The authors of this paper previously proposed a systematic method for end users to use in testing programs in spreadsheet applications. The method, a “what you see is what you test” (WYSIWYT) interaction, was not scalable; each individual cell had to be tested independently.
In this paper, the authors build on the belief that replicated formulas need not be repeatedly tested. They propose a new method that applies two approaches. A straightforward approach allows users to rubberband a number of cells to form a region, validate one of the cells, and deem the whole region to be correct. A region representative approach further allows the tasks involved in the validation of a complex formula to be shared among different cells. Algorithms for collecting static information in a region, tracking execution traces, validating the cells in the region using the WYSIWYT method, and adjusting test adequacy have been proposed for both approaches.
The approaches prove to be effective and efficient, helping end users test their spreadsheets in a clean and tidy manner while remaining user-friendly and practical. End users should note, however, that the proposed method does not reveal every type of error. For example, suppose cell A3 is defined as “A1 + A2,” and the formula should also apply in cell B3. This requirement is usually subject to two interpretations, namely “B3 = A1 + A2” or “B3 = B1 + B2”. The proposed method cannot reveal errors caused by such ambiguities if the user only validates A3 in the region.