More than the Sum of Its Parts:
Advancing Women at NJIT through Collaborative Research
Networks
The NJIT ADVANCE
Project
Funded by a grant from
The National Science Foundation
“To know who we are, we must understand
how we are connected.”*
PROJECT OVERVIEW
The NJIT ADVANCE Project pioneers
the use of social network analysis to
affect institutional change and
ensure the full participation of
women in academic science and
engineering.
enlarge women faculty’s professional networks
improve information flow
stimulate social capital formation
assess the relationship between network structure
& career advancement
create new data visualization tools to help faculty
manage their networks
transfer sustainable network analysis methods to
academic administrators
Universities are
more than
buildings….
and organization
charts….
…They are WEBS of human
interaction and perception
whose complex structure is
largely invisible to the people
embedded in them.
We make
the invisible
visible.
"The Old Boys
Network used to be a
metaphor.
Now it is a MAP….
a highway on which
we can track the flow
of social capital from
one human node to
another.”
Leveraging Social Network Data to Support
Faculty Mentoring:
Best Practices from NJIT Advance
THE GOOD NEWS ABOUT WOMEN IN STEM
“For the most part, men and women faculty in science,
engineering, and mathematics have enjoyed comparable
opportunities within the university.”
–NAS Report, 2010
"Overall, men and women are retained and promoted at the
same rate.“
-Kaminski and Geisler, 2012
“At NJIT, there are no statistically significant differences
between female and male retention. Men and women are
promoted in rank at essentially the same rate as well.”
-NJIT ADVANCE 2012
THE NOT-SO-GOOD NEWS:
The Arrow is Not Moving Toward Parity
The gender composition of the academic STEM workforce
is still profoundly different from the human population it
serves.
And likely to remain so.
IMPACT OF RETENTION FAILURES
National Data
Despite slightly improved hiring rates for women (up from 25%
to 27%) retention failures for both men and women are so
frequent…that "it may take 100 years before women are 50%
of the faculty in STEM departments.”
-Kaminski and Geisler, 2012
NJIT DATA
Tenure-Track STEM Faculty
2000: 22 women
(9%)
By 2010, 9 of those 22 women had left:
41% attrition rate
7 of the 9 (78%) left for reasons other than retirement
2010: 24 women
(11%)
10 Year Progress?
n= +2
WHY IS ACHIEVING PARITY IMPORTANT?
Because trying to solve important problems using only half
of the collective human brain is not smart!
A recent NSF-funded study shows that the collective
intelligence of a group is positively correlated with the
proportion of females in the group.
-Wooley, 2010
That is, the problem-solving ability of a group
diminishes as the number of women decreases.
RETENTION AND THE BOTTOM LINE
Academic hiring is expensive,
especially in STEM where start-up packages are large.
Failed retention costs money…
something that universities have very little of these days.
TRACKING STEM FACULTY
CAREERS
A DECADE OF DATA
FROM THE NJIT ADVANCE PROJECT
BASIC RESEARCH QUESTIONS
Does supporting collaboration
support women STEM faculty?
What is the relationship among
collaboration
productivity
retention
advancement in rank
?
ULTIMATE RESEARCH PROJECT
GOAL
To develop sustainable, scalable methods
of collecting & analyzing bibliometric data
that can be used to create
PREDICTIVE MODELS
of faculty career success
(and failure)
METHODOLOGICAL TACTIC
Using
COAUTHORSHIP DATA
as a proxy
for Faculty Research Networks
Coauthorship data is:

Public

Objective

Can be scraped from the internet or captured from
faculty CVs

Coauthorship requires a working relationship, unlike
citations or other publication measures

Publications are time-stamped… so we can track
the evolution of the faculty network over time.
