OECD bibliometric indicators: Selected highlights, April 2024
What increases (social) media attention: Research impact, author prominence or title attractiveness?
1. What increases (social) media attention:
Research impact, author prominence or title
attractiveness?
Olga Zagovora
Olga Zagovora, Katrin Weller, Milan Janosov,
Claudia Wagner and Isabella Peters
1
2. Factors affecting
Number of citations
“Paper related factors’’
quality of paper
novelty and interest of subject
characteristics of fields and study topics
methodology
document type
study design
characteristics of results and discussion
use of figures and appendix
characteristics of the titles and abstracts
characteristics of references
length of paper
age of paper
early citation and speed of citation
accessibility and visibility of papers
‘‘Journal related factors’’
journal impact factor
language of journal
scope of journal
form of publication
‘‘Author(s) related factors’’
number of authors
author’s reputation
author’s academic rank,
author’s productivity
international and national collaboration of authors
authors’ country, gender, age and race
organizational features and funding
self-citations
(Social) media mentions
Number of citations
...
title length (Zahedi &
Haustein, 2018)
title amusement level
(Subotik & Mukherjee,
2014)
Media popularity of topic
2
3. What matters for (social) media attention?
# citations
Discipline
Publication year
Affiliation
@rank
Country
@GDP on edu
Collaboration net
@PageRank
Length
# char, # words, #:
Sentiment
Polarity
Subjectivity
3
Research impact Author prominence Title attractiveness
4. Research Questions
Which features of publications are associated
with increased (social) media attention?
RQ1: What are the most effective factors?
RQ2: Which discipline-specific differences exist
between those features?
RQ3: Are the same types of publications popular
across different media channels?
(Social) media attention: Wikipedia, Facebook, News,
Twitter; Altmetric score [see also Altmetric.com]
4
6. Datasets
Publications
Multidisciplinary journal; control
for disciplines
High impact journals;
comparable papers in terms of
quality
Selected Journals:
PNAS 2004-2017
Nature Communications (NC)
2010-2017
Additional data
Web of Science (as of 07.12.2017)
DOI, Publication year, Title, Full
author names, Citation counts
PNAS website
Field of Science
Springer Nature SciGraph Data
Explorer
Field of Science
Altmetric.com (till 07.12.2017)
Altmetric score
Wikipedia
Twitter
Facebook
News
6
Discipline (OECD) PNAS NC
Agricultural sciences 221 62
Engineering & Technology 1 133 2 482
(Humanities) (0) (35)
Medical & Health sciences 12 813 2 649
Natural sciences 27 994 10 410
Social sciences 1 810 245
Total 43 921 15 883
7. Modified variables
Citation counts normalization by year &
discipline
Log transformation by Thelwall (2017); MNLCS
ln(1+𝑥 𝑖)
1
𝑛 𝑗=1
𝑛
ln(1+𝑥 𝑗)
, where 𝑥𝑖- citation count and 𝑥𝑗 is from the same
discipline and year as 𝑥𝑖.
Altmetric score normalization by year &
discipline
Normalized Log-transformed Altmetric Score (NLAS)
7
8. Regression models
Dependent var.
Media attention metrics:
Altmetric score, log-
transformed normed (NLAS)
# Wikipedia articles
# News stories
# tweets
# Facebook posts
Independent var.
Research impact
Citations counts, log-
transformed normed
(MNLCS)
Year of publication
Author prominence
First author PageRank
Last author PageRank
Title attractiveness
Length
Title chunks “:”
Sentiment:
