Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Implications for diet disease relationships.

Sir—Dr Tong suggests that we should have retained all the variables in the regression equation in which we assessed the contributions of potential confounders in the relation between mental development index (MDI) and type of infant feeding. Feeding data came from computerized hospital data obtained at discharge from hospital and from the Health Visitors' notes (the Health Visitor is a specially trained nurse who sees the child at prescribed times during the first two years after birth). We did in fact look at the effect of the excluded variables on the relation between MDI and feeding and could find none. The excluded variables were clearly