Effective Information for Managers: Using Multi-Dimensional Data Structures to support Research Management

Abstract If an organisation is to achieve its goals in the current “Information Age”, the work ofmanagers must be supported by high quality information. Systems that processorganisational performance data into managerial information commonly have eitherdimensional or relational data structures. This paper reports the findings of a study ofinformation support for university research managers and describes the informational viewafforded by the dimensional structure of data in contrast to the more common relational one.Subsequent discussion raises broader issues concerned with the effectiveness ofmultidimensional forms of information in supporting managerial work and the specificdifficulties of applying these to the problem of research management. Keywords Multi-dimensional database, relational database, research management, information support. INTRODUCTION The design of computer-based systems, to support decision making, has relied on the notionthat information of good quality, both in content and presentation, is essential for gooddecisions (McCosh & Scott-Morton 1978, Newell & Simon 1972). Our research into thework of senior managers (Hasan & Gould 1994, 1995, 1996, Hasan 1999), has confirmed thegenerally held view that their decision making processes are messy (Wagner 1995) andunstructured (Mintzberg 1989), and unsuited to support by conventional information systems.There is therefore a need to separate the day-to-day activities of production applications,using on-line transaction processing (OLTP), from processes of reporting and analysisrequired by managers, known as on-line analytic processing (OLAP). It has become evidentthat the process of management decision making is better supported by information arrangedby subject rather than by operational applications (Baum 1996). While the suitability ofrelational data structures for operational OLTP systems has been widely researched, it is lessclear what are the most appropriate data structures for subject-oriented OLAP systems. Acritical appraisal of their effectiveness is more difficult as these types of systems usuallysupport the messy and unstructured work of senior managers.The increasing demand by managers for business performance information, andorganisational knowledge, is of continued concern to IS professionals. Over the past threedecades, attempts have been made to satisfy this demand with management informationsystems (MIS), decision support systems (DSS), executive information systems (EIS) anddata warehouses (DW). The core of most MIS and DSS are relational databases, tied tooperational applications, whereas EIS and DW are often based on multidimensional data,concerned with subjects of interest to managers. The emphasis of most EIS research has beenon the functionality and data-handling capability of such systems, not necessarily on their