Sensitivity Analysis: A Necessary Ingredient for Measuring the Quality of a Teaching Activity Index

In recent years, and following the introduction of the European Higher Education Area, universities have developed measurement mechanisms to ensure improvement in the quality of their teaching and teaching staff. One of the measurement tools increasingly used in Higher Education to implement continuous improvement policies for university teaching is composite indicators, which are a mathematical aggregation of a selected set of suitably weighted indicators. Composite indicator building should be accompanied by sensitivity analysis to ensure good practice. However, this is rarely done. Sensitivity analysis helps to improve the understanding and, ultimately, the soundness of the composite. In most cases, sensitivity analysis shows that the weights assigned to indicators do not reflect the actual importance of those indicators in the aggregation to the composite because of the heteroskedasticity of, and correlation between the underlying indicators. This paper proposes a composite indicator for the teaching activity of academic staff in a Spanish university. As we shall see in the paper, the desired weights stated by developers rarely represent the effective importance of the components. Hence, we propose sensitivity analysis as a necessary tool for re-adjusting weights in order to achieve the desired level of importance for each component indicator.