| In this era of fierce international competition,the performance of scientific innovation decides the strength of a country.Continuous innovation of knowledge and technology depends on the persistent development of scientific research.In the context of "Big Science",the complexity and comprehensiveness of scientific research grows with development in every scientific research field.Major breakthrough in scientific research rely more and more on the interaction and collaboration among fields,institutions and nations.Research collaboration is now a need as well as a trend in the development of scientific research.The depth and the quality of scientific research decides the quality of scientific research outcomes.Current studies on scientific research have been mostly carried out based on theory from Economics and Management,mainly focusing on the collaboration between companies or between universities and companies.However,as one of the key elements in the knowledge innovation system,universities play the role of producing and transmitting scientific knowledge.Its non-profit-oriented substance decides that its research collaboration will show different patterns and mechanism from that of private companies or industries.Moreover,existing studies merely considered the impact of geographical space on research collaboration.As a result,this study aims at studying the spatial characteristics,evolutionary rules,network structure and the dynamic mechanism of the research collaboration between universities from the different geospatial perspectives.This may help improve the spatial understanding of research collaboration between universities,expand the scopes of research collaboration,and thus promote healthy and fast growth in research performance of universities,which will,in turn,promote the academic influence and the scientific innovation strength of universities and their countries.It will be meaningful in both theoretical and empirical perspectives.This study uses the United States(the US)as a case study,since it is one of the developed nations in higher education and scientific research.This study first reviewed the history of the higher education in the US and depicted the spatial patterns of the universities,then described the spatial agglomeration of the research universities and its scientific research,based on both its academic input and output.Based on the spatial distribution of the universities,10 university agglomerations across the US have been identified using spatial clustering methods.Secondly,employing all English coauthored papers published by the research universities in SCI and SSCI database from 2004 to 2013 as the indicator for research collaboration between universities,this study selected three research collaboration networks from the macro,intermediate and micro spatial level.The macro spatial level of the research collaboration network includes all the research universities in the US,the intermediate level includes eight universities in the Greater Boston university agglomeration,and the micro spatial level includes the research collaboration network of Biomedical Engineering field in these universities.Applying both geospatial analysis and social network analysis,with the help of ArcGIS,Pajek,UCINET,Gephi and other tools in network analysis,this study characterized the geospatial change of the research collaboration intensity and the research collaboration network structures over time.Thirdly,this study explored the dynamic mechanism of the research collaboration between universities following a multi-dimensional proximity framework,including geographical proximity,cognitive proximity,social proximity and institutional proximity.Using zero-inflated negative binomial regression and negative binomial regression on the basis of the gravity model,this study detected the key influential factors on the research collaboration intensity and its change throughout the study period at the macro and intermediate level;using Stochastic Actor Oriented Models,this study stimulated the dynamic evolution of research collaboration and elucidate its mechanism at the macro level.Fourthly,this study conducted surveys and interviews across universities in the US and drawing from the Theory of Planned Behaviour,this study dug into the complex mechanism behind decision-making processes of research collaboration between universities and its supporting mechanism.Last but not least,according to the current status of the higher education and the research collaboration between universities in China,this study came up with some suggestions to promote the research collaboration,strengthen the academic ability and influence in the international scientific research from the perspectives of evaluating and appointing system,supporting system and communication platform.The results include the followings:(1)The geospatial heterogeneity of the universities in the US is prominent and varies among different spatial scales.It is even more prominent when focusing on the research universities.The research universities in the US mainly agglomerated in the populated metropolitan areas with highly developed transportation systems and economies,forming several trans-boundary university agglomerations.With the help of OPTICS spatial clustering method,10 university agglomerations were identified across the US,providing a new intermediate virtual scale for the studies of university agglomerations,which is a more flexible and reasonable geospatial perspective.