21
NJIT ADVANCE researchers text-mined the Scopus digital
library to capture STEM faculty data
• Built a web crawler to search Scopus
• Retrieved 8395 faculty publications produced from 20002010—including 3608 coauthored publications
• Also captured counts of publications of NJIT faculty with
external faculty and grad students
• Compiled attributes from HR
22
Significant issues:
• NJIT faculty having same name as external
faculty/grad students (“Common Name
Problem”)
• Mislabeling of affiliations
• Different databases having slightly different
publication titles (e.g. using the word “beta”
as opposed to “β”)
23
http://www.cs.umd.edu/projects/linqs/ddupe/
24
Sustainability
Faculty at NJIT now required to enter yearly publications into
Digital Measures software (“Activity Insight”)
• Entering this data is tied to merit-based awards for faculty
Tactic:
• Use collected data to populate historical publications
• Retrieve new publications from Digital Measures
25
NJIT ADVANCE Database
514 tenured/ tenure-track faculty who worked at NJIT from 2000-2010-including a subset of 327 STEM faculty.
Attributes:
department
gender NJIT: 77 females, 437 males
NJIT STEM: 38 females, 289 males
rank progression
tenure status
hire date
separation date (if any)
retention status (left/stayed)
years at NJIT
years in the study
total number of publications
total number of coauthored publications
number of publications coauthored with other NJIT faculty
number of publications coauthored with Non-NJIT faculty
number of publications coauthored with NJIT grad students
Analytical Tools Used
For standard statistical tests:
SAS
ANOVA
To analyze network structure:
Organizational Risk Analyzer (ORA) software from
Carnegie Mellon
UCINET, a social network analysis program distributed
by Analytic Technologies
Analyzing the Relationship between
PRODUCTIVITY and CAREER ADVANCEMENT
2000-2010
“Productivity” defined as number of publications
and rate of publication
“Career Advancement” defined as retention and
promotion in rank.
Hypothesis Testing
H1. STEM faculty who publish more are
more successful in terms of rank increase.
POPULATION:
RESULT:
POPULATION:
RESULT:
All tenured/tenure track STEM
faculty (n=327)
Strongly Supported
A cohort of STEM faculty hired as
assistant professors from 2000-2003
Strongly Supported
Hypothesis Testing
H1. STEM faculty who publish more are
more successful in terms of rank increase.
Publication Rate & Advancement
POPULATION:
RESULT:
A cohort of STEM faculty hired as
assistant professors from 2000-2003
Strongly Supported
Those who did not move up in rank: average (mean) publication rate of
5.73 publications per year (SD=3.36).
Those who were promoted from assistant to associate: average (mean)
publication rate of 8.67 publications per year (SD=0.91)
Hypothesis Testing
H1. STEM faculty who publish more are
more successful in terms of rank increase.
Publication Rate & Retention
POPULATION:
RESULT:
A cohort of STEM faculty hired as
assistant professors from 2000-2003
Strongly Supported
Those who left: average (mean) publication rate of 3.46 publications per
year (SD=2.33).
Those who were retained: average (mean) publication rate of 8.78
publications per year (SD=0.94)
Analyzing the Relationship among
COLLABORATION, PRODUCTIVITY and
CAREER ADVANCEMENT
2000-2010
“Collaboration” defined as coauthorship
“Productivity” defined as number of publications
and rate of publication
“Career Advancement” defined as retention and
promotion in rank.
COLLABORATION, PRODUCTIVITY and
CAREER ADVANCEMENT
National/ International Data
"When publishing productivity is measured by...
a scientist's total number of publications,
collaboration is a strong predictor of
publishing productivity.”
- Lee and Bozeman (2005)
COLLABORATION & PRODUCTIVITY
NJIT ADVANCE Hypothesis Testing
H3. STEM faculty who coauthor more will publish
more than faculty who coauthor less.
POPULATION:
RESULT:
POPULATION:
RESULT:
All tenured/tenure track STEM faculty
(n=327)
Strongly Supported
A cohort of STEM faculty hired as
assistant professors from 2000-2003
Strongly Supported
COLLABORATION & RETENTION/ PROMOTION
NJIT ADVANCE Hypothesis Testing
H4. STEM Faculty who co-author more are more
successful in terms of rank increase.
POPULATION:
RESULT:
A cohort of STEM faculty hired as
assistant professors from 2000-2003
Supported
Analyzing the Relationship among
NETWORK STRUCTURE,
PRODUCTIVITY and CAREER ADVANCEMENT
2000-2010
NETWORK “CENTRALITY” MEASURES
Degree Centrality— The number of connections
(“ties”) a person (“node”) has. High degree
centrality indicates well-connected people who
can directly reach many people in the network.
