The Wall Street Journal
Do not believe it. The actual results of these experiments do not support the claims of the publicity campaign. Forty-eight children were fed diets rich in either saccharin, aspartame, or sucrose (ordinary table sugar) for randomized, three-week intervals over a period of nine weeks. Transient mental effects were not measured, although 43 tests of mental function were tabulated.
Wolraich, et al. make the claim that the parents and children did not know the identities of the diets. When asked to identify which of
the three diets were being eaten, "Only one parent correctly identified the sequence of diets.'' Even if we ignore the fact that these sweeteners have markedly different tastes and we assume that only one parent per child was asked to guess, the random chance that only one or fewer of 48 parents would guess correctly a sequence of three variables is only 1 in 600.The Wall Street Journal describes this study of 48 children for three weeks on each diet by crooning, "The results were resoundingly definitive.'' Exemplary were scores the authors report for "conduct'' of 8.1 + 6.7, 8.6 + 5.8, and 6.9 + 6.1 and "attention deficit'' of 10.2 + 7.0, 8.6 + 6.2, and 9.0 + 6.1 for sucrose, aspartame, and saccharin (their "control'') respectively in school age children. The higher score is "worse,'' so in these two cases sucrose is worse. This is described as "resoundingly definitive'' in proving no effect from sucrose, because no statistically significant difference was observed. Yet, look at the errors (which represent one standard deviation and thereby the range with a two-thirds probability of containing the actual values). Too few subjects and a very brief study have given them data so scattered that almost nothing can be proved.< /FONT >
Moreover, even this meager result is opposite to their conclusions. There are 20 unequivocal behavioral parameters for school age children (the experiment involved 25 preschool children and 23 school age children which further reduced its statistical significance) reported in the paper. Of these, 14 were worse with sucrose and 6 were worse with saccharin. This worse behavior in the sucrose group is significant at the p = 0.06 level. (We calculate the probability of getting 6 or fewer heads when flipping a coin 20 times.) In short, this is a poor study of too few children for too short a time. It proves nothing, "definitive'' or otherwise, and, if anything, it gives indication that sweeteners actually did affect the children's behavior.< /FONT >
S. A. Glantz,
Circulation 61 p 1 (1980) estimated that "approximately half the articles published in medical journals that use statistical methods use them incorrectly.''Why did the New England Journal of Medicine publish this? Why not? This journal has published articles claiming that civil defense doesn't work, because everyone in the World War II firestorm areas of Hamburg and Dresden was allegedly killed regardless of whether or not he was in a shelter. When it was pointed out that, in fact, virtually everyone in the better quality shelters survived, they refused to publish this. A publication of diminishing credibility as a result of its control by such groups as "Physicians for Social Responsibility,'' the New England Journal of Medicine apparently doesn't bother with serious referees, but The Wall Street Journal should know better.
|
|
Vol. 21, No. 7
Newsletter: Access to Energy Newsletter Archive Volume: Issues Issue/No.: Vol. 21, No. 7 Date: March 01, 1994 05:38 PM (For actual publication date see newsletter.) Title: Death of a Messenger
Copyright © 2004 - Access to Energy Newsletter Archive
|