The use of frontier estimation in direct marketing

✍ Scribed by Dominique Haughton; Jonathan Haughton; Alison Kelly-Hawke; Tim Moriarty

• Publisher: John Wiley and Sons
• Year: 2000
• Tongue: English
• Weight: 222 KB
• Volume: 14
• Category: Article
• ISSN: 1094-9968
• DOI: 10.1002/(sici)1520-6653(200021)14:2<33::aid-dir3>3.0.co;2-l

✦ Synopsis

A B S T R A C T

A common problem in direct marketing is to identify which physicians are the best prospects for an intervention that would encourage them to prescribe a drug. The standard procedure is to measure how far their prescribing behavior falls short of the level predicted by a regression line. We suggest that a better approach is to determine how far they fall short of "best practice," as measured by a frontier line. We discuss ways of measuring the frontier and apply the techniques to both simulated data and a live data set. The results show that frontier estimation is particularly valuable when the data are heteroscedastic, a relatively common situation.