Applying robust regression techniques to institutional data

Regression techniques are used frequently to analyze the relationships between university activity variables and the needs for different categories of resources. The ordinary least squares method (LS) has the disadvantage of being very sensitive to outliers. As an alternative the least median of squares (LMS) technique is discussed, which can resist a large fraction of contaminated data. To demonstrate the advantages of LMS, the parameters of some regression equations, estimated some years ago by means of ordinary least squares, and describing the needs for nonacademic staff and operating funds in a university, will be reestimated by means of this robust regression technique.