Miracles and Statistics: The Casual Assumption of Independence Author(s): William Kruskal Source: Journal of the American Statistical

Association, Vol. 83, No. 404 (Dec., 1988), pp. 929940 Published by: American Statistical Association Stable URL: http://www.jstor.org/stable/2290117 . Accessed: 21/02/2011 08:55
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at . http://www.jstor.org/action/showPublisher?publisherCode=astata. . Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Statistical Association is collaborating with JSTOR to digitize, preserve and extend access to Journal of the American Statistical Association.


progenitor of themodern computer. DMS 8404941. society. identified such. and and. careless or observation. and critical. Healy. to In Babbagehad views 2. in pecially secondary schools. work ticalaspects. moregenerally. so forth. Earl of Bridgewater.do notpropose that of for of and I certainly and be farmoreimportant theoriginal than that thebroadtheme religion science. uses One also finds fatuous superficial ofmathemator 0. University andthemathematical of Chicago. with on to history. author The thanks following. results poor. at treme outliers nonscientific It is widely of life. I.Religion. 71). and false If as prejudiced. and Nor.Miracles and Statistics: The Casual Assumption of Independence WILLIAM KRUSKAL* The primary theme thisaddress cautionary: of is Statistical independence fartoo often is assumed casually. Jr.Thisis a revision theauthor's IL of 1982 Thereare many suchconnections without statisexplicit address theAmerican to presidential Statistical Association.. second.Leslie Kish. religious import a miracle it than maybe farmoreimportant thedailylifein which are occurred. I. DonaldFiske.William Fairley. circularity argument. 1. Duncan. A.83. W. I have a moral. Offenbacher. S. William Christian.Krusdesirable attitudes thedomain ethics. D. Now narrow focus connections our to between religion * William of in a sciences. Judith Tanur. After initial discussion statistics religion. Safety. Some willbe surprised so argued a I further treating outliers investigation themechanism outlying normative statement. Arnold whotakemiracles given an example. Goodand nessof God.Joseph Gastwirth. and Zellner. M. in first. Stress is given often the tacit unexamined or assumption independence of among witnesses a putative of miracle. Edward David Owen.RonaldThisted.) (1982)expresses strongly: religious that "new forms I begin with of in Babbage'streatment miracles hisown Gilkey appearand reappear of and becauseof a scientific. ical terminology. R.the discovery penicillium often will exhibit of is intotheology.Benjamin SamuelKotz. Grier. ThenI discuss miracles testimony technological and a credible of system symbols giving structure. J.Presidential Address December1988. of the as signsof revelation. to especially Hume'scritique Babbage's and reply.and religion" 9).404. Kruskal Professor.in hiswillhe left 8. R. out thatof Hume's famous context. 929 . Chicago. R. as manifested Creation. esDo notmultiply lightly. often assumption independence tacit of among anddirection nature. Testimony.Hume. Zabell. for Hutcheson in by BrianGerrish. theaddress are contemporary assumptions independence: nuclear reactor include safety. M. Joel Cohen. and Other topics prayer. may study not aggressive incompetence dipping by deeply led to the outlier.Again. INTRODUCTION Francis HenryEgerton. in of polkal 1960). J. Kruskal. example. see Anscombe and over 1978.2 of itics. a Charles Babbage. in (A revised edition appeared 1838. No. Francis M. Grant.Neal Koblitz.wardapplying provided Keith George Barnard.to theparticular sharpanxieties-andeventerwitnesses. theself. R.to support are (p. by to calculations recent attempts (Brams1980)toand DMS 8601732. E. Brown Good.g. that fail and A Fragment. eight clerical scientific and notables each wrote and published treatise. without serious concern howcommon dependence howdifficultcan be to achieve for is and it independence related (or structures).I think a rangefrom This of and astronomical was supported National ScienceFoundation Grants SES 8303838. viewsin partstatistical. ? 1988American Association Statistical of Statistical Association Journal theAmerican Vol.Stephen Stigler. thatat leastone of far and themincorrectly arguedthatthepursuits scienceare of unfavorable religion. thetension of and between most itprovisionally. Antony Flew. are in to of debate. meaning. Norma E. great the Frank physicist (1956)wrote One analogy that is miracles likeapparent are outliers that that of howall scientific theories related "their are to fitness we study worry (e. Holland. stories.000poundsforthewriting and publication works "thePower. is of Department Statistics. thendiscuss I overcasual assumptions indeof age" example in contexts. Gilkey's and article about is Miracles relevant statistics a number ways. address of and the turns miracles. second. King. Farebrother. addition. RELIGION AND STATISTICS about miracles. Martin Marty. thedemand to for for mony miracles. 1724: "Benevolence directly theMoment Good of is as J.there misleading In apparent the of outliers misleading miracles. A current If it pendence other ofintersectionthedebateaboutteaching is evolution. in 1837 he So and generally twobasicrealms are of published whathe calledTheNinth Treatise: Religion science Bridgewater human effort can hardly to interact intersect. of on Wisdom. repeated measurements. part science general." in Undereminent auspices.Paul W. B.. specialemphasis theubiquitous. Savage. L.EugeneSeneta. thesameway. One might evensay thatmiracles theex. little attention these outliers miracles worth and are except of fortheir connections withthe psychological vagaries of on or the human observation with effects behavior fortuitous coincidences. of skepticism abouttestimony thepragmatic to accept and need KEY WORDS: Babbage. the who calendar criticisms suggestions: Baker. gametheory biblical to manBradburn. that eight felt the treatises not did carry their argument enough. M. B. Wright. Other of contexts and endswith casual of multiple testimony treated. is this: rors-of a technological (p. as Just forthose so. argument testiagainst culture response. Stephen Brush.' in died 1829. Stephen Portnoy. M. 60637. Langdon Gilkey. Kadane.

especially and (1974) for fathers).] Pearson's(1924Nightingale. he compared of There of with and brilliant interpre-family clergymen thelongevity others. andproposed randomized as an approach showing highly trials relevant 3.doneby "divine bullock.that and religion statistics. Suchuses have been roundly thandid for moreBritish prayers good health Craig. writes 1749 tween in son1931. generated defended useful as metaphors.dall(1871)andHenry Thompson.5 recent motivations connections.for of effects faith various on ature thecurative (1988) and Bartholomew (1984.His first ences(see Koblitz1981. put definition pened. MIRACLES of I digress discussstatistically to shapedstudies the Thereis a huge themselves. theargument apparently over contemmany results thatscandalized ofmathematical continues bitter in argument out with expressions negative as for as Academy Sci. pp. and a gives thorough and Weldon(1965) was double-blind by fear is A different theme theancient religious ofbeing of prayer Joyce 1969)gave study (Cullipp Another recent in in Thisappears several andinconclusive." morerecent A speaksof a in the powerbywhich opof wateroverit to makethetaskdramatically harder. 2.S. its description see (1986). 1872.[Galton's publication slightly stability statistical ratios"(p. modified.several chapters thelec. recent a and of obvious problems control compara recasting. 115). calledon Baal to start fire. tation Stigler by (1986) provides sympathetic and but superficial use ability.p. DavidHartley the of studies prayer. Pearson As however.-B = MIA" (Thomp.Without to returned thetopicin print. of perhaps much themethod theconcluovermembership theU.Haldane (1928. In another only proposed but mo.Recall that characterizationsa miracle. of if miracles. poured served nature.ThenElijahhad a bullock on hisaltar. themathematical of discussion of to a generally probability statistics. in book(see Galton the Ifwe specialize from sciences heincluded material his1883 further. Tyndall Yet the a sensemiracles. I have been able onlyto dip us for in interest evidence with prayers-arein intoit selectively a primary arguethat prayers-atanyrateanswered evidence. Madansky is Of questions. by may is of the priests Baal builtan altar.poraries. an overview entry 4. i. 36). (see wrote-and many have made thepoint-thatno Begin with DavidHume's1748definition Hume1975) others to probabilistic "could call one showerfrom heavenor deflect becauseHumewas one ofthefirst6 apply prayer argument and a towards a single beam of thesun" without "distur. 6) for many priests Baal (1 Kings 18).number devel. Indeed. see and bibliography. they Pearson(1978) carried sequenceoftheologically the one tivated Florence and whereGaltoncarried out. are nowto miracles We turn per efficacy prayer. 14-15).Marshall 1986). 286).p. becauseHume'sfamous us outin of of against is banceof natural as miracles framed terms one probability law. 1876). where = fearand L = love of God. The book also had a moregeneral and and intervention theposDe (1924)said. earliest knowwas bythegreat and W = loveoftheworld (see Stephen probably clergymen and family British the criticized also sonedthat royal and 57-58). thecase ofJohn In of of longevity members theroyal so whose1699calculation thedateoftheSecondComing others.p.34 the of ofthematter.930 Associafion. pp.of course. came In anycase. Original Designof theDeity. 2. it.1883). recorded trial sionofa lawofnature a particular empirical comparative Perhapsthefirst by of agent"(Hume between of prayer was the competition Elijah and the or by the interposition some invisible Hume variant of the (1962.although objectivity1975. another. 404). National in on publication thetopicwas in 1872. are.ideasto miracles.e.Again. often small especially validity testimonial (1871) miracles. Portnoy leavesme with general in senses.Vol. F Galton.1930)massive add we meaning might of cis Galton"(p.He reaFrancis I 1962. Brush of withtheological children clergymen their Bayes and RichardPrice. some literature miracles.put on it a sacrificed Aquinashad earliersaid thata miracle something obthe commonly and the beyond order agency Nothing hap. For on and and theybring near the topicof miracles. "he sibility ascertaining In thesecondedition. but to moreencouraging prayer.See Pomeroy of of itsreporting be questioned some. coursethere a longlitersee of and treatments religion statistics. has been frequently derided. PRAYER . Frank(1978). Moivre's whatGaltoncalled theocratic opments cometomind. recent empirical relatively Among thatW = F2IL. of See At leastone scholar (Dietl1968)hasbuilt this on example suspended. weighing of volition theDeity. added studies statisticians through (see the of describes study biography Galton that "with modified a almost Fran. of These studies interesting se. calledon hisdeity.1988). was determining frequency irregularities the thetheometric of from the of of the this grasping side ton-it offended sensibilities a number readers. for counted. Galtonpushedahead nonetheless. and "striking interposition divine are courseof nature overruled. is impossible understand history I believethathe never it to excontemporary and from andSiissmilch The Galton(1872) article other Derham statistics De Moivre through Debate (Means in in of were to Quetelet. response overwhelmingly erations theordinary The was fiery. of workon the central was limit theorem theological: because-said Galwas material omitted.are suffused evaluation medical A discussion.and sions. reprinted ThePrayer-Gauge principle the changes culminating the modern by was preceded Tynof place. 243-245)prosubstituting account(wrongly distorted tures those dealing Thomas vides a slightly with (Pearson 1978). 115-117).pp. For results partsof the Old Testament.December 1988 of Journal the Amercan Statistical berelationship happen-yetanother miracles really andinversely their as abilities. 346) fora or to treatment." Flew (1967. pp. Vol. quiteas serious thestoppage as a Humedefined miracle "a transgresan eclipse"(p. example a census.

