Technology incubators and knowledge networks: a rough set approach in comparative project analysis

Technology incubators have emerged in many places as a tool in facilitating the establishment and survival of high-technology firms. Some incubators develop quickly and produce a fast-increasing number of new ventures, while others remain stable in size. Despite a growing public investment in technology incubators, systematic studies of the factors determining their growth are scarce, meaning that policy decisions are taken without sufficient practical insights into critical conditions for growth. In response to that situation, we explore several factors in determining differences in growth patterns. We use a quantitative approach derived from the field of artificial intelligence that matches with meta-analysis and qualitative (and sometimes fuzzy) dataöthat is, rough set analysis. Benefits and challenges of rough set analysis are discussed, including experience with a stepwise procedure with various accuracy checks. The findings suggest that a strong performance of incubators mainly rests on diversity in stakeholder involvement and a location in nonmetropolitan areas. Rough set analysis turns out to be a helpful tool in comparative project analysis, but there is still a need for standardization of measures used in the interpretation of the results. DOI:10.1068/b3308 ôCorresponding author. employ both local and global knowledge networks simultaneously in their learning activity (Bathelt et al, 2004; Simmie, 2004). Incubators have been addressed as an effective tool for nurturing high-technology start-up firms within the framework of learning-region policies (eg Morgan, 1997), and this is supported by various empirically oriented studies (eg Castells and Hall, 1994; Keeble et al, 1999). However, little attention has been given to factors that explain the different performance of incubators in a systematic and comparative way. Researchers from management and entrepreneurship studies mainly address the role of characteristics of incubators, including their strategy, such as the selection criteria used in the access procedure and the expertise of the incubator organization. The influence of external factors, namely characteristics of the region in which incubators are located, seems neglected. Furthermore, many studies on the development of incubators rely heavily on a qualitative analysis of a few case studies (eg Hannon and Chaplin, 2003; Mian, 1997). Studies that approach the development of incubators quantitatively are virtually absent. This feature is caused by the fact that data are scarce and, if available, often somewhat imprecise and biased (fuzzy). In the context of knowledge-based economic policies, however, there is a strong need for using systematic, transparent, flexible, and authoritative evaluation tools. Policies for knowledge-based development are facing the complexity of a multiactor situation, including the influence of quickly changing fashionable ideas about particular policy measures, and the uncertainty of a lack of understanding of how the regions' knowledge economy and particular policy measures work (Jin and Stough, 1998; van Geenhuizen and Nijkamp, 2006). In response to this situation and these needs, we attempt to identify determining factors of incubator growth using a relatively young technique derived from the field of artificial intelligenceöthat is, rough set analysis. In addition, we aim to assess the usefulness of rough set analysis as a tool in policy making to enhance growth of the regional (urban) knowledge-based economy. The following questions will be addressed: which factors determine the growth of technology incubators using rough set analysis, and what are the benefits and challenges of rough set analysis as a tool in comparative project analysis, particularly in the context of regional (urban) knowledge-based economic policy? The rest of the paper is structured as follows. First, we take a closer look at the literature on factors that may determine growth patterns of incubators and we formulate various hypotheses (section 2). Next, we examine the requirements to which an evaluation tool in the policy environment of knowledge-based economic growth should respond, and we discuss the research design of this study, including the sampling of incubators and basic characteristics of rough set analysis, particularly a step-wise approach (section 3). This is followed by an examination of the empirical results, particularly those concerning our hypotheses, and an assessment of the accuracy and robustness of our results (section 4). Finally, a summary of the analysis of incubator performance is given, and some implications for policy making on technology incubators are forwarded, including challenges of rough set analysis as a tool in spatial project evaluation. 2 Incubators and the incubation process 2.