An Experimental Study of Spreadsheet Presentation and Error Detection

Spreadsheet Error Research Several well-founded concerns exist about the integrity and validity of electronic spreadsheets. One hundred thirteen MBA students sought eight errors planted in a single-page spreadsheet to discover if differences in the presentation format would facilitate error-finding performance. Five presentation formats were used. Spreadsheets were presented on the screen, both with and without formulas. Spreadsheets were also presented on paper with a list offormulas attached, or without formulas. A new integrated forinula paper treatment was introduced, with formulas presented in each cell directly under each calculated value. Subjects found, on average, only about 50% of the errors across all presentation formats. The on-screen treatments were clearly inferior to the paper treatments, whether or not formulas were presented. Practitioners should be aware of the d&Gxlties in jnding even simple errors, especially on-screen, and should develop training programs to facilitate spreadsheet auditors ’ performance. Many of the widely-reported problems in the management of end-user computing [I, 16, 251 are symptomatic of inadequacies in assigning roles to end-users and developers [ 1.51. Users face a confusing environment [6, 23, 321, employing tools that are often harder to use than programming languages [34]. Perhaps the ease with which users can perform basic functions [22] creates overconfidence in performing advanced functions or performing basic functions on a large or complex spreadsheet. Indeed, Brown & Gould [7] report users’ strong self-confidence in their spreadsheet work. Confidence aside, laboratory and field studies [7, lo] demonstrate that somewhere between one-third and one-half of all spreadsheets contain errors. It is relatively easy to find reports of serious errors in spreadsheets sampled from actual practice [9, 111. Error severity can be quite high; errors in actual spreadsheet usage are known to have caused losses of hundreds of thousands [21] to millions [ 11, 131 of dollars.