chapter  11
18 Pages

Regression lines and confidence limits

Inmanystudiesthecalculationofacorrelationcoefficient,inanyofthe waysoutlinedinthepreviouschapter,maybesufficientinitself.Thismay indicatethemostprofitablelinesforfurtherresearch,orprovidethedata fromwhichmapsofiso-correlationmaybedrawn.Inothercases,however,itmaybedesirabletotaketheanalysisastagefurtherbycalculating thevaluethatmightbeexpectedforonesetofdataifsomegivenvalue occursintheotherset.Thiscouldbedonebyseparatecalculationseach time,butitismoreeffectivetodrawonagraphthelinethatrepresents therelationshipbetweenthetwosetsofdata.Therequisitevaluescan thenbereadoffasrequired.Suchagraphicalrepresentationisequally valuableasadescriptivedevicetoillustratetheformoftherelationship betweenthetwovariables.