18.3 Why Spreadsheets?

Since their first appearance in the late 1970s spreadsheets gained a remarkable popularity in business as well as research and education. They are the most common software managers, researchers and traders utilize for data analysis, quantitative modelling and decision support.

Przasnyski and Seal (1996) find that most of the time series modeling done in the business world is accomplished using spreadsheets. Further research work suggests that those users have become so proficient with spreadsheets that they are reluctant to adopt other software solutions. Not even a higher suitability for specific applications is appealing then (Chan and Storey; 1996). An analysis of XploRe download profiles conducted in a data mining framework confirmed our perception of the statistical software market. A stunning majority of users are using Excel for statistical analysis (Sofyan and Werwatz; 2001).

A major difference between a spreadsheet application and statistical programming languages is the interaction model. This "direct manipulation interaction model" enables statistical computations e.g. by drag and drop (Neuwirth and Baier; 2001). In the cell based framework of spreadsheets the direct aspect of interaction means, that manipulations in one cell immediately effect the content of another cell. Of course this is only the case if the regarding cells are interconnected with appropriate cell functions and references. Especially in business applications the immediate visibility of numerical changes when cell values are modified, is an appreciated feature for decision making based on different scenarios.

Our approach is based on the philosophy to bring two ideals together: On the one hand an accurate and reliable statistical engine (which is represented by the XploRe Quantlet Server, XQS) and on the other a user friendly and intuitive Graphical User Interface like Excel. The numerical impreciseness of Excel has been discussed exhaustively in the literature. As a starting point the reader is referred to e.g. McCullough and Wilson (1999). The development of M D *ReX is guided by the principles of usability and flexibility. Hence we try to offer different modes of usage for varying user groups: a dialogue based and more "Windows" like appearance for the novice user of statistical software and a "raw" mode where the spreadsheet application merely functions as a data import wizard and scratch-pad for the advanced user.