" "All oflifeis thatforH.handof 12 hearts 13 is. (1975. eventhas been a it it to even supposing amounted a proof." R R of a in from scientists The God of Science. into airis a visible the of The raising a houseorship for . it seemsnot torical events" 224). have nothing do but did times. skeptical. I also be. propensity lie or be dupedis presumably the is by that allegedmiracle in fact an eventhappensthanif it likely if a "invisible an appropriate agent"-presumably benign verydifferent a highly more much is.than Swinburne 1968.Thefamiliar statementmadebya Barthian one as that probability theunconditional overthe Eichrodt (1967)stresses theOld Tes.and so little has for on. is every of aboutthehappening thecentral question testimony a bridge of of For in indepen. theprobability reporting a miracle. thenPr(H I R) = 0 = Pr(H)." to or willed thedeity. Humegoeson to like Everynuanceof definitions the above has been forremarkable us informs of and the For and examined analyzed.that"Justas I evena liar.Yetitis hard know questions volition a deity other see context. is Thereseemsto be remarkably be He that full wit. 80-81). is an inward the of miracle God's mercy" 8). and 6 = 1 .For in may Another toshow is Thisformulation be criticized many the explicating ideaof"lawofnature.. was that in believed theunrecognizability and is R lieved in the necessity. to amounts an entire religions. .p. These are psychological historical questions is of Hume'sdiscussion testimonycynical. Prior(1955) says that"Faith . the is calculation towrite column-wise central in Protestant Prior's is imag.He says.p.conditional dialogue. To be moreaccurate. and make it a just foundationfor of requisite wants everso little a force the of a feather. . by example. to regard all popular with He appearance.Pr(H I R) = pOl(pO + jiG). this to One can try formalize argument several abouta reported hypotheses two of by comparing explanatory excludethe raising a feather might Otherauthors of signifi. Let p.= 1 . was mistaken. that"no testimony anykindof miracle ever p in various settings. from denttestimony twoor morewitnesses. For a discussion theOld Testament reallyhappened. a natural thecourseofNature is if trustworthy = (p morenaturally. inary if Even sum. instead to for enough include nizablemiracle. 48-49).bility thereport correct misinterpreted."l do level. concludes human of testimonyso subject thatcannot neglected.2.Eliade (1981) in. wouldbe op. counts a miracle" as itself (pp.0 "The greatest or manas miracles themselves.89 and and p. of the camouflageof the 'ex. prove a miracle. say an angel. familiar Bayesianstructure. Eichrodt and that did mean. (p.as opposedto a demonor devil(see doesnot.Kruskal: Miracles and Statistics 931 events10 (1975. in takesmiracles the"widest possible tament Even result. thewitness wholly 163). as in well and on Yet I avoidsuchangelic temptations hasten to Thusthevalueofp might be taken differentthe to reason expect there of tworows. raising annihilation. I surmise. .pp. this substraction. as reala miracle. parts thereporter oneself) (perhaps mayappearin themostordinary thatmiracles sists the consider possitreatment might of everyday life. is theexistence thelawsofnature. handof12hearts I think. that. .p. Note that p = p. 163). usually stressed some.Thisis as biga miracle havea standard x 2 table: These are some quoas the newsof the Resurrection.has been bothunderscored deniedmany of experience contrary. 127). but be with might tried.5). or explanation? a fuller For (But improbability. Moreimportant.A likelihood (p. valuesfor0 and numerical of discussion reasonable to bias. idea say thatexperience.the validity humantestimony informs of the shouldbe striking. (1975. 7-10). onlyexperience.) Supposethen take or by that miracles Therearealsoviewpoints simply life science thoseeveryday for miracle that0 is the a priori probability H.p.are two samplepoints. fruit 2 at looking a miniature tree.Hypothesis is that miracle nothappen might the significance H whatreligious or tricked. thata miracle When"thesetwokinds the constancy thelawsof nature.pp.we generally notset At a morefundamental and much to amounted a probability.1970.miracle. approach to Turn Hume'sargument. of color witha fullchorussinging in urrection bright to we are and Messiah.self-interest. to of by (1967. lessto a proof. Then we a miracle. and of the trans-historichis. and therefore . with Hume'sdefinitionthat An important problem ways.A fuller There but so ofmiracle. (the miracle reported) in an in ceptional' the banal. passion. when wind to anysuchsystem religion" regard with not though so sensible is that purpose.forexample. with (Trinklein interviews scientists H TOa pO7 0 pp. thechanging waterintowineor thelevisay insisting an eventmusthave clearreligious that H the see for anchorite.Hume himself notsee theneed fora striking substractthe one fromthe other .a bridge as one of13 reported 12. 127). 162. Let p be theprobability correct Whenyoulookat an appleseed.p tobe a function 0.to permit analysis. thenPr(H IR) = 1. one. Swinburne tation a holy aside is putting so in (1970. particular allegedly than muchgreater a as forms cen. the proof" posedbyanother . example. testimony. . Res.p. .= terms. of us.youarereally p = Pr(R I H) = Pr(R I H). 115) in ways. . in of reporting: of Or. invisible or agent. a Hypothesis is that miracle of cancetobe counted miracle. said: can . traltopicofthisexposition. how can thatbe a signor evidence rich after miracles all. of statement thispointsee a simple.R (it is not). to as likely be reported 13. changeovertime. is of 1).chap. keep in mind Smart have a priori (1964. repeatedly us and of the remarkable. . I havedifficulty theunrecog. 421). no human testimony The miracle. as a 4 reportinghandwith hearts one of5 (see Venn1962. the report correct.up such modelsuntilafter surprising bias is Thusthere inherent or of from weakness testimony reported at leastdiscussed. example. book7 tations H pO pao 0 aboutreligion 1971.

