Response to the Letter to the Editor

The Annals of Applie d Statistics 2008, V ol. 2, No. 1, 432–433 DOI: 10.1214 /07-A OAS154 c  Institute of Mathematical Statistics , 2 008 RESPONSE TO THE LETTE R TO THE EDITOR By Clifford Spiegelman, S. J. Shea ther, W. A. Tobin, W. D. James , S. Wexle r and D. M. Roundhill T exas A&M University, T exas A&M University, F or ensic E ngine ering International , T exas A& M University, Hightstown Hi gh Scho ol and Chem Consulting This pap er has attracte d interest around the w orld from the media (b oth TV and n ewspap ers ). In a ddition, we ha ve receiv ed letters, emails and te le- phone calls. On e of our fa vo r ites wa s a v oicemail message asking us to return a ca ll to Australia at whic h p oint we would learn wh o really killed JFK. W e w elcome the opp ortunit y to resp ond to the letter to the editor from Mr. Fioren tino. Mr. Fioren tino claims that our “state men t relating to the lik eliho o d of a second assassin based on the pr emise of three or more separate bullets is demonstrably false.” In resp onse we would lik e to simp ly quote from page 327 of Gerald P osner’s b o ok Case Close d , one of the most w ell known w orks supp orting the single assassin theory: “If C onnally wa s hit by another bullet, it had to b e fir ed from a second sho oter, since the W arren C ommission’s o wn reconstructions show ed that Oswald could n ot h a ve op erated th e b olt and refired in 1.4 seconds.” Mr. Fioren tino also claims that the “second fatal fl a w is the u s e of a rather uncomplicated formula based on Ba yes T heorem.” Let E denote the evidence and T denote the theory that there were just t wo bu llets (and hence a single sho oter). W e used Ba y es Th eorem to h yp othetically calcu- late P ( T | E ) fr om P ( E | T ) and the prior probab ility P ( T ). In order to mak e P ( T | E ) ten times more lik ely th an P ( ¯ T | E ) , the ratio of the prior proba- bilities [i.e., P ( T ) /P ( ¯ T )] would h a ve to b e greater th an 15. Thus, we again conclude that this casts serious doubt on Dr. Guinn’s co nclusion that th e evidence supp orted just t wo bullets. Sadly , this is far f rom the fir st time that probabilit y has b een misun dersto o d and /or misapplied in a case of pu blic in terest. A n otable British example is the Clark case. See Nobles and S c hiff ( 2005 ) for details. Received Novem b er 2007. This is an electronic reprint of the original article published by the Institute of Mathematical Statistics in The Annals of Applie d St atistics , 2008, V ol. 2, No . 1, 432–4 3 3 . This reprint differs from the origina l in pag ination and t yp og raphic detail. 1 2 C. SPIEGELMAN ET AL. Finally , we welc ome and, in fact, encourage memb ers of the scien tific comm un it y to pr o vide alternativ e analyses of the data. REFERENCE Nobles, R. and Schiff, D. (2005). Misleading statistics within criminal trials: The Sally Clark case. Signi fic anc e 2 17–19. MR222407 7 C. Spiegelma n S. J. Shea ther Dep ar tment of St at istics Texas A&M University 3143 T AMU College St at ion, Texas 7 7843-314 3 USA E-mail: cliff@stat.tam u. edu W. A. Tobin F orensic Engineering Interna tional 2708 Little Gunstock Rd. Lake Anna, Virginia 23024-88 82 USA W. D. James Center for Chemical Characteriza tion and Anal ysis Texas A&M University 3144 T AMU College St at ion, Texas 7 7843-314 4 USA S. Wexler Humanities and Adv anced Placement Government Hightstown High School 25 Leshin Lane Hightstown, New Jersey 08520 USA D. M. Roundhill Chem Consul ting 13325 Black Canyon Drive Aust in, Texas 787 29 USA