1As already introduced in chapter four, the German universities are further classified in three different groups, which are as follows:

• Technical and non-technical universities63
• Medical and non-medical universities64
• Elite and non-elite universities65.

2This chapter now develops all relevant hypotheses regarding their collaboration behaviour for knowledge and innovation, distinguishing between the three different groups of German universities with the aim to draw differences regarding their individual behaviour. Thereby, special attention is again put on the spatial distribution of the German university co-authors with the overall aim to identify particular behavioural patterns in this regard, too.

3This chapter starts with the overall question to what extent cooperation patterns differ for the German universities with different functional orientations. Thereby, the overall hypothesis is as follows:

4Hypothesis 4: Cooperation patterns differ in case of considering the three different groups of the German universities [technical versus non-technical, medical versus nonmedical and elite versus non-elite].

5First of all, it is reasonable to ask whether there are differences regarding the number of cooperations in case of the three different German university groups. In order to underpin above hypothesis theoretically, one can, for example, draw on expenditures for research and development. In this context, the following figure indicates that there are differences regarding the amount of expenditure for R&D when different fields of studies are considered:

6As can be seen from the figure, while mathematics and natural sciences, human medicine and engineering have to stem high costs for research and development, it is reasonable to assume that they probably are more likely engaged in close network collaborations than the others are. Being aware of this, the following hypothesis is derived:

7Hypothesis 4a: The technical, medical and elite66 universities are more engaged in close network collaborations than their counterparts due to high cost of research and pooling of resources.

8First of all, it is illustrated how far the three groups of German universities have cooperated in terms of increasing numbers of co-authored publications compared to their counterparts. Being aware of the particular distribution of single- and co-authored papers, the chi-square test is applied to explore whether there are significant differences in the distinct data sets of the three groups and their particular counterparts.

9The chi-square test uses the chi-square distribution to examine whether there is a significant difference between observed frequencies and expected frequencies for the different data sets. The chi-square distribution is set up as follows:

10Overall, the chi-square test gives an indication of whether the value of the chi-square distribution, for independent sets of data, is likely to have occurred by chance alone. For all further conducted chi-square tests, the following null hypothesis is to be rejected: “the row variable i is independent of the column variable j”. Hence, the alternative hypothesis corresponds to the variables having an association or relationship where the structure of this relationship is not specified.

11For the test of independence, a chi-squared probability of less than or equal to 0.05 is commonly interpreted as justification for rejecting the null hypothesis. Thus, values of less than or equal to 0.05 are highly significant. In the course of this PhD thesis, the significance level, respectively standard errors are labeled as follows: “Standard errors: p<0.05*, p<0.01** and p<0.001*** ”.

12The SNA concerning the value of normalized degree centrality is further applied in order to verify and to explain the results of the chi-square test. Thus, it is proved whether possible differences are to be explained by the values of normalized degree centrality.

13Further, as Powell (1990) has already stated the sources of innovative capacity are mainly found between universities, research institutes and enterprises than inside them, it is also interesting to address whether there are differences in the institutional distribution of the German university co-authors regarding the three different groups of the German universities. Literature on this topic has proposed that technical universities are more likely to cooperate with industry. Organizations as TUM-TECH GmbH which is a transfer and consulting company that helps industrial partners to get access to excellent technological and scientific knowledge of the Technical University of Munich are good examples for the increased collaborations between technical universities and industry. Further, according to Mallon (2006), research centers offer a number of benefits to academic institutions in the field of academic medicine. Further, Trune and Goslin (1996) have shown that universities with medical schools and hospitals and research centers experienced the greatest profitability overall.

14Keeping in mind above consideration, the following hypothesis is formulated:

15Hypothesis 4b: The technical universities are more likely to cooperate with industrial partners due to a higher degree in applied research compared to their counterparts. In contrast, the medical universities do cooperate more often with knowledge-intensive institutions as universities and research institutes due to a higher degree in basic research-intensive bias.

16In order to shed light on the above hypothesis, a detailed institutional distribution of the German universities is shown. Hence, it is firstly pointed to possible differences within the distinct data sets of the three different groups of German universities. In order to verify the results, the chi-square test is used to illustrate whether the differences between the three groups and their counterparts are significant.

17After having presented the role of the three different groups of German universities regarding their knowledge, innovation and collaboration function, the last step of the analysis refers again to proximity patterns in this regard. Thus, it is shown whether proximity patterns differ for universities with different functional orientations. As already shown before, technical universities are more likely to cooperate with industry, which in turn has become more supra-regional, so did the elite universities due to their self image as global universities. In this context, Johnson and Lybecker (2012) have found out that within the biotechnology sector knowledge is more likely to diffuse over longer distances than it was true twenty years ago. Further, following Lu and Zhao (2012), trust plays an utmost role for medical knowledge and innovation collaboration. Hence, as trust can be best built in face to face contacts, medical universities are more likely to cooperate with regional partners.

18Being aware of above findings, the following hypothesis is developed:

19Hypothesis 4c: The technical and elite universities are rather engaged in collaborations with supra-regional partners compared to the non-technical and non-elite universities. The medical universities are more likely to cooperate with regional partners than the non-medical universities.

20After presenting the geographical distribution of the German university co-authors by illustrating the weights of regional and supra-regional located partners for each German university group, the chi-square test is taken to proof or disproof above hypothesis.

21Up to this point, it is assumed that the elite and technical universities have rather tended to cooperate with supra-regionally located partners compared to their counterparts. In contrast, the medical universities are to be more likely engaged with regionally located partners than the nonmedical universities. Besides, it has been assumed that the elite and the technical universities are more likely engaged with enterprises compared to their counterparts, while the medical universities are believed to have cooperated more often with universities and research institutes than the non-medical universities. Keeping this in mind, it is very likely that the elite and technical universities are rather engaged with supra-regional located enterprises, and the medical with regional located universities and research institutes. To shed light on this line of thought, the chi-square test is applied again. The results are displayed in the following chapter. Finally, in order to get an overview of the overall distribution of the German university co-authors according to their particular country code, it is further demonstrated how the German university co-authors are distributed regarding the five established groups of countries as introduced in chapter six. Literature on this topic has shown that medical- and technical-oriented research is rather found in the United States (US) or the European Union (EU) than in the emerging countries as BRIC. In this regard, is very likely that the medical, technical and elite universities are more likely to cooperate with partners coming from the United States than from emerging countries as the BRICs. After illustrating an overview of the country-based distribution of the German university co-authors, the chi-square test proves whether there are significant differences in the distinct data sets of the three groups of German universities and their counterparts.

22To sum up, chapter 9 provides a detailed overview and impression of the development path of the knowledge, innovation and collaboration activity of the German universities concerning proximity patterns, always focusing on the behavioural patterns of the three different groups of German universities compared to their counterparts.