uncertainties maries: neither In manwas. this equivalent is of is of ability themiracle less thantheprobability a lie appeals by or mistake. 89-91). dependent behavein a wholly if the observers 13 2. BABBAGE'S forgets sometimes that is. fact. to Supposed Have Subsisted theChristian pp. whospeaktruth can witnesses be found. miracles. Midof of will treated in thework Dr.it chanceofmisobservation.pp." (Babbage1838. pft 1 the within very werelimited that these persons ten posing to pursuitable their of ten narrow range only fabrications ~P~0 that wouldall fix the posesofdeception: probability they is miracle mentioned 1/(10)1o'' (pp. probto 0 < -. requirement (Indeed.] But I do wantto describe exin dependence-nocollusion.)In thesameway too andmultiplies casually.presented event rather at a tailprobability decreasingly than more able events under one hypothesis. sc. the From Earliest in Church.andthat he that observers n independent agreein reporting it hapto withevery disposition deceive. It also assumes of the in is extreme. Youngmisread a on observers R will ofn independent produce conditional is He interesting. that the ones. p is edgessometimes. 122. 130-135) might-butI think not-have quotedthe skeptical made the point. reader find Which Are Powers Into "A entitled FreeInquiry theMiraculous dleton.] Price(1777. Ifp >75.it seemsto me that his n an howsmall. (or viewof huof and to experience.inR. without collusion. Hume bage-like and ofwavesofskepticism belief.932 Journal of the American Statistical Association. suptestify the same thing.theconditional to to thefact that saw a dead manraised life.but supposethatthereare several (1844." saysYoung(1844). did (1857. 66-69. answer Hume. also scoldedLaplace on of developing uncertainties to terms-however earlyand rough-ofprobability. "Suppose.Babbagesaid.pp.Chalmers of much was Hume'sargument metbya storm protest. it longpredates Christian see Grant the others Not surprisingly. gument thedifficultyestablishing absence colAt independently.p. to applies independent argument of . if independent thanfalsehood. independence. a time superadded theskepticism Humebythegreat to and in of La Place" (p. 2). given Indeed. up arbitrary aboutsetting thetwo-pointsameas fora single there something is as decrease n increases. thedegree credit be assigned have which dependon facts circumstances. improbability thefalsehood of whose conindependent of shall be greaterthan thatof the improbability the testimony curring miracleitself.but tenotherpersons pened.Indeed 203) [That Hume alongBab(1952) fora criticized"4 era.a first are and observers. p.Undersome sum. as Humeargued.. Chalmers about Christianity.p. advance more claimed second Young's close for probability H thatis arbitrarily to 1?" Wellof pointsout thatbothHume's and Babbage's treatments that senseprovided p > in that's course right an algebraic of initial reports mirin to miracle. that theunconditional is. dead therevived to similar those calculations Babbage wentthrough of makeslittle senseto look at theprobability a simple as and previously concluded follows: of prob. .andso forth.cusedhis"mathematical ofreasoning" itwas whatmiracle because meansthisnoise?/Fellow.. theopposite behave observers wayand lusion.g.I shallreturn Laplace.givethesame answer. pages114-116and 142-143. act in ofselection thevery ofmodel observer. of the witnesses. anyfixed > 0 no matter is. fact.[One account by Brown too extensive report to inassumes R. 1846. A little is. 198-199)madea smallerror calculation does 0 for that. upon the particular of discussion the is 0 this where assume < 0 < 1. Chalmers wordsof Shakespeare. somecases. largerthe audiencethe greater waternorwine. then probability363-365). skeptical"2 position notnew. an incorrect forinstance.Second Part.Muchearlier. conditional formulation. 80) is thatBabbage (1838.Chalmers exhad Gloucester "What gave a similar say who Babbage.. Isn'titobvious In affect mainconclusion.it is ALWAYS possible to assigna numberof frequently him.ineffect.. 418) briefly detailed history lines. p.is to Now. December 1988 is probability the the Similarly. 66). shows that calculation that that Pr(H IR) < . Still "the best fitted neutralize mischievous to at name case of Humeputtheskeptical with specialforce.Thisdevelopment an early we It of statistical problem selection. James Youngcarefully themathematician Babbage's (1984.whoagreein reporting miracle. is here. thata persontestifies thatall of probability H. Conyers the subject. that for observer 0 is less than youmaybe right a single claimby Young lations muchlikeBabbage's. Coins do land on their at observable magicshows. acles refer previously to unspecified one. to are and Price Chalmers careful require of REPLY I although do not see properdiscussions how strong 5. familiar as a function n is monotone increasing thelimit p requirement to that this Notethat assumes p is thesameforall observers tempt get aroundthe Hume-Babbage in weakness Young'sarthat But of coursetheprimary crucially the > of and nota function n. A complication that might samplespace. to Young'scalcuof the amining strength thatassumption. thewhole this to to of and of possibility thecollusion. e. hislanguage-without "David Hume. is also an atto of unity. dependent that in pp. thelike). 202Successive Several AgesThrough .. Centuries. The witnesses. style influence the claim?" (Henry theSixth.He argues anecdote. pp. act 2. anygiven under witnesses of the to collected become subject mathematical beensufficiently notyet bear on thispartof the which Some of thoseconsiderations inquiry.a phenomenon own have witnesses their in-between andmoreoften the the abouthow circumstances. 366-367. inp5. I interpret that observation = Pr(H IR) > Pr(H I R) Pr(incorrect I R) observation = Pr(correct I R). refer a specified than a is observation more probable that that correct is. before argument dost thoupro. explanation hiscynical was mannature. can find integer suchthatagreement we Babbageand thatthetwoagree.

for variables. in which notorious Collins done. eral or manyinterviewers maybe sensible regard it to estimable One answeris ignorance. example.Forfurther discussion HeydeandSeneta(1977. for eviwith and among New morerealistic thatdealtcarefully empirical the Testament Gospels. p. peoplewillwrite the interviewed thesameenumeramong respondents by out the thesetwoprobabilities-that corewilldry under ator. especially will and conditions that accident safeguard systems failto interviewers wemay decide regard interviewer to only. 140-141)as he chides Humefor shifttestiincluding mentof modelsfortestimony generally. 395). else every or bridge handwouldbe astonishing.In mostof thisworktherewas little or to reference byitself. In effect. The pointhad been raisedearlier. can but often isrealized. there set and are one porary examples. see 742-743). position. maywell ask whytheassumption further For in complications.g. abilistic of legaland legislative. 1973.which oversimplificadiscusses had a rebirth. in anycase. 2. example. a turning to reconciling. exampleby for the We witnesses. havetraced developbriefly historical Chalmers (1857. 327) as "la scandaledes Mathematiques.conditional often turnindependence.17 of of versity Chicago. or (Lapp 1974.1.fying equalprobabilities independence.so the investigator relatively ferential fertilities the next.thedevelopment stochastic models theassumption of especially thelatter's in famous Philosophical Essay-cal. analysis latent and structure deliberations legislatures. ErnestDeWittBurton. pp. about as (1972. More recent yearshas suchwork-done farmorecarefullyrecent related works McArthur are (1965). dence. president the Uni. and are factor theory. 491-499). 7.p. devoted much hisscholarly of probagainst direct energies Therewas.in particular citeCournot (1838) and Bienayme I exampleis TheLifeofJesus Strauss by (1892. 3..kindof incantation and of it. book3. to I p. haveseenthat central our lenging ability setup a usableprobability to spaceat is that independence-or of independence-among alleged all. in analyses matters the Gospels(e.pp. decisions aboutlaw-court cases. thecondition with culations something thosesketched like earlier testi. see Stevens and Burton1905).. FRANCE tension in Thereareintermediate positions theconstant Beforecoming contemporary to I A grand theme in back between examples.g. in he what a against was There eventuallyrevulsion (no matter howmucha liar) report with of Mill Stuart calledthe"opprobrium mathematics" it probability John qV > . samplepointin whichall report revival-assuming an expression (1888. 72).with may the See also theexposition Eggleston (1978).19 toFrench with thought. Vennshowshowp maywell about withquestions I pause to recapitulate. to Before circumstances. or point of ideas out stated. by translated French Bertrand into the independence. 28-34). (as he putit) finding and only harmony among. popular and is A example the are (1977). 1 in 1 million" Thereis a similar in of ambiguity somediscussions covariin Far moreimportant simple than ignorance.A widely readnineteenth-century dence. fundamental problems ques. That leads to a likelihoodratio of (V/. Twoexamples hypothetical monyto miracles.inwhich by tiontherelevance probability at all. see pp.1)10 for the (Mill1952.4. .18. multiplicationalleged of case.In myviewthere a central was to is obligation facesquarely to whichto refer him.Kruskal: Miracles and Statistics 933 Thatsomeprobabilityvery is small sayslittle nothing utor. more that dependence make hope a tremendous difference. any of the reflood core-is the product the two individual effect fixed thatthere no stochastic as so is dependence.see Fairwas probabilities characteristics casually of 6.empirical simplion and greatreliance a priori evidence One naturalstatistical of formalizing way over Young's ap.I think on thereare episteSometimes it difference.and probabilities error depending ciron not of witnesses factsaw a dead manraisedto life.sec. manipulation ginal probabilities. is many references. even todayegregious Yet possibledependencies amongthe Gospels. (1838). one page and considered bearing as unknown fixed difneglectsdepensmooth. 538).ingon unobserved. thetaskwas im. as a but of is assumption independence carefully ambiguities framing model. would cumstances. of Poisson a majorcontrib. and is of Starting Condorcet Laplace. . in a to without attention itsstrength due "the individual presented himself. ingreference populations. assumptions: proachis to consider alternative the that.orhopesthat makeslittle issues16 thatremain be sorted here to mological out. of in thus chalissue one we for testimony miracles. myopinanceanalysis where in nearby plotsofground agricultural marIt simplicity: is so easyto multiply ion. turn and verisimilitude simplicity. we have to selecttheconditions . RECAPITULATION ones" (p.especiallysec. Many otherwise themas randomly chosen. chap. Starting dependon thenature theevent question.In other with one or two circumstances. ."hasten The literature miracles on to includes much qualitative add thatthere was also workof theera thattried be on consistency among witnesses.if one of the people.e. extensive In serious statistical thought abouttestimony.pp.18 and thedifficultyconfirming posed uponus of selecting requisite the groups series or of frequency of a plurality moreor less suitable out of 8. p.McCormick most other applications. example. in turning contemmony judicial So thereference maybe questioned. 353. surveys with sevis ofindependence so widespread.is seductive are to fertilities on experimentation taken havecorrelated is formulas and simplify. OTHERFUNDAMENTAL ISSUES exand (1974) forcitations a tolerant ley and Mosteller An in discussion thelegal literature.15 Perrin tions madeinlegalsettings. probabilities.thusintroducing dependence "The combination of for that.As Venn(1962) putit.

