ction.Thismeansthatourfirsthypothesis‘A1:Full-servicecustomersaremoresatisfiedthannonfull-servicecustomers’isclearlysupported.Infact,thereisasatisfactiondifferenceofapproximately12%betweencustomerswhospendmorethan25%withothersuppliersandfull-servicecustomers.Inordertoanalysetherelationshipinmoredetail,wegrouptheactivityvariableintotwolevels:fullservice(100%)andnon-fullservice(lessthan100%).Thisisdoneinordertoreducemisclassificationwithinthetwogroups‘Upto25%’and‘Upto50%’.Furthermore,weintroduceanumberofdemographicvariablestosupplementtheEPSIRatingwhenexplainingtherelationship.Ourmethodofanalysiswillbeabinarylogisticregression.ItappearsfromTable1thatEPSI,asexpected,ishighlysignificant.ThehighertheEPSI,thehighertheprobabilityofbeingafull-servicecustomer.Italsoappearsthatbothgenderandeducation