On the impact of the types conversion in java onto the coupling measurement
Miguel Sánchez López, Alain Abran, Grégoire Seront, Naji Habra · 2005
Software measurement represents an important topic heavily discussed within the software engineering community. Since thirty years, software measurement has become an important domain where interesting debates have occurred. Internal measurements of software do not necessitate any execution. Since these measurements are automated, it is commonly accepted that during such measurements errors cannot occur. Indeed, such measurements have no random or probabilistic aspect. The current paper aims at showing that other sources of error or uncertainty exist in the software measurement. Sources of uncertainty can appear before the measurement itself, that is, at the measurement design level. Indeed, mistakes related to the design of measurement can occur, and therefore affect the measurement results when executing the measures. The current paper extends the notion of uncertainty to the measurement design level, and highlights the impact of the design uncertainty onto the measurement result.