it. Healy pointsout thatthe idiotBayes procedure Mostaccident analysts understand. difficult estimate" 25). is.Markand Paul" (p.forexample the butpasseslightly for service overjudging "The improbability production pseudorandom it: of numbers selective of coincidental is for similarity proportional the number to of lotteries 1971). of and in independent of (Thesetwocases beto re. givetheclassically ingenuous misleadingly Burns(1981.are not. orously argues responsibility showing itworks that Kiefer's one do rather thanthereader's. temptation assume called by Healy (1979. whatwe do and do notknow. Exactly In factmany (House of Bishops1986)deals in considerable with (p.pp.NewTestament theinstitutiontheEucharist. grandirony thatinto of A in and we want naively multiplying probabilities.continued The interest theBishopsof theChurch Engby probability belief of is the aboutreligious ability suchcommon-mode failures ofcourse most land. thatan ore depositwillbe formed . WhyMarkand Paul are taken see is proach.somewhat vague.Thus. which was in models intermediate The procedure.934 Association. saysthat"if we and of and inserting guessesabout manycomplexconditional bear in mindthe dependence Matthew Luke on in witnesses the wheredirectempiricalevidenceis non. Swerdlow was or (p. pletedependence. 57-61) findlittle no byMachol(1975)inconnection thesinking theSS association with of hapof polar between putatively traits positive . 61).cussion. served up to for to The events leading accidents.pp. hullandwatertight dependence.sometimes surprisingly butKiefer for that ple.Kiefer an entirely the of Machol. becauseofherdouble-bottom unsinkable of in mistakes a priori conclusion against assumptions in..21 Estimating probabilities In mostreal cases thereis noticeable bedependence likethelatter stretches thelimit-perhaps two to beyond tween when for phenomena.p. equal to 2n-1'' (p. radiooperator 9. Estimates accident probabilities nuclear of in pseudolikeliand are important. in the semblances shapeof. (1979.usually life related random sampling. example. is it simplifying assumptions-the simplest independence. and wereinsufficient of and Over-easy acceptance independence. In fact thoseknown forms are es. and in randomization (see Fienberg traits similarity. with lateJ. of withsentiments contentworking socialpsychologists How important dependence be is vividly can described ment(see Bradburn or 1969.ofcharacteristics having spores having amyloid (like on few turn the relatively seriousaccidents cidents. a it in thatcase the probability the same as thatof the in 1980. the in those us.December 1988 Journalofthe American Statistical hit and Before Titanic an iceberg sankin 1912. course.toand starting accidents. I think.allocation treatments experiments.maybe found somehistorical to a in literature.form thesample for point. 169) in the medicaldiagnosis Bayes asmust strong. 230).we go to great in the painsto achieveit.p.forexample.fora critical may analysis thatap.p. with taxonomy classification mushrooms. 294-295 and elsewhere) cited (1946.thereare onlytwo independent probabilities or of namely existent tenuous. is determined thesebooks contain muchmaterial relevant our disto ability the that bymultiplying probabilities each essential ingre. Lewis(1980. ferent levelsofcare.5mentioned of triggering event. the of and to study robustness Titanic. It is ironic thatMacholtooktheother different domain. exam. We it-the appropriateness probabilistic of ideas. 529-530) We see thensomeof thetrapsin gratuitous casual takesa contemporary of is historian astronomy taskfor assumptions independence. contains fullbibliography. forn suchchar.forexample.miracles testimony them. central a a without-replacement of and a torpedo. cause of their structure. addition. conditional not. after explaining multiplication underindependence. literature. as independent not made clear. 28).It is explicit but One facesthedifficult choicebetween pendenceand independence. arbitrary Or to averages extreme pos. the of to timating probablity extraterrestrialor theprob.forexample. Bethe(1976)saysthat cannot always that:"There or notis theproposer's in can be 'common where one event article appeared themycological mode'failures triggers analytical critical and of Macholreplied twoor three failures essential of elements thereactor. Singer. dient present" 68). example. A recent statement episcopal to so. assigning suggested in reactors gether had strongly of withR.) closely of war ability a world next year. is I from In theearlier natural discussion prayer. when really dependence Lorenz(1974) recognizes need forindependence.and one does notgetanybenefit of themultiplicationsmall of numbers. its usage is. Otherexamples be in point that are is mayreadily found they aimedat jointdistributionsknown bothordinary scienceand exoticareas. and on uncertainties setting physical lamellaewidth). shark. He refers striking independence speaking. Bothof by . there sidein a debate about with complexities notassuming continue the of it. part aboutdeand for accident are and pathways inbetween independence com.Deffeyes MacGregor and as (1980). strugglesfast. 296). muchin thenews. dolphin.The difficulties mushroom speciesto genera accordwith acmarginal probabilities other by as formed multiplying making thoseestimates.20 or gash on one side thatfloodedfive bergcut a 300-foot the In adjacentcompartments." is in factthe correct if are howcan one geton with job? the indepenprocedure all indicators assumed signment of thatone dent. I givea few examples beyond already mentioned. seems rare nature.for Garraghan pp. For (1979)vigcannot works well. form: "theprob.strictly aimedat full acters.somewhat example. of of sibilities be taken. CONTEMPORARY EXAMPLES was goingtoo shipwas asleep. be otherwise context independence various at points "idiot'sBayes.shewasdeclared the Unfortunately icecompartments. are surprised.Mark.the Titanic on a nearby lifeboats. prob. safely multiply marginal probabilities. with low-probability hood ratios or an ob. The same vagueness of at for It is notdifficult find variety statements dif.

recognizes gapsin this the counterargument stereotyped ways.. might facto. of Aye. suchindependence fragile. multiplying and comments: weak correlation not the toother is [Cautionary (a) againtwoyearsby eight provides effective an advantage 16 years! of sameas independence. 266). Notliterally first. from visible divine ofwitnesses.] Burns outline Hume'sseveral of acles-thoseexceedingly events-arise rare obvious. centrality persistence Hume'sargument between apparent nonetheless 2. lemsomemeasuring devices havearrangements blind article for ofWales. and Turner's article provides shifts expected value.beganwith Charles BabbageandI end British MedicalAssociation 1956. that half theremaining only of prayers for are length lifeas opposed of boons. 149-150).treats pp. of discussion over Page (1978) givesan excellent overview. of In Debate aboutmiracles continues within Church England. (1974. that had thepowerto change really further inthechain.pp. He wastheeighth lastEarl. reotypic femininity (gentleness concern others) sayhalf. was at (1939). Therewas a use of theprayer-gauge by Stead(1891)in his abouta then-current scandal on Prince centering Albert Edward. 857).p. "wellaccredited miracles theJewish Christian of it (1744. literature with another focused thePrince Walesand prayer-this on of controversy valuin illness hisrecovery. 312-313). 1699] transmission p. 1. is and Berkson. p. which unusually interesting is bishop Canterbury theBishopof London. Marty (1874)bookstresses similarity says approached "rather a random sample prestiof the ers and miracles constructively Trinklein.onlyto havehimbecometheobjectofa virulent attack press further using different for scales separate Benson(1930.velously. 211-213)gives lucidaccount a of by observations. of save forthefew (about Annet).especially 64) is striking. more lessquantitatively or (pp. be sure.a renowned and I later by of canals. Stephen "The says or is miracles and "irregular discrepant fact[s]"generally givenby fact theargumentbeing that is answered this proves itsefficacy to day that McKinnon is notexhausted. sec. out of other of an one examples. A philosophical discussion therelationship of it of As makes a goodpoint departure. the of Thereis a related on earlyrelated! substantial literature sympathetic of Steadexplains Prince's the life superficial as the consequence boredom processcontrol. oneaboutserious able further abouttheargument efficacy prayer. halfagain. Bishops. ofthat analysis shows Helmreich of 1978)putatively traits stereotypic polar masthatclergy abouttwoyears Rolive longer thanlawyers doctors. Act 2 of The Yeoman theGuardFairfax ortheprayer-gauge otherwise.has collected by carried bythePresident theRoyalSociety. A strongly pro-miracle viewpoint given Berry by (1986)toNOTES with further In context dependence of references. alleged and at that cheating baccarat. xxi)andBarnes(1965. the Brown Grier Illinois (Northern University) assisted theArch.I havenotedno refin to Galton's wise En. For Chalmers (1842)has an ingenious of the between efficacy prayer the (taken Burns(1981) treats of explain apparent inconsistency predecessors Hume. probably of of questionable thosemeasurements independent.4).with The testimony one eye I particular.23 Mahalanobis (1947)suggested carrying this hislife.butsomediscusses dependence.theincident. analysis.7.showedhow laboratory blood-cell "theurgent request thepublishers" 448). discuss the primarily argument and calledbyBurns a priori the epistemological ara and mony probability. example. had been to to ready swear him.manessaysthatonlya fraction prayers clergy truly of by are sincere. much he and more deceive senses the byunobservable divine intervention ipso simply. .See Babbage(1838. scientists . Vol.6. Similarly (Spence and holeinreceived Galton's Part 265-266)against empirical analysis. Steadcalculated at least haphazard resetting thezeropointbetween of successive a thousand million for of had prayers thePrince Wales beenmadeduring observations. In hispreface the of (p. the gether many among eccentric family witnesses. 3B). another debate There stated In glishman. 59-60. [Annet's and 89 and dispensations" treatment (p. through testimony. published or claims independence howthey eminent established of and inflated insidious and Commission auspices (see Archbishops' 1958. 262). is that Everynewassault a tacitadmission previous answered prayers small assaults as 3. thework Hooper[1689(especially of and of on of chap. 1.] he is saysthat givesit to show"thatthestatistical method nota trustAn almost in universal models for worthy assumption statistical instrument which gaugethephysical with to efficacy Prayer" of repeatedmeasurements real-world of quantities that (p. aboutthefundamental Christian miracles.and and for that of say turnout to be almostvanishingly correlated givensex.An account perhaps is by widow (1908). he putsSteadwith attackers infactSteadis toward Prince-ifnotto theeffectiveness prayer.Kruskal: Miracles and Statistics 935 natural andargues law.pp. Chalmers pointed that areunable trace probability that point backfor more a and prayers answered that.77 (Woolston). I notefurther he wrote first that Bridgewater Treatise! the (1977)callstheTrinklein "obscure almost book and answered naive. givesa systematic arguearlier ments I basedon testiintervention. surements thesameobserver boundto be affected saysthathe finds by are idea of bymemory prior observed values. the botheyesat once-this and that. a miracle are if worker nature marinnature than few steps. But whenit is corroborated theother. Earl'stestamentary The intentions were (1942)gives colorful 6. Marty. onlya fraction thoseareforclergy.To avoidthat prob. It may an oversimplification be toregard havenotdemolished hostile the works" (Stephen 1962. Galton describes omission the ambiguously. samedistinction beenmadebyChalmers The had aboutmiracles. (1967. Two outspokenly skeptical statements byThomas are descriptive (1985)andKramer (1985.vol. pinessand negative discontent. The opinions heard that miracle an answered or dilemma a proper prayer . in Archbishops' Commission (especially thelatter's chapter 3).erence these publications Chalmers's explanation. according Pearson to (1924-1930. He to theme. but the when each with variant scale markings-perhaps even nonlin.5. 174-187). I canbe damaging. plainly but "nothing recant" theomitted to in chapters. Hurn accuracy given Romanes's 4. to distinction chap.citetheChurch's (1986)Report theChurch England of of is naught-hemaylie. The omission. and culinity (self-confidence competitiveness) and versus ste. aroseafter goodevidence nonemaygainsay. miracles. . he wereas varied those as seemsto teara gious. although lackseventhe causative chains rough the out we to idea Humeuses. Inrecent on there haveagain beenBritish commentariespossible years to So we havecomefrom miracles. money. 80. (Wollaston). . largely from building operation English the and a account.shallreturn to a 1986statement theHouseofBishwith 1.. (1857. between and commentaries under relationships prayer healing.see pages10. who the consecration thereof now-Bishop David Jenkins.overcardcheating. p."And thoudidst all this?"Pointanswers. 59-63) describes distributed Muchofthat deals observations. Repeated mea. forexample.Annet stresses shrewd 237). theaverages (b) differ standard in to and Romanes. William by quoting Schwenk Gilbert.pp. 75 forgranted) theobserved and uniformitynature.Falk ops. human The and of testimony. arecriticisms-mildly of MedicalAssociation yetperhaps deep-of the British report the by see asks. the sense of independent in identicallyandlackofserious Turner responsibilities. is it by soon it YorkMinster setafire lightning after was by that in Here are bothpresent court. is 86-89).22 we know are yet that theinfluence CharlesDarwin. didnothesitate repeat favorite gument. Magath. Romanes's is under aboutreligion thinking fluctuated. Multiplication aneightfold gives dilution.pp.Ramsey 1971). (especially x.secondedition." the between Romanes's pray.pp. back Real mir. to countspresentsurprising correlations. 90-95.

