T ime Series and on Index Num b ers, there are sev eral on Non- parametric Statistics and on Multiv ariate Statistic s. The system of grouping articles together imp licit in these ti tles and an ela b orate syste m of cross- referencing make s the mappin g of the field b oth a fascinating glimpse in to Bill’s o wn m en tal represen- tation of statistics wr it large and a usefu l guide to a stu d en t tryin g to connect parts of the field. I w as luc ky enough to b e the first suc h student to ha v e as an assignmen t reading these encyclopedia articles— and reading them in revision, and reading them after cop y editing, and reading them in galley , and reading them in page p ro ofs! Bill’s consideration for m e while we w ere in d a y to d a y c ont act w orking on the Encyclop e dia w as enormous. In p rofessional matters, b ecause I was so “y ou n g and pliable” and in deed so muc h ju nior, Bill could easily ha ve u sed me as a glorified secretary , but he insisted on treating me as a full p artner in 1 2 J. M. T ANUR the en terp rise, often asking m y advice and consen t at times wh en it would h a ve b een m uch easier to just go ahead with his own judgment. In addition, b ecause I had to tak e r esp onsibilities that I often felt were b ey ond m y abilities, I learned b oth statis- tics and diplomacy . Bill’s consideration for me also extended to p er- sonal matters: Let me illustrate with a story ab out a min or p oin t instead of a more d ramatic one. I r e- mem b er Bill’s visiting my little cubb yhole of an of- fice at F ree Press and asking if it would b e all righ t to smok e one of the little cigars he fa v ored at that time. I said, “Of cours e,” and hoping to put h im at his ease somewhat m ore, added that m y h usb and also smok ed cigars. “Y es,” said Bill, “but that’s wh y y ou get out of the house to work, isn’t it?” Bill and I contin ued to w ork together after the In- ternational Encyclop e dia of the So cial Scie nc es ap- p eared. W e s p ent five y ears negotiating a con tract with F ree Press that would p ermit us to pu blish the statistics articles from that Encyclop e dia as a sepa- rate International Encyclop e dia of Statistics . Then w e wo rked an other fiv e y ears to up date and cor- rect the articles and solicit sev eral new articles to fill holes in the original so cial science encyclop ed ia. The con tract took so long to negotia te b ecause Bil l’s p erfectionism had rubb ed off on m e: toget her w e in- sisted on havi ng con trol of ev ery step of the pro duc- tion pro cess. T h us we work ed on th e system of sig- naling wh at mate rial w as new, on c hec king the bibli- ograph y , and sup ervised and en d ed up constructing the index ourselves—pro ving to Bill once and for all that ev er y v olume can ha ve a fine index. Man y of us learned that lesson fr om Bill; the vo lumes w e write and edit are the richer for it and our readers o we a debt of gratitude to Bill f or their conv enience in using su c h volumes. It wa s at Bill’s su ggestion that F red Mosteller in- vited me to b ecome p art of the team that pro du ced Statistics: A Guide to the Unknown —another won- derful learning exp erience for me. Again I learned more ab out statistics, learned more ab out exp osi- tory skills and gained more adm iration for my co- editors. There w ere those who said th at a readable v olume ab ou t statistics for the la y p ublic couldn’t b e d one. Gentl e p ersu asion of authors, goo d exam- ples from their o wn p ens and extreme p ersistence, esp ecially from F r ed and Bill, made it happ en. An d the vol ume has a w onderfu l index. A new fourth edi- tion, edited b y a completely new and y ounger team, has ju st come out. I did not b ecome a m emb er of the C ommittee on National Statistic s until some time after Bill had serv ed as the Committee’s first c h airm an , bu t I knew of Bill’s dedication to th e in tegrit y of federal statis- tics thr ough articles we had edited and thr ough copies of h is corresp ondence that he had forwarded to me o v er the years. In some sense that exp osure had b een Bill’s wa y of giving m e some basic t raining in the imp ortance of federal statistics an d th e work- ings of federal statistica l age ncies. My service on the Committee serv ed as a p ostgraduate course in these matters. I ndeed, us ing service on the Committee— either by mem b ers hip or on staff—to edu cate neo- phyt es ab out the imp ortance of federal statistics w as an imp ortant part of Bill’s agenda. This is y et an- other mo d e of Bill’s teac hin g and one that contin- ues to b e emulated by the Committee on National Statistics. Those of us w ho knew and lo v ed Bill we re on his mailing lists, and so w ere our friends a nd colle agues, and so w ere th eir friends and colleag ues. . . . Because he was a true p olymath, Bill’s reading was br oad and eclecti c, and h e w as able to mak e surpr ising co nnec- tions b et we en and among p eople and ideas. So, as w e all know, when he found something that might in terest one or more of his p en p als, he photo copied and mailed. Over the y ears I was am used, edified, de- ligh ted and a wed b y this b readth of inte rests. Those mailings stopp ed some years ago—a nd I h a v e missed them. I h a v e often thought how d eligh ted Bill w ould ha v e b een if he h ad b een y oung enough to tak e ad- v an tage of the electronic wonders of scanning and e- mail. Then, how ev er, our elec tronic mailb oxes w ould ha v e b een as ov erwh elmed as our p ap er fi les indeed w ere b y the volume of Bill’s corresp ond ence. Lik e our files, our minds are fuller an d b etter fu rnished for having had Bill Kru sk al in our liv es.