A testing time for spreadsheets

Despite users being dangerously error-prone, training in best practices is rarely given.

But are these mistakes that matter? Are they of the “don’t let the perfect be the enemy of the good” type; or more like “oops, earnings restatement”?
To investigate, the Dartmouth trio asked five organisations to provide five spreadsheets each. The error taxonomy was applied and the impact of each mistake estimated. The files of one organisation, identified as a “small consulting company”, were described as “works of art: thoughtfully designed, well documented, easy to understand, and error free”. The mistakes in half the files of two big companies, however, led to calculation errors of more than $10m. The biggest impact was just over $110m. But these companies also each had one error-free spreadsheet. Within a single organisation, “spreadsheet practice can range from excellent to poor”, say the researchers. This, if anything, is the ugly truth about spreadsheets: how good you are at them, and how likely you are to make serious mistakes, is a function of the culture and practice in which you learnt to use them. In the absence of formal training, at university or later on, this seems unlikely to change.