Degree
NETWORK “CENTRALITY” MEASURES
Betweenness Centrality— reflects the extent to
which an individual has the ability to broker the
flow of information in the network.
Betweenness
NETWORK “CENTRALITY” MEASURES
Eigenvector Centrality— reflects the extent to
which an individual is connected to wellconnected people in the network.
Eigenvector
Hypothesis Testing
H5. Faculty with higher Betweenness Centrality
will publish more than faculty with lower
Betweenness Centrality.
POPULATION:
All tenured/tenure track STEM
faculty (n=327)
RESULT:
Strongly Supported
SUMMARY OF RESULTS:
For NJIT STEM Faculty (2000-2010)
Collaboration (co-authorship)
Was positively correlated with
Productivity
Retention
Promotion
SUMMARY OF RESULTS:
For NJIT STEM Faculty (2000-2010), the most
important predictors of productivity were:
1. co-authoring with non-NJIT researchers
(“cosmopolitan collaboration”);
2. co-authoring with NJIT graduate students
(the “worker bee connection”);
3. co-authoring with NJIT colleagues;
4. Betweenness Centrality.
Preferential Attachment & The Worker Bee Connection
The Case of Assistant Professor Y
Publication rate for successful assistant professors
= 8.67 papers per year
Publication rate for Assistant Professor Y
= 39 papers a year!
His secret?
Y was a post-doc—and now junior colleague--to one of NJIT’s
most prolific and well-funded researchers, “Professor Z,” a man
who from 2000-2010 produced over 150 publications, including
more than 100 co-authored with one or more of his large stable
of graduate students.
PRODUCTIVITY AND GENDER
The Case of Professor Y helps to explain one seemingly anomalous result
in our data analysis:
THE GENDER PUBLICATION GAP
NATIONAL DATA:
"Women publish between 70% and 80% as many articles as
men."
- Lariviere, 2011
NJIT STEM FACULTY DATA (2000-2010):
Female mean publication rate:
12.08 (SD=8.83)
Male mean publication rate:
19.19 (SD=13.34)
i.e. The female pub rate was only 63% of the male rate.
NO GENDER DIFFERENCE IN P&T
NJIT 2000-2003: No statistically significant difference in
retention and P&T rates for female & male STEM faculty
Assistant Professor Cohort Hired 2000-2003: No
statistically significant difference in publication rates for
successful assistant professors.
Gender
Av # Years in
Study
Av Pubs/Person
Av Pubs/Year
Female
Male
9
8.27
34.50
33.27
3.81
4.10
The most productive assistant professors, both male and
female, tended to collaborate more--and to co-author more
papers with graduate students.
SO, WHY THE OVERALL GENDER PUBLICATION GAP?
ONE ANSWER:
HYPERPRODUCTIVITY AT THE TOP:
The Worker Bee Connection
NJIT T/TT STEM Faculty 2000-2010
20 men
--5.7% of the total STEM faculty-produced 33% of all publications
A majority of these Top 20 Males are directors of one of
NJIT's large funded research centers (like "Professor Z")
or
Associated with a research center (like "Professor Y“)
HYPERPRODUCTIVITY AT THE TOP:
The Worker Bee Connection
All Top 20 Males are full or distinguished professors.
A majority already held that rank in 2000.
All NJIT Research Center directors are male.
55% of the Top 20 Males‘ papers
had NJIT grad students as co-authors.
If we include both genders, only one woman appears
in the Top 20 list of highest producers.
WHAT WE CAN DO
Leverage the mechanism of "preferential
attachment" ("the rich get richer")
IN MENTORING:
Connect new and marginalized faculty to graduate
students (and undergrads looking for research
experience)--and to senior colleagues who have large
worker bee colonies
IN GRANT FUNDING:
Fund clusters of women researchers, post-docs, and
graduate students, building the network around existing
"stars"--senior faculty who have high centrality values.
Leveraging the Benefits of
COLLABORATION
The Benefits of Collaboration
Increased productivity and advancement!
Division of labor;
Partnership with colleagues who have complimentary
expertise;
Access to expensive equipment;
Access to graduate student RAs;
Intellectual stimulation;
Collaboration among people with different intellectual
tool kits drives knowledge creation and innovation;
Access to new and novel information;
Access to tacit knowledge;
 Devil's advocacy;
Internal referring weeds out unfruitful approaches;
Safe reality checking;
Diminished social isolation;
Increased social capital.
