university-industry research collaboration13/bozemantalkjune2013.pdf-.147 collaborations with...
TRANSCRIPT
University-Industry Research Collaboration: Findings from the Individual Level
Barry Bozeman School of Public Affairs
College of Public Programs
Phoenix, AZ 85004-0685
Presentation to I/UCRC Evaluator Meeting
Arlington, VA, June 6, 2013
Acknowledgments
The research reported here is based on data developed in three projects supported by the National Science Foundation: “Assessing R and D Projects’ Impacts on Scientific and Technical Human Capital Development” (SBR 9818229, Barry Bozeman, PI); “Collaboration in University Research Centers (SBR 98-18229, Barry Bozeman, PI), Department of Energy Basic Energy Sciences “Research Value Mapping” (DE-FGO2-96ER455562, Barry Bozeman, PI) and “NSF CAREER: University Determinants of Women’s Academic Career Success” (REC-0710836, Monica Gaughan, PI) and. The author gratefully acknowledges the support of the National Science Foundation and the Department of Energy. Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
Collaboration Studies: Past, Present, Future
• Past: Collaboration Strategies and
University-Industry Collaboration
• Present: Collaboration Outcomes
and “Nightmare Collaborations”
What is Research Collaboration?
• Organizational level: university-industry, public and
private laboratory, multiple institutions.
• Individual level: co-authors, knowledge producers and other research enablers (lab assistants, technicians).
• “Mixed”: collaboration patterns of individuals from different institutional settings.
Different approaches to Studying
Individual Level Collaboration
1. Publications based (e.g. Lee and Bozeman,
2005) or patents based (e.g. Dietz and
Bozeman, 2006).
2. CV-based (e.g. Gaughan and Bozeman, 2002).
3. Questionnaire Based (e.g. Bozeman and
Corley, 2004; Bozeman, 2010).
The Data:
Survey of Academic Researchers
• Questionnaire data supplemented with CV data and website data (2004-2005, supplemented in 2009).
• Questionnaire sent to 4,916 targets, yielding 1,794 respondents after three waves of administration (adjusted response: 37%).
• Our target population was tenured and tenure track STEM faculty members in Carnegie research extensive (Research I) universities.
• Women were over-sampled. Results reported here either employ statistical weights or statistical controls for sex.
The Data: “Collaboration Ethics and Practices”
• 77 semi-structured interviews with university researchers
• N= 641 University Researchers in “Research I,” Carnegie Extensive Universities. Stratified sample (by field, gender).
• On-Line questionnaire and bibliometric analysis. • Focus on “most recently published” article
– Provides anchor – Assuming randomness, gives good indication of incidence and
propensities
• Also questions about career experiences in collaboration.
Question 1: Who Does NOT Collaborate?
Answer: Nearly Everyone Collaborates at Some Point
(98% for Tenured STEM faculty), but:
Not everyone spends the same amount of
time collaborating.
Explaining Percentage of Time Working Alone
The Loners: Oldest cohort, math/computer field, untenured, and…male. NOT research center affiliate,
Question 2: Who are the “Cosmopolitan” Collaborators?
• Physicists and chemists
• Lower percentage of time teaching
• Higher percentage of time grants supported
• Male
What does Cosmopolitan mean???
Who are the “Cosmopolitan” Collaborators?
Who are the “Cosmopolitan” Collaborators?
Male
Higher percentage of time grants supported
Lower percentage of time teaching
Physicists and chemists
Collaboration Criterion Scale Scale Origin: Bozeman
and Corley, 2004)
Note: Men and Women Differ
Findings Related to Research
Centers
Measurement: University Research
Center
• URC Affiliation: responses to questionnaire item, member of center with at least 10 researchers, at least two disciplines, not part of on academic department.
• Questionnaire responses AND audited by website check.
• About 32% affiliated with URC
Measurement: Industrial Involvement Scale
• Responses to questionnaire items about types of industrial involvement (e.g. patenting, working at a business, collaborating on article).
• Weighted, additive scale, with weights based on the residual (e.g. if 10% patent, score value 90; if 30% exchange research, score value 70).
– Applications of scale: Lin and Bozeman, 2007; Boardman, 2009; Bozeman
and Gaughan, 2007; Boardman and Ponomariov, 2011
Items for Industrial Involvement Scale (Bozeman and Gaughan, 2007)
Findings: Number of Collaborators
Findings: Number of Collaborators, by Sex
Men and women share predictors: • being grant active • affiliating with a multidisciplinary URC • greater industrial involvement. • following a mentoring strategy
Men experience collaboration gains via three collaboration strategies: Instrumental, Experience and Mentoring. With women, gains only with Mentoring strategy.
Current Study: Bad Collaborations
• What do researchers think of as “bad” or
“good” collaboration?
• Do early bad experiences have long range
effects on careers and propensity to
collaborate?
• When the weak are exploited (e.g. not
credited, insufficiently rewarded) what
conditions encourage exploitation?
• Ghost authors in STEM.
( Inexact) Estimate of Percentage of Articles
including Industry Co-Author
Revised Estimate:
Approximately
9%
Discipline Group % Including Industry Co-Author
Math, Computer Science 9.8%
Economics 5.6%
Biology 8.3%
Physics and Chemistry 4.1%
Engineering 19.4%
Propensity to Include at Industry Co-Author,
by Discipline Grouping
Is there a Gender Gap?
Bad Collaboration Outcomes
No Yes
Some DESERVED co-authorship, did not receive
91.5% 8.5%
Someone DID NOT DESERVE co-authorship but received it
97.3% 2.7%
A co-author made NO contribution at all
98.5% 1.5%
Bad Collaboration Outcomes
Tau Correlation with Industry Co-Author
Some DESERVED co-authorship, did not receive
Not Significant
Someone DID NOT DESERVE co-authorship but received it
Not Significant
A co-author made NO contribution at all
-.147
Collaborations with Industry LESS likely to
have Ghost Authors