. excan amples be more than matched many casesoffalse. especially 222-230). of not calculus. of events are this However we of we must sensitive aretothevagaries testimony. What areyour if miracles.and in passing.by Stephen (1962. Alleged ofargumentanother question arises is basic circularity that YetI must comment passing all ofus takemost in that at related the testimonyface in various ways.Middleton's independent Of bookwas widely it for influential.Thatis temperature down goes enough. and hallucinatory. we of on. 638).In scientific it is impossible eachof for Prince. is worth reasonable means canuse oftheattainment truth" 638).or at least. 32-49. Hume1975. majorreasonforhis Did thecreative start in divinity theuniverse a grand and was design then skepticism that "practically mainsource error suspicion our of and sit simply backandwatch spin?Or does divinity it tinker interfere is thepossible and of existence somekind collusion."however many they are but [thewitnesses] be.1969.Millsays use in of by volition. 15. Before it going however.251) and Heller ironic thathisviewsaboutmultiple witnesses should undercut arthe (1958). ation the happenings. seemsto go theother he that wayandargues nearunanimity with indeed. they without use of means.pp.17) tooka dimviewofprobabilistic corresponds the miracles. There a related is literature about of surprising degrees agreement 11. ball and sage(pp. (p. Hume. exactly Van Leeuwen(1963). Humeusedtheconcept probability a rather of in curious (p.an earlyEnglish deist. 528-529. even critical is summary byZabell (in press). by events. In presumably interms explicit probability Mill that programming.dence.121)butthereferences brief (Mason1967. excellent effect An discusses the of dependence and among views theFathers.An their Trench. pasevancein ice. namely. 144-145. so forth of observations apparently extraordinary find of however. is the evidence on whicheven the wisestreceive all those of of .15. thelike.. inaccepting opinions others: of "since there much is more falsehood and wrote 1730.theIndianPrince of most it. . that puter the of for program gives sequence natural numbers a thousand Jean Rousseau earlier a Jacques to presentedskeptical approach such and years then shifts another to suddenly in to sequence obedience initial calculations.dependence tavern" 31). uninterrupted as or of the of succession a natural order theworld" in theproblem but disappears. repeated it" A nineteenth-century of miracles eloquent supporter thebiblical and Some thinkers seemto takebothviewpoints different at central for was places. [but rather]not con. 224." asked:"Whowillventure tell does notgivea reasonable for interpretation miracles (Mill 1969. . As "Au(see Locke1959. theolo.9. ought be skeptical require to and strong but alleged a testimony.6) pointed in1690 dangers is validated the faith. calcuage-old debate aboutfree anddivine will intervention. 233). they always a few may tionist position appears pages178. writes a footnote we"cannot circularitythefollowing: NewTestament in that as The "pretends contain to a . when exists cannever submitted get of introducing to of Babbage(1838.5) interms theKing Siam. (p. 157. is yetmore realshifts-rather.p.as of character Christianity? first observed Pliny verified "every and has by by nurse that carried a 10.See also Sobel opposite view that to of expressed most by defenders thebiblical miracles (1987) and Owen (1987) forrelevant on variation accounts natural.116.December 1988 Journalofthe American Statistical . however. is of was of whoseexistence at first denied truths science. 1. chap..349).in many cases."It's an odd jumble prove truth a bookbythe in of to the error among men. themore traditional interven. me howmany are eye-witnesses required makea miracle to credible? a that miracle "must produced a direct be 473-478). 429). meteorites.For child" 232). argues theencompassing law-of-nature he concept "The Creedofa Savoyard to Priest. He howcan one reconcile timeless a universe thespecifically and historical forexample.pp. . couldnototherwise We on carry ourlives.Something thority . likethat thestory meteorites. lightning. thequestion independence addressed Middleton.doctrines be true in to becausecontained that in book" (Stephen 1962. ofHume'sstatements. to truth thedoctrinescontains.atleastprovisionally. 535). ample.936 Association. on theother in side of thefence. draws to of analogies contagion illness. p. Matthew out the As by Tindal. Hartman see (1976.pp. p.Prot. parochial The continues true life.pp. appears methat as regards miracles preordained in apparent shifts natural in unfitted consideration Probability eventhat for than law. performedproof your doctrine. and statisticians-along lawyers. Trench position suspicion is about varirough way. atthesametime conclude of it and to those be Heathens Japan. accept We time A saysHume(1975). 168-171)seemsto takethefirst it combination testimony view.of which or in or havenot they foryears(see Westrum 1978). A basicdivergence theliterature miracles in on to Venn(1962. (p. wouldproduce (pp.ofopinion historians. in Mahometans Turkey. 16. certainly ofdaily life. are pp.Trench put by that (1860)wrote is on but viewpoint taken pages480-484.p. 112. especially Bayesian (that human. estants England.Thereis an apocryphal Hume-like an.find paradoxical life I that hadreported meteorite in 1807. Yet Locke's 117). that"It is easierto believethattwoYankeeprofessors wouldlie than 13.1970)picksup themultiple-witness art 1878.Vol. . Whether p.among Fathers theChurch especially presenting algebraic the of (see Daille 1843). on and 498-499.and Peirce(1958. fresh problems then arise.Daille in mulasfortruth testimony and under various conditions. 217-241. 61). Thereis a sparse extended but literature for. simply be following Hume. ofa more parts general His analog to a com. 170. is testimony" 428). others-by the strengthens case (see pp. Yet.236-237. of 479-480)ontheweakness testimony thebiblical for 9. coursetheseex.He simply never to had observed water and more at mathematician cannot check confirm thana small fraction of lowtemperatures.Another discussions the of state probabilityorbefore of at Hume Burns see of authoritative accounts historical among (1981).Eveninmathematics is an analog: to there Themost and energetic stringent formable it" (pp. sec.whoincludes bitter a pp. on menofhigher and than intelligence larger experience ourownis the 8. One theme thisessayhas beena skepticism of abouttestimony. The Stewart discussion deals pointand wonders whether Humehas overstringent of standards eviexplicitly supposedly with observers.is discussed.268-270. SouthIndianPrince.is toldthat water becomes hardandsolidwhen of we the to checkmorethana little what havebeentoldor read.pp. peoples' behavior is "not contrary his experience . pp. Alexander Campbell Fraser. 113-114). 470-489).g.p.chap. from randomly 38 selected patrons anyfriendly of with authority experts the of neighborhood dispense and. . example.he tonumerical calculation. sibility extraordinary that notmiracles. increases credibility. lightning exists beenquestioned years has really for (Garfield 1977. clination training and listen with to specialcaution human testimony. ecdote about Thomas Jefferson: informed two When that Yaleprofessors persistent skepticism would make impossible.ofsingle law.book4. outside time. 407).butnot .In Maimonides's the law guments forward other (1963)GuideofthePerplexed general of critics Hume. . may called. in is and material. is wryly It importance Christianity Richard is example Maimonides. mayfind one suchboasts of editor. theBookin turn in Locke(1959. in and Lutherans Sweden"(p. example The Lockein 1690 statements proofs themathematical goesbackatleasttoJohn and in literature.Perhaps most fundamental is thevaliway of valuemost thetime. a fall is said Jefferson tohaveremarked andpuzzling. 305-306). dation faith miracles of via described a HolyBook. 263). Trench's For may is discernmentin anycase questionable.474). of notably by mostmiracle enthusiasts. asserts. Swinburne (1968. chap. of Sincewe canseldom from time time suddenly to by miracles? and it it be entirely rid thedanger. Mostof thesewouldnotbe calledevenpotential miracles by 14. mentioning thepos. thelastchapter hisbook. primarily becausethere little is religious rel. in. haveneither norenergy question. Millsays. the aqueous ustorecapitulate than tiny more fraction other of observations.p. facts history in life. of opment probability concepts. I do not. whoregard of gians creation theuniverse timeless.compared themultitudes attest fact For with a who which excludes their fact. Humein fact refers multiple to witnesses several at placesin his that stones would from fall heaven" ball essay(e. menhavereason and .of anymeanswhich simply themselves the if so require much proof?" (Rousseau1911. Thusprofound. Therewerealso restatements Hume'sargument. by tendentious. in Papists Spain. peopleare lighter that whenawakethan whenasleep. interpretations the Patey(1984)describes devel.most again (p. 403. Uman 1982.personally examined proofs" the (Mill 1969.thantruth knowledge .putitcrudely: lations To of A abouttheveracity jointtestimony. In of 12.par.Mill(1885. Stew.

