Food colors and behavior.

meant by the term "'problem behaviors." Nowhere was the subject population clearly described. The strategy of allowing each parent to pick items from five different behavioral checklists leads irrevocably to a case study approach; that is, 22 "studies," each with an N of 1. Such an approach precludes the possibility of deriving scientific generalizations. Moreover, each of the behavioral rating scales alone, administered according to standardized procedures, has known reliability, validity, and factor loadings. What was the validity of using these diverse items in nonstandardized combinations? And what kinds of items did the parents select as being "aversive' or "positive"? Such information was lacking. That teachers' observations were discarded when they proved "undependable" was an unfortunate design decision: teachers would probably have been less biased than the parents who had already put their children on the "Feingold diet." It is also disturbing that in the one child who responded "dramatically" to the color challenges, the mother as data collector knew when her child had received the challenge, ostensibly because of the child's behavior, but other hypotheses cannot be ruled out, for example, that the mother discovered some physical symptom in the child or some telltale sign in the beverages that provided clues to the presence of the colorings. In addition, the time between ingestion of the drink and parental ratings was not rigidly controlled, but allowed to vary randomly over a period of up to 3.5 hours after consumption of the test beverage. Thus, potentially valuable data on the time course of the responses under investigation were, apparently, not collected. At best, the data of Weiss et al. suggest that sensitivity to food coloring is a rare phenomenon: at worst, the methodological problems prevent them from adding to our understanding of the issue of food colors and behavior ROBERT L. BRUNNER CHARLES V. VORHEES RICHARD E. BUTCHER Institutefor De e Children's Hospital Research Foundation, Cincinnati, Ohio 45229 References the beholder. Behold: Swanson and 1. B. Weiset al., Science 27, 1487 (1980). Kinsbourne (5), in the same issue of 2. M. A. Lipton, C. B. Nemeroff, R. B.Mailman in Nutrition and the Brain, R. J. Wurtman and J' Science, reported adverse effects offood J. Wurtman, Eds. (Raven, New York, 1979), dyes in 17 of 20 hyperactive children (85 vol.'4, pp. 1-27. percent). And the largest study of the JuneNw1O"o *& ~all ten boys in a preschool sample imBrunner et al. (1) ascribe to us (2) a positio di.not adop, te tt it. proved on the diet (100 percent). Animal position we ddnstudies also reveal behavioral toxicity We never claimed to be testing the therafrom low doses of food dyes, especially peutic efficacy of the Feingold diet. Nor in lonates o d,. did we claim to be estimating population in nenae (7. Some criticisms by Brunner et al. are parameters. We wrote: due to a hasty reading of the report. For