asuring rates of adoption of innovation and how they change over time (Mansfield et al, 1977). However, new approaches have emerged in more recent years, with evolutionary, path-dependency and industrial districts theorists all emphasising different factors in the diffusion process (see Lissoni and Metcalfe 1994, for a review of these developments). Many economists are coming to recognise that any single theoretical approach will only explain part of the picture. As Jacobson and Androsso-OCallaghan (1996, p.176) have recently argued, "human, social and even cultural factors are increasingly being accepted as important in technology, innovation and diffusion", and innovation economists are being drawn beyond the traditional boundaries of their own discipline in the search for fresh insight.Towards a framework for multi-level, interdisciplinary researchThe thematic convergence across levels and disciplines reflected in the foregoing review points us towards the kind of framework within which multi-level, interdisciplinary research on innovation might be usefully developed, based on systems thinking, learning theory and dynamic analysis.The first conclusion that we can draw from the foregoing review is that the innovation process can no longer be fully understood at any single level of analysis. This is perhaps most dramatically illustrated in the developing literature on systems of innovation, where very clearly the nature of innovative activity at firm level is seen to be inextricably linked to institutional influences operating at national and sectoral levels. Increasingly, we consider firms in less discrete and atomistic terms in the context of industrial agglomerations like clusters, districts and filieres, with innovation emerging through supplier partnerships, strategic alliances and other forms of inter-organisation relationships (Jacobson and Androsso-OCallaghan 1996, pp.116-122). In short, the study of innovation increasingly req...
