Abstract
This study investigates co-author and co-inventor collaborations using scientific articles and patents to measure collaborative knowledge production. This paper assesses how a scientist’s position within the joint co-publication and co-invention network affects its production and citation impact. Our findings reveal that number of publications is strongly associated with the scientists’ position in co-author/inventor networks and that a scientist’s technological production actually increases with collaboration in such networks. These academic relationships have a significant impact on the future number of publication citations and appear to benefit the number of patent citations in the same measure.
Notes
2 We extracted data originating between 1985 and 2005. Our reason for choosing that end year is that we aimed to have enough citation years after the sample’s end date, given that we were examining three periods for citations, i.e. three, five and seven years after grant year for patents. It is not uncommon to find patents that took five years to be granted, and then to count five years of citation periods up to 2015. We also chose 1996 as the start date of our sample because too few nanotechnology papers and patents could be found prior to that date. In addition, the Scopus database substantially changed around 1996 to include more journals.
3 According to a study of difference between these databases by Minasny et al. (Citation2013), Scopus metrics are found to be slightly higher than that of Web of Science. Yang and Meho (Citation2006) compared the citations of Scopus, Web of Science and Google Scholar and found that Google Scholar has some technical problems that users need to be aware of in order to accurately use the number of citations.
4 This database covers data from three agencies in Canada: Natural Sciences and Engineering Research Council of Canada (NSERC), Social Sciences and Humanities Research Council of Canada (SSHRC) and Canadian Institutes of Health Research (CIHR).
5 Our approach to developing a nanotechnology bibliometric search consisted of three steps: first, we created a pilot field scope to define nanotechnology search terms; second, we asked nanotechnology experts to modify, add and retain their field scope; and third, we evaluated these terms and tested them with publication and patent data.
6 We checked with five-year intervals and the results for our sample were the same.
7 We also added an ordered measure to our set of instruments for type of funding (Award), which equals 1 if a scientist receives funding through an award and 0 otherwise, but this did not yield significant results.
8 We chose the cross-section versions of these regression models to account for repeated observations as opposed to their panel versions, mainly because the average number of years for each observation is relatively small (only two years).
9 Vuong tests on zero-inflated negative binomial regressions favoured the non zero-inflated version of the regression models.
10 We also added the number of co-inventors, but this variable was never significant.
11 We first started with two distinct networks, but following the suggestion of a referee, we joined them because they were not independent of each other.
12 The linear effects are all positive and significant. These results are available from the authors upon request.
13 Although they do not appear to be strongly correlated, both Age and AvgFunding interact with each other, since older scientists generally raise more funding, but that relationship is non-linear. Therefore, both variables were introduced in distinct regressions to avoid their interaction. We formally tested various forms of interaction/moderating effects but none were robust enough to be included in the analysis. More research is required in this regard, but that is beyond the scope of this paper.