Give us. and the possibility a miraclefollowsat once" (Stokes of 1891. and theproduction hisdirect by volition of an effect . "Statistical eds. DavidJ. .vol. . p. is. 32)]. (1978). Lewis's argumentwas not entirelynew. accumulative..experimental that only an ancient disciplinelike number theorycan respectability bestow" (p. God impliesmiraclesimpliesGod. app. 2.Special Problemsof: OutAnscombe. Apparentlythe independenceassumption ually in a contextwhereit could hardlyhold.E.(The first Treatise: Fragment A Bridgewater (1838). 25. The opprobriumstatementseems to have appeared first the in second (1846) editionof Mill (1952). (1930). highly but geometrical. vol. writesthat"Statisticalindependence. In a modest probe.I cite as examplesLewis of may (1849..F. yet alike in the frequency withwhichtheyare ignored. one B. p.vol.Perhaps we are all forcedwilly-nilly cirinto cularitiesthatwe dignify callingthem"coherences" or "interconsisby tencies. duringa trialover a was used cascontestedwill. by Benjamin and Charles S. 18.. F).). to worn themissing and by girl.. 109 in 1947 ed. For discussionsee Mascall (1956. 123-125) used theFrenchlegal example as the basis forhis critiqueof casual independence. 7. erences. and so on. assumption-independence. . in his well-knownbook on miracles. 6. [See De Finetti(1982) foran intensely personal statement The truly basic idea forDe Finettiis not independence key references. be reckoned with as a serious possibility"(p.ifwe believe in God. . 553).Kruskal: Miracles and Statistics divine revelation.. ideas applied to tribunals.butmultiplied hundreds multiple or thousands" (pp. Tanur. 352). at 351-352). an unhappy of record.as part is by of hisrecapitulation Hume. Andrews (Cook 1821. A. 8. An example of extraterrestrial between Donaldson and Pollard (1980). although do not findit in the 1848 I Americanedition(see Mill 1973. and Walker(1975. merely that was of not discussion).I inspectedthe six of and counted 11 reviewsof introductory of and graded theirtreatments independence these books in our library no tolerantly: A's. in in has a literary endorsement Casual multiplication a legal setting Poe's "The Mystery Marie Roget" (1904). p. calculationis the correspondence 21.). p. about independenceis 20. Following the opprobrium thereis a brief. Discussions of such a device for measuringblood pressure are and Dore (1970) and Labarthe. and threeF's. arithmetical.(FirstpubWestern lished in 1937.. . achieved the of gamblers. 106. sec. Otherdiscussions circularity be found. (with Statistics Theology" and (1988). The quotation may remindus not to sneer at what appear obvious of circularities argument. two C's.belief in himdoes not facilitate beliefin miracles"(Stephen 1962. pp.. especiallypp.Charles World(3rd ed. Cooper. thereis no difficulty believing in [in] miracles. 162). I in return ratify will in the yourfaith uniformity regards overwhelming as majorityof events" (Lewis 1960. Perhapsone reason formisunderstandings in inadequate training classrooms and lectures. . The two testaments by form a double mirror. Poe writes of that"If the feet of Marie being small. J. . pp. W. when he writes"Theology 'Admit God and withhim the riskof a fewmiracles. however."That statement by JohnCook.As WeWere/A to London: Victorian Peep-Show.) (2nd ed. Among the problemswithtreatment independencein these the in difficulty even finding independencediscussions: books. 339of 346). what analyticoptionsthereare. p.That option relevantjoint distribution at does not affect primary the themeof thispaper: to look critically any whatever-and exchangeability.and in some cases I gave up aftera fruitless search of the index and table of contentstogetherwithrapid weregiven..] of but that (plus identicaldistribution) exchangeability. 7-8. London: JohnMurray. how to recognizeit. 1987. London: ChurchInformation editionappeared in 1837. 1039-1944. . Board. Peirce (fatherand son). 72). Journal the of RoyalStatistical thisshoessuchas she had been known wear . to all or Add Society. p. pp. invariance the of underpermutations variables. see also chap.we seek for nothing further-whatthen if two or three. 8. thatsimply viewpoint 19. 501-502) give a fascinating storyof 1867 experttestimony two greatAmericanmathematicians. pp. Benson. . therefore.. Kac (1982).Hawkins. no evidencecan prove a miracleto anyone who did not previously believe the existenceof a being or beings with supernatural power . the increase 937 If we seek for nothingfurther.theremight mention independence randomvariables apand observations. One textsforcasual assumptions independencein statistical could easily go on. of chapters. Mill says. simplifying fails. hypothesis . is no longera purelyarbitrary .makes a virtueof the circle. I Association Statistical of American issues the of Journal the looked two at textbooks. Variantsof Tindal's criticism come from internal the structure of the Bible. chap. constitute of the evidence that this pretensionis foundedin truth. Once admita God. flowers thehatcorrespondingthose in Green. or more? Each successive one is by evidence-proof notadded to proof. of Archbishops'Commission (1958).Revised October May1988. 287. Analysis. p. 385). Chalmers(1857.no miraclecan prove to us theirexistence"(Mill 1973.In at most one of the books was thereanything proachingresponsibleconcernforwhat happens when thereis dependence. that"the existence God cannotpossibly of of be proved by miracles. 180. . 129).pointeddiscussionof overstatement. p. thatof Frenchmathematical-political thought. An important exceptionis the positionof De Finettiand his folwith lowers. each reflecting other"[from reviewbyKermode the the (1982. of a personal God.Whentreatments independence readingin likely examples. ofprobability thebody that Mariewould be . Lewis. 538. Mary's College. one D." There is.London: M. especiallyp. Peter (1744).withoutmentionof Stephen or Mill. for avoids the verisimilitude example. Lyttle(1982) provides an example of the blitheassumptionof independenceto calculate the probmissile. In additionto decreasingor avoidingdedevices of thiskindmay also pendence among contiguousobservations. If God is thecause of order. p. There is also a somewhataloofmathematical questions emphasized here. to worry about consequences when the assumption 23. Eddington(1935. 34. especiallypp. prois fessorof divinity St.. those of the corpse were also small. "How do we knowthatthe Gospel story is true? Because it confirms propheciesof the Old Testament. thatsome of us are willing to examine our own circularities. whichHume has made out .. 285-286). New York: Free Press. only one flower. 17. The Church's Ministry Healing. decrease or avoid sources of bias and unwantedvariation:integerprefand so on.and Reminggivenby Wright ton (1973. Longmans. The relevanceof thissubtopicis thatit presents methodological a tangle different fromthose of dependence of testimony and reference set. 151. The famous mathematicalphysicist G. For a fine treatment abilityof accidentallaunch of a strategic accident probabilitiesfor liof the dependence problem in estimating see quifiednaturalgas transportation Fairley(1977. "All.But the how do we know thatthe Old Testamentpropheciesare true?Because theyare confirmed the Gospel story.the apparent miracle can always be accounted for on a more probable hypothesis. For moreon Mill's distaste simpleprobability forformalism. . St. Bartholomew. Mill pointsthisout again whenhe writes. A third circular argument criticized Mill (1969).137-178. Kruskal Encyclopediaof Statistics. and. Related problemshave arisenin examination acousticevidence about the assassinationof JohnF. H. I cite first Indexes were absent or skimpy. liers. Stokes had givena variant:"Admit the Existenceof a God." in International and J.TheNinth Babbage. Meier and Zabell (1980. London:SCM. 82-93). 477). 1. 22.) and Elmer History the of Harry (1965). (1984).1 [Received REFERENCES of Annet. sec. F.An Intellectual Cultural Barnes. 185). . Kennedy(National Research Council 1982. 24).it gives a narrative the veryfactswhich . M."iProbability.and says in effect. Ser. then. . 7.Hume's answeris plain. and some simplydeny themwithout study.. [and] must .thisdifference. If we do not already believe in supernatural agencies. is that.once a shadowypartner has scientists and statisticians.. New York: Dover Publications. including see Schabas (1983). p. chap. vol. G. . 625). in support of this high pretension. The Resurrection Jesus.). p. In vulgarbrevity.GodofChance.forunless a God is alreadyrecognized.observerexpectations. plus a fewsimplecombinatorial theytendedto be definitions be for of Withluck. pp. Frye (1982) asks. Stephenmakes a similarpointin discussing Hume and Paley: "When Paley calmlysays. Portnoyand Petersen (1984) criticizea studyof biblical of analyses.

