A first look at Germany, the UK and Japan
Hanna Hottenrott
TUM School of Management
Technische Universität München
Cornelia Lawson
International Research Fellow of the
Japan Society for the Promotion of Science
Kassel, June 2016
Multiple Institutional Affiliations in Academia
• Affiliation to an institution linked to
o identity and prestige (Di Leo, 2003; Long, 1978; Long and McGinnis,
1981; Fox, 1983)
o creates bonds but also exclusion (Di Leo, 2003)
o resource access and research infrastructure (Stephan, 2012)
o collaboration and opportunities (Melin and Persson, 1996)
• Institutional affiliations highly influential in academic culture
• But what about multiple institutional affiliations?
2
Why are affiliations interesting?
3
• Universities want to attract leading talent (assessment / competition)
e.g. Universities in China or Saudia Arabia created special part-time
positions to attract leading foreign scholars (Xin and Normile, 2006;
Bhattacharjee, Y. 2011)
• Researchers want to increase opportunities
e.g. additional resources, visibility, network, option to return to academia
• Side effects of academic employment market
e.g. documented by increase in part-time and fixed term positions amongst
junior academics
Country and field-specific factors may affect the extent and the forms of
multiple affiliations
4
Why do we see multiple affiliations?
• What is the extent of multiple affiliations in
• bioscience, chemistry, engineering
• Germany, Japan and the United Kingdom?
• What types of multiple affiliations are most common by
• institution type combination
• institutions’ geographic locations?
• Are publications with multiple affiliated authors different in terms of
impact through
• network
• visibility
• access to resources or
• selection?
5
In this talk
• Journal publications (articles, proceedings, book chapters) from web of science
• Making use of the new author institution tag available since 2008
6
Data
• Journals selected based on 2013 journal citation report
• Sorted by eigenfactor score, a rating of journal importance based on the
number of incoming, journal-weighted citations
• Bottom 50% discarded
• Five journals randomly drawn from each quartile of the eigenfactor
distribution for each scientific field
20 journals per field stratified by eigenfactor score
• Manual checking of addresses
• All articles with authors in Germany, Japan or UK
• Period 2008 to 2014 (citations until April 2016)
7
Sample selection
• 29,582 items
• 36,035 authors with at least one address in Germany
• 57,604 with an address in Japan
• 31,648 with an address in the UK
• Semi-manual coding of institutions by type
• Maximum number of addresses observed for one author is seven
Analysis article-based
• author count
• funding acknowledgement
• citations received up to April 2016
8
Data
9
Share of articles with multiple affiliation author(s)
10
Affiliations by type of institution
Cross-sector affiliation articles as share of articles with multiple affiliation
author(s)
11
How international are the multiple affiliations?
Share of articles with multiple affiliation author(s)
12
University-university affiliations by location
Japan Germany UK
0
10
20
30
40
50
60
70
International Cross-Univ Domestic Cross-Univ
13
University-PRO/NGO affiliations by location
Japan Germany UK
0
5
10
15
20
25
30
35
40
45
50
Domestic Univ+Foreign PRO Domestic Univ+Domestic PRO
Foreign Univ+Domestic PRO
14
International affiliation articles as share of multiple affiliation articles
15
What about impact?
• Measured as the number of citations
• We take the log (plus the unit) to normalise the highly skewed
citation distribution
highly field and year sensitive: we follow Lee et al. (2015)
and consider papers that are in the top 1% (10%) of citations in
their field in each year as papers with high scientific impact
• Only 761 articles in our sample are in the top 1%, we therefore also
consider the top 10% as an alternative measure (5533 articles)
16
Citations per publication
Discipline Country
t-test of mean
difference
Single-affil only Multi-affil
Bioscience Germany 28.14 49.24 ***
Japan 9.53 20.81 ***
UK 30.48 38.58 *
Chemistry Germany 24.73 27.38
Japan 12.08 20.69 ***
UK 22.28 20.35
Engineering Germany 7.81 10.00 *
Japan 7.14 9.32 *
UK 8.59 11.20
17
Citation distributions
Kerneldensityestimatesofthenumberofcitationsbyaffiliationtype
18
What about impact?
• Accounting for other factors explaining citations
19
OLS citations counts
20
Probability to be in top 10%
21
Dependent:Top10%cited Japan Germany UK
Domesticmultiaffil 0.034*** (0.009) 0.054*** (0.013) 0.010 (0.016)
Internationalmultiaffil 0.068*** (0.012) 0.071*** (0.019) 0.038** (0.019)
Reference:Engineering
Bioscience -0.059*** (0.023) 0.112* (0.058) 0.037 (0.044)
Chemistry -0.023 (0.025) 0.055 (0.055) -0.032 (0.054)
Reference:JournalQualityQuadrant1
Quadrant2 -0.050 (0.063) -0.066 (0.072) 0.001 (0.060)
Quadrant3 -0.073 (0.068) -0.135* (0.072) -0.134** (0.062)
Quadrant4 -0.120** (0.053) -0.070 (0.061) -0.143*** (0.053)
Authorcount 0.007* (0.003) 0.015*** (0.004) 0.016*** (0.004)
Fundingacknowledgement 0.054*** (0.019) 0.123*** (0.034) 0.141*** (0.038)
N 11886 9594 9076
Probability to be in top 10% by country
22
Dependent:Top10%cited Japan Germany UK
Univ&Univ 0.035*** (0.011) 0.045*** (0.015) 0.030 (0.019)
Univ.&Industry 0.029* (0.016) -0.001 (0.031) 0.006 (0.041)
Univ.&PRO/NGO 0.063*** (0.015) 0.094*** (0.015) 0.046** (0.022)
Univ.&Other 0.018 (0.034) -0.029 (0.047) 0.035 (0.045)
Reference:Engineering
Bioscience -0.063*** (0.022) 0.111* (0.058) 0.036 (0.044)
Chemistry -0.024 (0.025) 0.055 (0.055) -0.033 (0.054)
Reference:JournalQualityQuadrant1
Quadrant2 -0.048 (0.063) -0.066 (0.072) 0.001 (0.060)
Quadrant3 -0.071 (0.068) -0.135* (0.071) -0.134** (0.062)
Quadrant4 -0.119** (0.053) -0.071 (0.060) -0.144*** (0.053)
Authorcount 0.007* (0.003) 0.015*** (0.004) 0.016*** (0.004)
Fundingacknowledgement 0.053*** (0.019) 0.122*** (0.034) 0.141*** (0.038)
N 11886 9594 9076
Probability to be in top 10% by country
• We find that multiple affiliations have increased in all fields and countries
• reported by authors on 23% of academic papers in 2014, up from 10% in
2008
• Cross-sector affiliations are higher for Germany and Japan, reflective of the
stronger public research sectors in both countries
• Cross-country affiliations are highest for the UK
• Articles with multiple affiliation authors receive more citations which may be
indicative of their higher impact
• Especially international multi affiliations and
• cross sector and with public research organisations
• … more insights soon …
23
What’s new?
24
Thank you for letting me get
to this slide!
Contact:
Hanna Hottenrott
TUM School of Management
Technische Universität München
hanna.hottenrott@tum.de