(2)The spatial agglomeration of the universities in the US is coupled with the geospatial heterogeneity of scientific research and research collaboration.From 2004 to 2013,the amount of research collaboration between universities increased dramatically.More and more universities participated in the research collaboration and the research collaboration between universities were tighter and tighter.Not only in the geographical space but also in the research collaboration network,the "core-semi periphery-periphery" structure were outstanding,indicating the existence of the Matthew Effects.Seen from the spatial structure,the research collaboration network of the universities in the US gradually evolved from "there cores,one belt and four points"pattern to "one leading star,multiple undertaking cores and multiple supporting points'pattern during the study period.The members of the core groups continuously increased and so did the overall complexity of the research collaboration network.The spatial pattern of the research collaboration intensity and the research collaboration network was becoming homogenous.In the Greater Boston university agglomeration,8 research universities form a multilevel research collaboration structure,with Harvard University,Massachusetts Institute of Technology and Boston University as a triangle core along the Charles River,while Tufts University and University of Massachusetts-Boston forming the semi periphery and the rests as periphery universities.The research collaboration network of the Greater Boston Biomedical Engineering was structured with three core researcher groups as well.As seen from the network structure,although a few universities collaborated most in the network,the small-world and scale-free character of the research collaboration network at the macro level was not dominant,showing the typical phenomenon of"Club of the rich" and the polarizing trend.The number of the published papers and the degree centrality,betweeness centrality and closeness centrality of the universities were correlated,indicating the universities with high academic performance would be more active in research collaboration,control more resources,and be more dominant in the network.Meanwhile,the research collaboration network at the micro level matched the standard structure of a small-world network.But different from the macro network,the degree centrality wasn't correlated with its betweeness or closeness centrality.Based on the network structure,this study concluded four types of network structure,called Sole-core Type,Dual-core Type,Bridge Type and Growing Type.The research network normally evolves from the Sole-core Type to the Dual-core Type and to the Bridge Type,then eventually into the Growing Type.The spatial areas covered during this process is enlarging and the complexity of the network is going up in the meantime.(3)The model results implied that at all levels,a tendency among researchers to collaborate locally has been clear,meaning the research collaboration can easily happen when two collaborators are located in the same state,the same university agglomeration and in the same university.The academic performance of universities,the geographical,cognitive,social and institutional proximity between universities are beneficial for the research collaboration nationwide.Nevertheless,the influence of geographical proximity isn't significant at the intermediate level,and the role of institutional proximity isn't clear due to the limitation of the study case.In addition,the evolution of the research collaboration network is the result of three effects combined,including the network structure,the characteristics of the nodes and the proximity between them.Among which,cognitive proximity,institutional proximity and social proximity are the most important dimensions.The perspective of multi-dimensional proximity serves as a crucial framework for understanding the dynamic mechanism of the research collaboration between universities.(4)Besides the second-hand sciencemetics data,the results drawn from the first-hand survey and interview data showed that research collaboration has been an essential component of the daily scientific research activities in universities.Professors in their 30s that work in the Science and Engineering collaborate the most.In research collaboration,the researchers' positive opinions towards the research collaboration motivates them to participate;being influenced by the academic evaluating and appointing system,the aim of the research collaboration is usually to increase academic outputs;valuing the resources the collaborators have and the personality of the researchers,as well as the social and cognitive proximity between them and the potential collaborators could boost the researchers'confidence in collaboration.Notwithstanding the benefits that geographical proximity and institutional proximity bring to the research collaboration,they are not the top priority in the research collaboration decision;the supporting system for the research collaboration provided by the universities and the government could increase the researchers' willingness to collaborate,but it has great potential to be enhanced in the US.By studying the research collaboration between US universities using the framework of multi-dimensional proximity,the novelty and the potential contribution of this study are:(1)defining and providing a method to identify the university agglomerations on the basis of their spatial density,implying a new geospatial scale in the study of the research collaboration between universities;(2)depicting the geospatial evolution of the research collaboration intensity and research collaboration network,summarizing four types of research collaboration network and their evolutionary features;(3)from the multi-dimensional perspective,examine the dynamic mechanism of the research collaboration intensity and the research collaboration network;(4)understanding research collaboration between universities at the micro level using a self-conducted survey and interviews instead of second-hand scientometrics data. |
PDF Full Text Request