CrossTalk Only - - - - - - Mission Staff Contact Us Subscribe Now Update Cancel

This article describes our work in developing software engineering devices for spreadsheet developers, one of the largest classes of end-user programmers. Software engineering research has focused on aiding programmers throughout the software development and maintenance process. However, this focus has been on professional programmers and has largely ignored the sizeable end-user programmer community. It is predicted that by 2005 in the United States alone there will be 55 million enduser programmers compared to 2.75 million professional programmers [1]. The programming systems used by these end users include spreadsheets, web authoring tools, scientific visualization languages, and graphical languages for creating educational simulations. It should not be surprising that a high percentage of end-user programs contain errors that can have significant economic impact. For example, a Texas oil and gas company lost millions of dollars in an acquisition deal because of spreadsheet errors [2]. In error data collected from field audit reports of real-world spreadsheets, Panko [2] reported that 20 percent to 40 percent of the spreadsheets contained errors, and errors were as high as 90 percent in some of the financial models reviewed. In empirical studies involving both experienced and inexperienced spreadsheet developers, he found that over 60 percent of the spreadsheets created by the participants contained errors. Compounding the reliability problem is the unwarranted confidence of end users that their spreadsheets do not contain errors [3]. What is surprising is that software engineering research has paid little attention to spreadsheet programmers and other end-user programmers. Our research has focused on the spreadsheet paradigm, the most widely used and studied end-use..