Frank. Proof and Probability. Walther (trans. Bertrand.June 294. M. Garraghan. (1973)." Scientific Theories. L. pp. T. "Statistical of Norman (1969). Paul (1968).Francis(1872). publishedby OxfordUniversity on theRightUse of theFathers(2nd ed. Nidditch.322. I: Singer). Press. Stuttgart: ematical Models (with referenceto a methodologyof Machol and Klotz. D.R." Science. Fairley and F. New York: FordhamUniversity Joseph Glanvil Gilkey.) Belief (Report of Heller. Literary Supplement. A. 5).67-71. Eichrodt.Oxford."Human Nature. it also appears to say that the Discourses was first Hartman. of Philosophical Medical Times. "Science and Theology as Art Forms. 3-18. by Testament [Published in 1964 (Vol.An Inquiry Cook. Review of The Great Code: The Bible and Literature. John Theory Heyde. Eddington. It appeared Daille. (The catalog of the BritishMuseum gives a range of 1836-1842 Essays. Morton. R. Evidence." American pp. The Nature of Christian pp. Well-Being. I. and Pollard. quees. 1976.] Games: Strategic J. S. 369-379.). C. Statisticians.ed. 62. London: Author. publishedin 1748. Eliade.. "Maimonides' Theory of Miracles.& Human Values. 15. (1928). Glasgow: Collins. Theories.-A. J. 85-106). G. Frank. Theologyof the Old Kermode. Thomas. Donaldson.71. C. Selby-Bigge. "Probabilite des Jugements des TemoinP. "Probability House. (Reprinted in 1985 in On Being the RightSize and Other 262. East. A. Anticipated.in 1977 in Statistics 32. East and West. York Times 41.] Vol. Calcul des Probabilites. Technometrics.(There 21. L. February 10-73. "The Necessity FissionPower. CatastrophicAccident From the Marine Transportation Liquified Kramer. MA: Addison-Wesley. "The PrayerTest. of D. New Pathwaysin Science. 479-490. Resources. 109-131.Hooper..TheBridgewater Falk. p.KennethS. Chalmers.Biblical Brams.and Mathas Theologie des Alten Testament(5th ed.David (1976). Alexander. J. 321-322. Paris: Gauthier-Villars. London: East & West Library. John(Jean) (trans. langlez. of Planets?" AmericanScientist. Tomorrow." of Study." Cullipp. (1981). (1971). M.Inquiries Human BritishMedical Association (1956). MA: Addison-Wesley.Miracle Natural M. 137-182." Berry. 1." in Mathematics West. William B.Thomas (1842). 1974-1976).) of Christian adelphia: JewishPublicationSocietyof America. "LetterFromEurope.Philipp G. Northrop (1982). pp. thisis fromthe posthumous1777 ed. H. ages. April and Public Policy." Nature. (1976).(1857). (1888). et Bienayme [Jules] (1838). J. "WorldUranium Hume.Richard The AboutCollins.Arthur(1935)." in Statistics Kruskal. "Mathematics as Propaganda.ed. 120. and Mosteller. pp. "When Citation Analysis StrikesBall LightScientist(Vol.December 1988 Journalofthe American Statistical The Berkson. "On the Consistency Law in Graeco-Roman and Robert (1952). M." Mycologia. (1977). of Amsterdam:North-Holland."Nuclear review this of [See tory. Fairley. Autobiography. R. (1956). ed. sec. June9.Frye.PA: Bucknell University of Inquiryand Belief. R. New York: Springer-Verlag." in Extraits Proces-Verbaux Seances.J. pp. 257-334.Bruno (1982). Darwin R.WilliamB. 201-204." in The Validationof Scientific 21-31. Debate Miracles/From on Burns. M. 130-134.C. Steven (1980). "The Fienberg.. of Bradburn. London: Eggleston.Margaret(1939). "A Calculationof the Credibility Human Testimony." Chiswell. January. cytometer. Reading. Gani. 93-96...U.London: Church and My Life.Magath. New York: Springer-Verlag. York: Harcourt. (1977). Mircea (trans. by W. William G. (1980). With Revelation Viewed Connection Modern Theology Haldane.) Belief. New York: Harper & Bros. 3. in Works(Vol.Independence. "The Varietyof Reasons forthe Acceptance of Bethe." in An EnquiryConcerning 66-76." Chicago: Aldine. TheNatureof Christian De Finetti. Frye.New Series 12. 68. San Francisco: Harper& Koblitz. Joseph (1958). 2. HisMillions/A CandidFamily (1942). Errorof Estimateof the Blood Cell Count as Made Withthe Hemo128. August. Cambridge.London: Judiciare. (1979). London: Hutchinson. in the Old Testament. ed. and GeneralScience). 18." UniversityChicago 41. Between the Efficacy Summer. C. "A Conversation Law Review. (1946).Thomas B. American.. (1960). Baker) (1967). and MacGregor. Hawkins.938 Association.revised by P. A Guide to HistoricalMethod. 242-253. 234. to David Hume. George (1689).. Mosteller. pp.Ian D." Scientific Understanding (3rd ed.). InquiriesIntotheEfficacy Prayer. 367-377. "The Medical PowerofFaith. The JournalCardiology. January. Fairley. London: ChurchHouse.MA: MIT Press. "The Objective Efficacy ofPrayer:a Double-BlindClinicalTrial. January 255-261. MI: Brown. "Sur les Applicationsdu Calcul des Chances a Pures la Statistique de Journal Math6matiques etAppli.] in TheTimes . (Firstpub1855." in Essays of an Information ConMind. 186. "Randomization and Social Affairs: 22. of (1699).along withother relevantmaterials.Eugene (1977).97. (Firstappeared in Current Eerdmans. (1969)." in WeAddedDiscourses Between IllustrativetheConnection of Possible Worldsand OtherPapers.40-47." in The Making of Statisticians.Joseph. "The Creationist the R. 2). (First Healy. . New des des Code: TheBibleand Literature.Stephen E. 10. "Does Medical StatisticsExist?" BulletinIn Revelation the of Statistics."Journal ChronicDiseases.K. Applied publishedin Edinburghin 1814.) lished in 1631. "Commentson Taxonomy. New York: Springer-Verlag. 7). Miraclesand theCritical Philadelphia:ISI Press. Weidenfeld& Nicolson. 331-353. of has been disagreement about the authorship thispaper.) Between Doctors and Clergy. 234237-252. "Evaluating the 'Small' Probabilityof a of 61. Grand Rapids. William H. and Seneta.PlatonJ. H.) tents. NewYork:Carter. 111Row.] lomatique de Paris. ed.Reading. pp. (1974). Richard (1978). 13. Cournot. 1970 Draft Lottery.ed.546of suring BloodPressure. Steen. Mark (1982). 3-20. "Life on Other Joyce. 112-127. (1980). and Weldon. Correspondence(1975). 30-33. To in Astronomy! Which and Early ChristianThought. "Of Miracles. J. Ricketts) (1981)." AmericanPhilosophicalQuarterly. 125-135.Philpublishedin 1818.Bernard or Folly?" New The Salvation Nuclear book Lapp. J. and Evidences the Miraculous Internal . 62-72. Jekyll)(1843). Philadelphia: Westminster.New York: OxfordUniversity press. New York: Macmillan. Paris: Societe Phi. (1821). Smithand G. Natural Gas. 6.Jerome (1978). (1974). Neal (1981). Kac. "What to Believe About Miracles. Human Deffeyes.Frederick(1974). J. L. 10. "Evaluation of Performance Selected Devices forMeaof [Reprinted." in Between Church of England (1986). AmericanJournalof Physiology. Maimonides:Torahand PhilosophicQuest."Science.Jane(1985). pp. Churchof England Bishops). and Hurn. IntotheBooksof theNew Testament. 309-323. the(anonymous) Magazine. 77.: Clarendon Press. "Some Remarks on Wild Observations. Technology. May 12. pp. Gani. The New Republic. A. 346-353. "On Miracles. ening. "The Search forthe Meaning of Independence. (1838). Review. London: Bohn. Journey pp.Galton.) 5.GilbertJ. N. Stephen G. B. Frank (1982). J. July24. Antony(1967).TheGreat [ReviewedbyKermode (1982). E. 561-563.238. August. J. Policy.Langdon (1982). Smith. A. Commissionon Divine Healing. (1965). M. J. ed. A Fair and MethodicalDiscussion. (1979).J. Boston: Beacon. "The Efficacy Prayer:a Triple-Blind Transactions theRoyalSociety." TheNew Yorker. Journey 1907-1937.. ofPrayer-and theUniformity Nature" (Discourseson theChristian Grant. "Miracles. 2.TheStructurePsychological M. 343-378. Lonand Into Faculty ItsDevelopment. DeBrush. 6-7.TheGreat Controversy: Interrelation Press.Colin (1984). 63. Flew.] andPublic 553. Labarthe.On the and AuthorityItsRecords." in Encyclopediaof Philosophy(Vol.A.ed. and Remington." memorandumto the Archbishops' Garfield. New York: Macmillan and Free Press. 1-3. Amer. (1883). "Divine Healing and Co-operation don: Macmillan. [Reprinted The Fortnightly of A Analysis Stories in Means (1876. Ehrenfried Kiefer.).RalphE. Frank." pp. New York: Springer-Verlag. C. 359-365.) for the Works." in The Makingof House ofBishops(1986). (First Dietl. pp. A." AmericanScientist. BienaymelStatistical Edinburgh:Waugh & Innes. David (1975). A Treatise Press in 1757 and in Hooper's Works. Lewisburg. (1986). 73-89. (Second editionpublishedin 1907 by Dent. of Scientific ican. B.Brace & Jovanovich. 128-131.