Polarity
Subjectivity
8
10. Click to add Title – but what??
10
What matters? Nature Communication PNAS
Short titles all all
„:“ Wikipedia
Altmetric score
(e.g., Natural sciences)
1.Twitter
2. News
3. Facebook
Positive sentiment Altmetric score
(e.g., Natural sciences)
1. Twitter
2. News
-
Objective Altmetric score Altmetric score
(e.g., Natural sciences)
1. Twitter
2. News
3. Facebook
11. Co-authorship network and PageRank
11
10 papers in PNAS
14 coauthors
1 paper in PNAS
4 coauthors
1 paper in PNAS
0 coauthors
Olya JohnProf. Frink
High page rank Low page rank Lowest page rank
12. First author
12
10 papers in PNAS
14 coauthors
1 paper in PNAS
4 coauthors
1 paper in PNAS
0 coauthors
Olya JohnProf. Frink
High page rank Low page rank Lowest page rank
No significant relations
BUT :
PNAS Medical & Health
Sciences
13. Last author
13
11 papers in PNAS
15 coauthors
2 papers in PNAS
5 coauthors
1 paper in PNAS
4 coauthors
Olya JohnProf. Frink
High page rank Low page rank Lowest page rank
14. Last author
14
11 papers in PNAS
15 coauthors
2 papers in PNAS
5 coauthors
1 paper in PNAS
4 coauthors
Olya JohnProf. Frink
High page rank Low page rank Lowest page rank
PNAS Journal:
Altmetric score
overall
in Social, Medical & Health
sciences
News
Twitter
Facebook
Nature Communications:
Facebook
15. Summary
Features associated with increased (social) media attention:
Articles with shorter titles are more likely to appear in media
Polarity of title matters for
Twitter attention to NC papers – Positive
New collaborations attract media attention
News, Twitter, Facebook
Journals inconsistencies
What’s next?
Supplement feature list with
Authors' academic age and productivity level
Google trends of topics
Effect size comparison
Causality
where we studied factors affecting social media attention to research papers.
- I would like to start with brief overwiew of what have been done so far in the area. There have been dozens of studies on factors affecting n. of c. And probably that is why we are here; we all study these factors from different perspectives.- Overall, one can define 3 pools of factors. ...- But what are the factors/features affection social media attention? First of all … It have been studied that … - Mediators or confounders There ve been few studies looking into some of those factors and checking whether they affect mentions in certain media.
- Zahedi & Haustein (2018) showed that Mendeley readership counts have a small negative correlation with the title length of paper in most disciplines.
And here we also have one more factor that …In our ongoing study within the frame of metrics project we are going to identify and quantify the most effective factors affecting s. m. a. In this paper we presented the first steps with this regard.
explanatory analysis; Focused on from our … might be …most important
..take into account..Stephen Hawking; network centrality features of authors in co-authorship network ... due to multicolinearity ...
So we studied which features of p. are ...
( coverage of scientific publications varies across altmetric aggregators but also across disciplines and social media-platforms. Therefore, in-depth analysis of publication features has to consider the particular discipline or media-specific particularities. )We studied features effects and differences along 3 dimensions: dicipline; social media platforms and here were are moving to the 3rd dimension with an example
At the time when we run this study, paper on „How diversity works“ was rated as the most tranding one according to it altmetric score.It is important to keep in mind that popular science magazins, like Scientific American, might receive more media attention than any hard core scientific journal(And overtall, there might be differences in interst to certain journals.) So, controling for journal particularities is mandatory .
So, our 3rd dimension is journals.
So, which journals to select? multidisciplinary;quality; due to competitive peer reviewing process;desent amount of publications
(NC & PNAS) publications have almost 3 times stronger positive effect of citation impact on number of tweets than number of newspaper articles. Nature of encyclopedia
Lets move to our preliminary results with regards to Author related factors .Based on a publications made in PNAS and NC journals, we created co-authorship network and calculate centrality measure, pagerank.Let’s imagine that Prof. Frink
studies should control for the journal; the relationship of media attention and publication features may strongly be influenced by journal marketing strategies.Ant still there are many things that should be still done. And we are going to improve our model and supplement it with more feature …
President Donald Trump’s immigration order ; travel ban ; 7 countries
Scientific community reacted to that …
People who are different from one another in race, gender and other dimensions bring unique information and experiences to deal with tasks
This is not only because people with different backgrounds bring new information.
Interaction ; forces group members to prepare better ; alternative view points; reaching consensus will take an effort
Simply interacting with individuals who are different forces group members to prepare better, to anticipate alternative viewpoints and to expect that reaching consensus will take effort.