ultsagesto.Weexcludedotherracesbecauseofsmallnumbers,aswellasimputed,unreliable,orsurrogatedata.Resultsarebasedonweighteddata,permittinginferenceaboutthetotalBlackandWhitenoninstitutionalizedUSpopulation.Groupmeansandstandarderrorswerecalculatedusingsoftwareappropriateforcomplexsamplesurveydata.StandarderrorsarefrequentlylargeforBlacks,duetotheirsmallnumberinthesurvey.Therefore,resultsforBlacksshouldbeviewedwithcaution.Conversely,standarderrorsforWhitesaresmallduetotheverylargesamplesize,sothatsmalldifferencesinintake,whilenotbiologicallymeaningful,couldbestatisticallysignificant.Dietaryinterviewswereconductedbyinterviewerswithaknowledgeoffoodpreparationandnutrientcomposition.Ameasureofportionsizewasobtainedusingthree-dimensionalfoodmodels,aswellasdishesandglasswareofvarioussizes.Subjectswereaskedtoreportallfoodanddrinkconsumedinthe-hourperiodbeforetheinterviewday.AlmostallrecalldaysmicsResearchDivision,AgriculturalResearchService,