Toronto: University Toronto Press. (1977).David L. pp.: Theoryand Literary Press. W. St. S.K.Eliot (1986). and Toronto: Strauss. the Utility Religion and Theism.No. The Conceptof Miracle.54. "On the Importanceof Christianity. M. The Life of (1973). May9. ininityl (2nd ed. 535-536. Correction(1982). eds.Uman. New York: Lewis. Tucker. [First People. Notes on the Miracles of Our Lord CambridgeUniversity (2nd Amer."Interfaces. For a fullversion see the 1969 Toronto edition. John(1959). John0. Cambridge. gregational Publishing in Leslie (1962). Jo (1985). Antecedents.). JeanJacques(trans. (1974). 144. "A Surveyof Recent Gospel Research. don: Macmillan. (Firstpublishedin 1842.TheResurrection Norman Perrin.75." The Reviewof Reviews. Uni1846. 402-404. 879-887. Plagues.). Peerman. Smart. "Meteorites. (1861). 11-21. 234. New York: Dover. Boston: Con23-34. 289-321. LonRomanes. "On Biblical Censuses. York: Ronald. London: CharlesW." Harvard Magathe Price.London: Longmans. 429-455. P. (First published in 1835. A Scottish Patey. Paul. Foxley) (1911). (1962). Nature zine. S. 1192-1194." AmericanScientist. StatisticalAnalysis: Zechariah and Beyond. Evans)] (1892). "Toward the BluntBishop. Fraser New York: JohnWiley. of Austin: University Texas Press. Prior. (ed. (1977). London: Swan Sonnenschein. eds. October 8. Johnson. Trench. Nature. McKinnon." Biometrika. Ralph S. 218. John Stuart (1885). of Mahalanobis." The Philosophical Swinburne. Natural Theology. and ReligiousStudies.4. 8). E." The Mill News Letter. Harold W. Philadelphia:Fortress. "Statistics Portnoy.Douglas Lane (1984). November 15. (FirstpubAssociation. 127Swerdlow. and Burton. lished in 1876." Journalof Biblical Litence. Newton's The Crimeof Claudius 133. "The Mystery Marie Roget. and Petersen. and Miracles. L."4The C. Edgar Allan (1904). [Firstpublishedin 1843.) London:Parker. 521. Philosophers for Sobel. "Process Control. New York: Harcourt. New Schabas. An Essay Concerning Sciences (Vol.) FightGoes Public. osophical Quarterly. "Science and a Future State. New York: Collier. and ReligiousTruth. W.5. (6th impression).New York: Macmillan. I. of (1969). Journalof Philosophy SCM. (1956). (1965). 81. E. Kotz and N. 4." The AmericanScholar. S. Cleary. JordanHoward (1987). (1984). Wiener. ence. Karl (1978). (1971). A HarStudy ed. Mascall. pp. (Firstpublishedin 1947. [Firstpublishedin 1874." in Four Dissertations Coincidence.London: and Literary Form/Philosophic Thompson. (Lecturesgiven 13. 1-11. (1891). Richard (1777). and Helmreich.TheLifeand Letters George ferent Laboratories. S. S. 211) of Debate. (1980). 809-812.David Friedrich[trans. "Comments on Taxonomy-A Response. Emile." PhilosophicalQuarterly. and the Princeof Wales: Press. pedia of Statistical A. 661-662.Grand Rapids.D. Das Leben Jesu. TheirPsychological Dimensions. NewYork:Charles (4th for mony the of Gospels Historical London: Longmans. 166-186. eds. The God of Science. T. 7)." Scientific Pomeroy.U. S. H. 5. American.ed." Archbishops' DoctrineOccasionalPaper 1. Mason. 229-234. [Firstappeared in Australian Madansky.Balfour (1878)." Journalof theAmericanStatistical Century (3rd ed. RobertE. "Our Understanding Prayer. From of London: Longmans. 31-38.Harold William(1931). subtitled"Three Essays on Religion" Scribner. 53Machol. of of Lewis. (1968). P.) (Vol." The in New TheologyNo. byJ. Pearson. (1974). 185.] SimmersDown.. "JohnStuartMill to WilliamStanleyJevons: An UnpublishedLetter. turies. "Biblical Textsand Lorenz. "Hume Versus Price on Miracles and PriorProbJohnsHopkins University Press. George J. and Zabell. Philosophical & McCormick. Jr. Alastair (1967). 2. Margaret(1983).) Ptolemy. Quarterly.." Values a Universe Chance/Selected in of WritingsCharles Stanford. "Prayerand NaturalLaw.) pp. C." of acles: A Bayesian Interpretation David Hume's Analysis. to According Matthew. Pines) (1963). P. R. 1222-1224. (1958)." in Fragments Science Tyndall. "The Princeof Wales.Kruskal: Miracles and Statistics Mark. 1-12. MI: Trinklein." SaturdayReview.) (1876). Page." AmericanPhilStead. LawsofNature Hume's and Argument Against forUnscientific London:Longmans. A. The Life. London: SCM.242. March-April. Commission Christian on Lecture on the Combinationof Data From Tests Conducted at DifRomanes John of Romanes. "Academy Membership Rousseau. ed." in EncycloStephen(1988). Richard Chenevix (1860). G. p. Green. Pearson. Stephen(1986). (Firstpublishedin 1767. 16. (1960). (1891). ed. A System Logic/Ratiocinative Inductive. Handbook of theLaw of Evidence Spence.Review BallLightening BeadLightening.) pp.Brace & World. G. "The Apocalypse Equation. "'Miracle' and 'Paradox'.) of Mill. Stewart. American Scientist. MN: West Publishing. Martyand D. (4th ed.Green. C.K. Macmillan. E. OxfordUniversity Practicein theAugustanAge. The quotation is from"The Creed of a SavoyardPriest" (book 4).M. in as National Research Council (1982). 497506.U. pp. (1955).) 37. Barry. (Includes the BurneyPrize essay for 1873. 187-202. New York: Appleton.New York: Free Press." Journalof British in Pearson. Griffin."ASTM Bulletin. Thomas. and of 80. London: SPCK. 561-569. C. Marty. "Ptolemyon Trial. 320R. of University Toronto Press. editionpublishedin 1894.103. ChristianTheologyand Natural Science. 24-28. (1980). Stanford CA: Uni. Sandy (1980). 399-408.). "JohnCraigand theProbability History: see treatment the 1973 edition (Vol." The Princeton (Vol. [Ascription authorship Means. Turner. London: Cadell.Albert (1986). Curve of Error. . 46-65. Study(2nd ed.Charles Tilford(1972). (1924-1930). Dissertation4. Stevens. Quarterly.Green. Second edition Stigler. 218. Maclntyre. 308-314. N. History EnglishThought theEighteenth of Stephen. Encyclopediaof (1970). The Prayer-Gauge and frominternalevidence. in versity Press. "Summaryof Ramsey. ed. (1979).Portnoy. (reportedby J. L. Harvey K. 201-221.] Society. of 48. David (1987).1).Noel M." in International eds. Brian (1967). Owen. "Reexamination of Acoustic EviTuibingen. (1971). "Historical Note on the Origin of the Normal FrederickE. Miracles/APreliminary and Luke. Paul. 55.London: Black. "On the Evidence of Testimony MirMcArthur. of HistoricalEvidence. William Arnold. Probability Press. The Guide of the PerPrayerand General Laws. July19. NewYorkTimes. Peirce." Monographs.).Bradford(1982).G. "Can Religion Be Discussed?" in New Essays in Philosophical Theology. 29-35." Journalof Official (1942). 248-327." Sci.Ninian (1964). A System Logic/Ratiocinative Inductive Review. (First published in 1690. George Cornewall(1849).Chicago:University Chicago Press. Lyttle. Flew and A. No.pp. (1975). Locke.19-20. Eerdmans. London: Deut.MartinE. For fullscholarly the Death of Christto the Birthof Laplace.Frank M.] and (8th ed. 10 of the CollectedWorks).London: 72. Masculinity Fem& Correlates.] in published Mountaineering of Miracles. of Lewis. G. 18. M. Follow-up (1987).] of Statistical Association." in Works ." Statistics. Speech 29.pp. "Rainfall. G. "Benjamin Peirce and the Howland Will. abilities:Testimony and the Bayesian Calculation. of (1952). Peirce (1839-1 914). T. comes fromBenson (1930. CollectedWorks. Konrad Z. December 10. 236. of John(1871). Maimonides. (Essay-review R. of Jesus. Meier. 679-683.B. Eliot (M. "The Titanic 5. Moses (trans. Karl (1924). in 1921-1933. 2.] Statistics. A. Christian of plexed. Stokes. E. 2. W."Science. in Matters Opinion.(Written 1901.) 54." SciMarshall. (1982). 523-531. ed.: CambridgeUniversity of A Chapterin the Conflict Religionand Science. "Hume on the Testimonyfor Miracles.). E.). Fraser.London: Macmillan. 37. "The Safetyof Fission Reactors. Kruskaland J. 89. 328.18. Ernest DeWitt (1905). 53-65.) (1947).RobertL. erature. Lettersand Labours of Francis Galton. "Miracles.) StephenL. An Essay On theInfluence Authority Poe. "Analogy as a Source of Knowledge.Green. A. (1978). Cambridge." Journalof theAmerican versity Toronto Press.). 63-65. The History Statistics the17thand 18thCenof Studies." Mycologia.) dence in the Kennedy Assassination. ThatcherIs Irreverent. (1978). March. Ethel(1908). JanetT. Man of Feeling. (Firstpublishedin 1964. Tanur. Human Understanding. (1874). "Science Versus Religion: An Old Squabble [Firstpublishedin 1762.

and Journal Statistical of 493. R.(First pubon Someofthe SouMiracles. S."A Random-Zero G. (First published 1866."Scienceand SocialIntelligence Analysis Testimony.) in AboutAnomter & Law. P." Lancet. 14. 20.3rded.An Elementary & (4th and Practical ed. Planning Inference.J. February 337-338. The manometer. and Magic/From Ficino CamIs Study of Advantages Mathematical to Which Added panella. John (1962). 1888. (1844). 461.940 Association. of 1. (in press).IN: University Notre of Dame Press. and Dore.Zabell. (1975). TheProblem Certainty English Wright. M. Ron Westrum. Sphygmoin Van Leeuwen. Hague: Nijhoff." of alies: The Case of Meteorites.. Venn. L. F. South Bend. NewYork:Chelsea Young.December 1988 Journalofthe American Statistical B. C.Three Walker. (1978). Henry (1963).). London:Souter Law. in Publishing. London: Argument Against of an Examination Hume's lished 1958."The Probabilistic of 8.) College of Addressed the to Students Belfast to Lectures (1846). (1970). on Theoretical Treatise Algebra. the ThoughtIl630-1690. .Spiritual Demonic D." Social Studies Science.TheLogicof Chance (4thed.).

Sign up to vote on this title
UsefulNot useful