Measuring Non-standard research impact?

advertisement
Mendeley Readership Altmetrics
for Clinical Medicine and
Engineering?
Ehsan Mohammadi1, Mike Thelwall1, Vincent Larivière2, Stefanie Haustein2
E-mail: e.mohammadi@wlv.ac.uk
1University
2Université
of Wolverhampton. UK.
de Montréal, Montréal, Québec, Canada.
Citation is incomplete
• Limited
• More
to authors’ perspectives.(Schloegl
&
Stock, 2004).
•
appropriate
for
theoretical
publications.
3-5 years times are needed for receiving
citation
• Citation indicators are not able to give the
full picture of research impact
Usage Metrics are tasteful but..
• They mainly employed local usage data.
• Downloaders are unknown.
• Data aggregation is not easy.
Altmetrics as a solution?
Altmetrics is a new movement which tries to find complementary
measures for traditional metrics based on scholars’ activities in
social web platforms (Priem, Taraborelli, Groth, & Neylon, 2011).
Data collection is faster.
Data is more accessible.
Data coverage is global.
Diversity in data type (not limited to authors)
Why Mendeley?
• Massive users
• Diversity of users
• Huge size of database
• Open API
• Global coverage
Research Question
•
•
What proportion of Clinical Medicine and Engineering articles
are covered by Mendeley database?
Are there significant, substantial and positive correlations
between Mendeley readership counts and citation measures
in Clinical Medicine and Engineering specialties?
Method
• Montreal university’s in-house version of the Thomson-ISI databases were used for data collection.
• Based on NSF classification, the most productive specialities of engineering and clinical medicine were selected
• All bibliographic information + citation data of the journal articles of the year 2008 were downloaded.
• 145,536 for clinical medicine and 109,390 for engineering.
• Statistics data related to Mendeley readership for the WoS articles were extracted using the Mendeley API.
• The WoS data set and Mendeley readership data were matched and duplications were removed.
• Spearman correlation tests were applied to the ISI citations and Mendeley readership counts.
% of WoS articles fro 2008 from Clinical Medicine and Engineering specialities
in Mendeley
90
80
70
60
50
40
30
20
10
0
Unique WoS articles covered by Mendeley
Duplicated records in Mendeley catalogue
Articles with readership statistics in Mendeley
Findings
•
•
•
We considered the coverage in Mendeley for different
specialties based on the available unique records in Mendeley
catalogue.
Clinical Medicine articles had the higher coverage (71.6%) in
comparison to Engineering and Technology (33.7%) papers.
1.5% of the overall founded records of both Engineering and
Clinical Medicine were subjected to duplication.
Spearman correlations between WoS citations and Mendeley
readership counts (non-zero only) for 2008 Clinical Medicine and
engineering articles
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Findings
•
•
•
•
There is a significant correlation between Mendeley
readership and citation counts in all the investigated
specialities.
The correlation for clinical medicine overall (r=.561) is higher
than for engineering (r=.501).
Cancer (r=.604) and Materials Science (r=.682) had the
highest correlations among clinical medicine and engineering
specialties.
Surgery (r=.451) and Computers (r=.414) had the lowest
correlations among Clinical Medicine and Engineering and
Technology specialties.
Comparison of citation and Mendeley readership median for
2008 Clinical Medicine and engineering articles
WoS citation median
Mendeley readership median
Findings
•
•
The median Mendeley readership counts were higher than the
median citation counts for mechanical and computer
engineering papers.
This is due to that mechanical and computer engineering
papers were read more and citied less.
Limitations
•
•
Readership is limited to the individuals who choose
Mendeley for their reference manager.
Our studied sample is restricted to journal articles while
conference papers are important document types in
engineering disciplines.
Conclusions
•
•
•
In almost all disciplines, the correlation is not strong enough
to conclude that Mendeley readership and citation counts
measure the same aspect of research impact.
A likely explanation is that Mendeley captures broader
scholarly activities from a variety of readers’ perspectives in
comparison to citation counts.
Hence, Mendeley readership data could be a useful
supplementary measure to remedy some limitations of citation
analysis for some applied specialities.
Download