A book called "Other Losses" was published in Canada by James Bacque (1989). It received considerable media attention mainly in Canada and Germany since it asserted that closed to one million German prisoners of war held by the Americans and French during the World War II period, perished through the deliberate efforts of General Eisenhower to satisfy a personal hatred of Germans (a priori difficult to believe given that Eisenhower's ancestors were German). What piqued my interest was a comment by the N.Y. Times Book reviewer, S.E. Ambrose (1991) who stated that the author's statistical methodology was hopelessly compromised. It appeared that most of the purported "statistical" evidence used to buttress the author's argument that Eisenhower starved the German P.O.W."s to death depended on missing records, missing bodies, missing numbers, and lastly, missing orders on Eisenhower's part to reduce prisoner rations. The author, James Bacque, neither historian nor statistician but previously a writer of fiction, documents his case by imputing, adjusting and devising incriminating numbers, especially where data are missing to support this extremely unlikely hypothesis. Was this compromised statistical methodology? Another reviewer, A.E. Cowdry (1990), who reviewed the book in a journal of medical history and whose statistical competence exceeded that of the Times reviewer, correctly labeled it as just bad arithmetic, inability to read a table of numbers and a misunderstanding of definitions among other things. Apparently compromised statistical methodology is now social science jargon for faulty arithmetic.

Click here to download the complete PostScript technical report.
Other formats include the Macintosh Microsoft Word version of this technical report.