简介:Thepresentpaperemploystechniqueofgeographicalweightedregression(GWR)tomakeanempiricalstudyofChina'sR&Dknowledgespilloversatcitylevel.Conventionalregressionanalysiscanonlyproduce'aver-age'and'global'parameterestimatesratherthan'local'parameterestimateswhichvaryoverspaceinsomespatialsystems.Geographicallyweightedregression(GWR),ontheotherhand,isasimplebutusefulnewtechniquefortheanalysisofspatialnonstationarity.ResultsshowthatthereisasigniicantdifferencebetweenOLSandGWRinestimatingtheparametersofR&Dknowledgeproduction,andthattherelationshipsbetweenlevelofregionalinnovationactivitiesandvariousfactorsshowconsiderablespatialvariability.