1) NIH (1994 v 1982) inverse-quotes itself concerning dyes & additives:
NIMH1994DecadeOfBrain states that removal of chemicals like tartrazine form ingested foods "only seemed to help about 5 percent of children with ADHD, mostly either young children or children with food allergies." And the document then goes on to liken food chemicals as like TV in being unrelated to ADHD.
The reference given to justify this summary dismissal is NIH1982Consensus, which states several times that "controlled challenge studies do not appear to have addressed adequately the role of diet in hyperactivity."
2) AACAP (1997) on tartrazine:
The official practice recommendations by AACAP (the American Academy of Child and Adolescent Psychiatry) failed to note the significant difference in test protocols and dismisses the difference in results for the tartrazine studies of Rowe1994 and of Levy1978a.
Rowe1994 finds causality. Levy1978a et al does not. This is a radical difference likely due to the tightening of test protocol by Rowe & Rowe.
Rowe1994 responders were:
83 % = 19 of 23 "suspected reactors"
27 % = 3 of 11 "uncertain reactors"
10 % = 2 of 20 "control subjects"
The AACAP refers to this as:
"A small number of children may respond negatively to tartrazine..."
Levy1978a responders were:
0 % = 0 of 22 "with parent ratings indicating responsiveness"
This Levy study is noted only indirectly via review articles by the AACAP.
3) Nutrition Foundation (1980) dismissive of strong reactor:
Weiss's ten-author paper reported a "dramatic" reactor (p=.0001) to colorings. The Nutrition Foundation was dismissive, saying "There were no teacher observations in this very young child nor were there independent evaluations by trained observers. Whether this change in behavior noted only by the parent would be noticeable to other observers cannot be answered by this study. Without this information, it is difficult to tell whether the behavioral change reported by the parent is of clinical significance." This was for a two year old.
Note: Retired Pennsylvania psychiatrist Barrett seems to have read summary papers, not originals, and missed the distinctions pointed out here. He derives a strident defense of the originals, and dismissing other results as "absurd" and "nonsensical."