Lead, IQ and social class

Abstract

Social class (SC) conveys information about a child's potential lead exposure (PB) as well as other, independent determinants of cognition (IQ). Thus, depending on the way in which SC is handled in statistical analyses, the PB-IQ association may be either 'overadjusted' or 'underadjusted' for SC. Two assumptions that underlie the inclusion of SC in correlation/regression analyses of the PB-IQ relationship are: 1) SC is an interval scale and 2) the PB-IQ relationship is homogeneous in all social strata. Simulation analyses are presented to illustrate the impact that different values of the bivariate correlations PB-SC, IQ-SC, and PB-IQ have on the estimate of the PB-IQ adjusted for SC. Alternative approaches to addressing these issues are discussed.