Academic social network sites (SNS) are booming. A recent large-scale survey published in Nature indicates that almost 90 percent of researchers in science and engineering and more than 70 percent in the social sciences, arts and humanities are aware of ResearchGate – next to Academia.edu the largest academic SNS with more than 6 million users. However, only limited research has been carried out on academic SNS. Although a vivid community creates and implements alternative measures of scientific impact with social media data, little use has been made of the potential of academic SNS as a data source. Consequently, few studies employ person-based metrics that cover users’ social capital in the form of structural indicators and network statistics (centrality, density, homophily, clustering). This contribution draws on extensive data from ResearchGate to address this issue and add a relational component to altmetrics research. It includes a follower/following network of 302 nodes on ResearchGate: the complete faculty of a Swiss public university who are members on this academic SNS as of early 2014. We describe the overall network with classical metrics of social network analysis and compute the centrality of each individual node. Results indicate low density, high institutional homophily, a skewed degree distribution and many isolates. We then compare the structural properties of individual nodes with other metrics of influence. To do so, the network data is complemented with detailed attribute data, such as department affiliation, gender and position within the university hierarchy. Moreover, we collect researchers’ activity on ResearchGate, bibliometric information, webometrics and altmetrics, i.e., the prominence of their publications on general and specific social media platforms. We evaluate whether the relational aspect of influence in the form network centrality correlates with activity, bibliometric, webometric and almetrics indicators as well as personal attributes. Significant and intermediate correlations between activity and centrality are found, while the correlations between centrality and bibliometric as well as altmetrics are weaker but still significant. No significant correlations between webometrics (coverage of publications on general social media platforms, like Twitter and Facebook) and network centrality occur. The analysis suggests that network centrality is distinct but correlated with (bibliometric) output metrics and therefore worthy of inclusion in future altmetrics studies.

1. Connected for Success?
How Network Centrality on ResearchGate Relates to
Bibliometrics, Altmetrics and Webometrics
Christoph Lutz
INSNA Sunbelt Conference
27th of June 2015

2. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 2
Background:
Social Media & Science1

3. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 3
What do we know about social media in science?
• A few studies on general social media
use (Gruzd & Goertzen, 2013; Procter et al. , 2010)
• Some research on academic blogs and
Twitter in science
• Active research field «altmetrics»
• Very few empirical studies about
academic SNS

4. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 4
Popularity of «altmetrics» on Google Trends

5. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 5
Current research on altmetrics
• Comparing 11 webometric indicators with WOS citations: correlations found for
Twitter, Facebook wall posts, research highlights, blogs, mainstream media and
forums; not enough evidence for Google+, LinkedIn, Pinterest, Q&A sites and Reddit
(Thelwall et al., 2013)
• Mendeley readers a relatively good altmetrics indicator for citations later on
(Haustein et al., 2014; Mohammadi & Thelwall, 2014)
• Very few studies about ResearchGate (e.g., Thelwall & Kousha, 2015)

6. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 6
Network study by Jordan (2014) about academia.edu

7. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 7
What about social capital and relations?
• Relations with other researchers are
important resources
- Promotion
- Publication
- Invitation
- Collaboration…
• Social capital matters
 Bringing in the relational aspect in
impact measurement… beyond
citations

8. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 8
Framework
• Resonance of publications
• Resonance of the person
• Public perception
Altmetrics
Influence
Webometrics

9. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 9
Sample
• 302 users from Swiss university
• 68.5 percent male, 31.5 percent female
• 40 percent PhD, 41 post-doc/assistant
profs and 16 percent full profs (rest:
admin)
• Collection of data in March 2014

10. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 10
Scientometrics 2.0:
Results2

11. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 11
N=885
0
100
200
300
400
500
600
700
Mendeley readers
0
50
100
150
200
250
300
350
400
Scopus citations
0
2
4
6
8
10
12
1
37
73
109
145
181
217
253
289
325
361
397
433
469
505
541
577
613
649
685
721
757
793
829
865
citeulike bookmarks
1. Altmetrics

12. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 12
2. Influence: ResearchGate network metrics
Activity and network of the
researchers, attention,
resonance of individual publications.
Person:
#Followers, #Followees, #Q&A,
#Profile Views, #Impact Points/
RG Score
Publications:
#Papers, #Uploads, #Views,
#Citations, #Requests,
#Downloads

13. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 13
44 = Sven Reinecke
29 = Miriam Meckel
55 = Marcus Schögel
117 = Tobias Jenert
54 = Joachim Schelp
23 = Jörg Hofstetter
8 = Christian Hoffmann
10 = Christoph Lutz
80 = Michael Lechner
185 = Monika Bütler

14. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 14
Results match with academia.edu study
Jordan
(2014)

15. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 15
Schools
= SOM
= SEPS
= LS
= SOF
= SHSS

16. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 16
Position
= Full professor
= Assistant professor
= Post-Doc/Procect leader
= PhD student
(No Isolates and Pendings)

17. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 17
Network statistics
• Density: only 3 percent of all possible connections realized
• It is a relatively small community: average distance = 3.12;
diameter = 9
• Internal networking most prevalent : On average, members
of the UniSG have 16 followers, 50 percent of which are
UniSG colleagues. There are a couple of outliers and isolates
• High homophily along institutes

18. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 18
3. Webometrics
• We analyzed mentions of publications on:
 Wikipedia (3 quotes)
 Blogs (6 quotes)
 Facebook (Comments on public pages with reference to
4 articles)
 Twitter (227 tweets)
 Delicious (100 bookmarks)

19. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 19
Distribution of number of tweets for articles
0
5
10
15
20
25
30
1
19
37
55
73
91
109
127
145
163
181
199
217
235
253
271
289
307
325
343
361
379
397
415
433
451
469
487
505
523
541
559
577
595
613
631
649
667
685
703
721
739
757
775
793
811
829
847
865
883

20. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 20
Correlations between indicators
Position1
Publication
success2+
Publication
resonance4+
Centrality3+
Altmetrics5+ Webometrics3+

21. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 21
Summary
Activity Centrality
Publication
success
(h-Index)
Publication
resonance
(RG)
Position Altmetrics
Centrality 
0.3
Publication
success

0.2

0.15
Publication
resonance

0.3

0.35

0.7
Position X 
0.2

0.5

0.45
Altmetrics X 
0.15

0.65

0.6

0.35
Webometrics X X 
0.35

0.2

0.15

0.25

22. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 22
Conclusion
• Altmetrics have a strong correlation with traditional measures of impact.
• The correlation of centrality measures with traditional measures of scientific
impact is weaker. A dynamic perspective might be necessary here.
• Impact can barely be «forced» via communication efforts. It has to emerge from
the recognition and reaction of other users.
• Webometrics only have a weak correation with traditional impact measures (in line
with current bibliometric studies, e.g., Costas et al., 2014).
 Altmetrics and webometrics allow more differentiated view of impact.

23. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 23
Thank you for your attention!

24. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 24
References
Costas, R., Zahedi, Z., & Wouters, P. (2014): Do “altmetrics” correlate with citations?
Extensive comparison of altmetric indicators with citations from a multidisciplinary
perspective. Journal of the Association for Information Science and Technology (JASIST),
early view. Online: http://onlinelibrary.wiley.com/doi/10.1002/asi.23309/abstract
Haustein, S., Peters, I., Bar-Ilan, J., Priem, J., Shema, H., & Terliesner, J. (2014): Coverage
and adoption of altmetrics sources in the bibliometric community. Scientometrics,
101(2), 1145-1163. Online: http://link.springer.com/article/10.1007/s11192-013-1221-3
Jordan, K. (2014). Academics and their online networks: Exploring the role of academic
social networking sites. First Monday, 19(8). Online:
http://firstmonday.org/ojs/index.php/fm/article/view/4937/4159

25. Sunbelt 2015
Connected for Success?
Christoph Lutz
Page 25
References
Mohammadi, E. & Thelwall, M. (2014): Mendeley readership altmetrics for the social
sciences and humanities: Research evaluation and knowledge flows. Journal of the
Association for Informations Science and Technology (JASIST), 65(8), 1627-1638. Online:
http://onlinelibrary.wiley.com/doi/10.1002/asi.23071/abstract
Priem, J., Taraborelli, D., Groth, P., Neylon, C. (2010). Altmetrics – A Manifesto. Online:
http://altmetrics.org/manifesto/
Thelwall, M., Haustein, S., Lariviére, V., & Sugimoto, C. R. (2013): Do Altmetrics Work?
Twitter and Ten Other Social Web Services. PLoS One, 8(5): e64841. Online:
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0064841#pon
e-0064841-t003
Thelwall & Kousha (2015): ResearchGate: Disseminating, communicating, and
measuring Scholarship? Journal of the Association for Information Science and
Technology (JASIST), 66(5), 876-889. Online:
http://onlinelibrary.wiley.com/doi/10.1002/asi.23236/abstract