Study to secure reliability of measurement data through application of game theory

Abstract

Recently, many corporations have applied CMM, SPICE or other software process maturity models to developing software products. These models are recommended to change from qualitative process management to quantitative management as they reach a certain level of maturity. Quantitative process management is possible only when reliable data exist. If quantitative process management are conducted when data's reliability are not ensured, adverse effects can be caused by the false management activities. That is why so many methods have been studied to analyze reliability of collected data. Most of them, however, are analysis methods for already collected data. Therefore, if the analysis finds that the existing data are not reliable, the data should be abandoned and new data be collected again, causing need to consume much time and effort. Therefore, as a way to secure reliability of data to be used for quantitative process management when they are input in the first place, this paper suggests a method applying the second level meta game of non-cooperative, bimatrix games to collecting data for quantitative measurement.