When it works well, strategic collaboration
offers the very advantages that women
faculty need most in order to thrive:
the ability to do more high quality work
in less time
and
the ability to signal the value of their work
to the research community as a whole.
BUT…
1) Collaborations have costs as well as
benefits;
2) Not all collaborators are equal.
In social networks, as in real estate, it is
often “location, location, location” that
creates value.
"TRANSACTION COSTS” OF COLLABORATION
International collaboration involves:
time and money for travel
transaction costs resulting from differences in language and
cultural expectations
Internal collaboration across lines of ethnicity and gender involves:
cross-cultural communication issues;
Reaping the "assembly effect" (1+1=5) requires:
a sophisticated understanding of small group process;
expertise in effective project management;
experience in conflict resolution;
(and skill in psychiatric counseling?)
"Teamwork" can mean losing credit for your work;
Pro-rating papers by number of authors;
The "Matilda Effect" (Rossiter, 1993);
"Teamwork" can mean loss of the "alone time" needed for creativity.
Most of all,
collaboration involves the
"transaction cost” of
locating and assessing
potential research partners.
New online data visualization tools
from NJIT ADVANCE
reduce the transaction costs associated with
professional collaboration.
The Research Interests Map
provides an interactive display of NJIT faculty research interests and
methodologies, allowing the user to identify high-value "targets.”
The Faculty Connections Visualizer
maps actual collaborative interactions among faculty (co-authorship, co-PI
ties), enabling the user to identify efficient “pathways” to the targets.
STEP 1
Identify “Destinations”
i.e. potential collaborators who have
• Similar research interests
• Complimentary research interests & methods
• Equipment (instruments, software) they might share
• Institutional knowledge and memory
Use the Research
Interests Map Tool
(RIMap)
STEP 2
Identify “Pathways”
i.e. colleagues who can broker introductions
to the “destination” faculty you want to reach
Use the Faculty
Connections Visualizer
Tool
You
FACULTY CAREER ADVANCEMENT
NETWORK
NETWORK DATA VISUALIZATION—
A GPS FOR CAREER MANAGEMENT
THE MENTOR AS
BROKER & PATHFINDER
A Human GPS SYSTEM
CONNECTING
COLLEAGUES
TO THE RESOURCES
THEY NEED TO
SUCCEED.
The NJIT ADVANCE Strategy
Text Mining
Google Scholar/Scopus
Digital
Measures
Interface
Faculty Network
DB
Interview Data
Updates
SNA (ORA/UCINET)
Research
Interests Tool
Career GPS Tool
FACULTY MENTORING
DEPARTMENTAL MANAGEMENT
NJIT ADVANCE PIs 2010-2012
Nancy Steffen-Fluhr, PI (HUM/ Murray Center)
Katia Passerini, Co-PI (SOM/IS)
Brooke Wu, Co-PI (IS)
ADVANCE Team 2010-2012
Consultants:
Roxanne Hiltz, IS, Emerita
Anatoliy Gruzd, CS/Information Management, Dalhousie
Research Assistants:
Regina Collins (IS doctoral student)
Mingzhu Zhu (IS doctoral student)
NJIT ADVANCE External Advisory Board
Dr. Laura Kramer, Montclair University (Emerita)/ Former NSF ADVANCE
Program Director
Susan Metz, Co-Founder, WEPAN
Dr. Caroline Haythornthwaite, Director and Professor, Archival & Information
Studies, University of British Columbia
Dr. Ellen Townes-Anderson, Professor, UMDNJ Department of
Neurosciences, and director of the UMDNJ Faculty Mentoring Program
External Evaluator
Dr. Katherine Mayberry, Vice President for Special Projects, Rochester
Institute of Technology
This work is sponsored by a grant from the
National Science Foundation ADVANCE Program
(Awards 0547427 & HRD-1008549)
* Christakis, N. A., and J. Fowler. 2009. Connected: the surprising power of our social networks and how they
shape our lives. New York: Little, Brown and Co
.
The NJIT ADVANCE Project
To download this presentation and our full paper, go to
http://advance.njit.edu/