network & cluster analysis of departmental libraries used by faculty at a large university...
TRANSCRIPT
Network & Cluster Analysis of Departmental Libraries Used by
Faculty at a Large University Library
JoAnn Jacoby
Associate Professor of Library Administration and Anthropology & Sociology Subject Specialist
Spiers Estate, South Africa
13-16 August 2007
7th Northumbria International Conference on Performance Measurement in Libraries
& Information Services
• Major research library – big collections (10 million vols.) & great librarians
• Decentralized library system (~40 libraries)• Pressures catalyzing change• User expectations multiplying &
increasingly diverse
Setting: University of Illinois Library
Purpose
• Map connections between departmental libraries from the users’ perspective
• Gain insight into the use of the ~40 libraries by faculty across disciplines
• Evaluate how the library’s structure might be changed to better serve faculty
Methods• Mined data from survey of all faculty,
academic professionals & staff, 2006 <http://www.library.uiuc.edu/assessment/>
• Responses to “Indicate your primary library/ (ies)” were used to create network maps and cluster trees – Respondents could choose as many libraries
as they pleased
Status
Total Head Count
Number of Respondents
Response Rate
(Est.*)
Tenure-Track Faculty 2035 413 20.3%
Visiting Faculty 172 37 21.5% Academic Professionals 2133 647 30.3%
Other (Postdocs, GAs, Classified Staff) 8763* 253 2.9%
All Faculty & Staff 13103* 1350 10.3%
Response Rate
*Distributed by a campus-mediated “massmail,” the survey was sent to 12,399 (not 13103) individuals because the headcount at the time was less than the annual maximum. The response rate was therefore slightly higher than shown.
Social Network Analysis• Seeks to reveal the patterning of
people's interaction (Roger Brown 1965)– Web of social connections key to understanding
individuals, groups & societies
• Studies patterns in the relations social actors have with particular others (Grannis 2005)– Actors can be people, firms, nations, etc.– Relations = whether actors go to war, have sex,
use a particular library, etc.
Methods
• Sociomatrix = Data for each relation presented in a two-way matrix, where the rows & columns refer to the actors making up the pairs.– Format used by most network analysis
software – Adjacency matrix = type of sociomatrix
used for dichotomous data (presence or absence of affiliation)
Methods
Setting up the matrix
MethodsTwo-mode affiliation network
1st mode = 1350 faculty & staff respondents
2nd mode = 47 libraries
“Connections among members of one of the modes are based on linkages established through the second mode”
Affiliations and Overlapping Subgroups. In Social Network Analysis: Methods and Applications. Wasserman & Faust (1994). Cambridge University Press.
Methods
Setting up the adjacency matrix- Quantifies the ties for the relation in question
Network map of connections among all libraries. Lines were drawn when >5 respondents reported
using both libraries. Selected nodes are shown in color.
Software Packages
• International Network for Social Analysis (INSA), “Computer Programs for Social Network Analysis”
<http://www.insna.org/INSNA/soft_inf.html>
– Comprehensive list of programs for data conversion, network analysis, data collection & project management
The INSA website -- a great resource
Pajek <http://vlado.fmf.uni-lj.si/pub/networks/pajek/
default.htm>Free, well-documented, Windows software
TouchGraph <http://www.touchgraph.com/>
Commercial source for software & consulting
BibApp <http://code.google.com/p/bibapp/>Maps researchers & citations 'Institutional
Bibliography'
Drawn by Hand in Adobe Illustrator (>20 only shown here)
Size classes by number of
connections
- Used natural breaks
- Basis for circle size
Results
Network maps were generated showing the connections among:
– individual faculty maps– specific departmental libraries – groupings of related
departmental libraries
Individual Respondents
Pink – Cinema Studies, Assistant ProfessorIndigo - Physics, Professor
Green - Agricultural Science (ACES), Academic ProfessionalLight Blue - Spanish, Italian & Portuguese, Associate Professor
Frequency Hisotgram of Number of Libraries Selected
0
50
100
150
200
250
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
Number of Libraries Selected
Fre
qu
ency
Specific Libraries
City Planning & Landscape Architecture Library
Modern Languages, Modern Languages, English, English, Classics & Classics & Latin AmericanLatin American
Health Science Library
Broad Disciplinary Groupings
All Divisions coded by color
Life Sciences
Area Studies Arts & Humanities
Physical Sciences & Engineering
Social Sciences
Allied
Special Collections
Central Public Services
Mean Number of Libraries Used by Division
Cluster analysis of all
libraries
Confirms some patterns in the network maps, but is of limited
utility in identifying potential mergers
Correlations with other indicators of activity level
Relationships between Number of Primary Library Designations, Circulation & Web
Transactions
So what?
Faculty don’t come in to the library anymore….
…if they can get what they need online
…..right?
Comments: Library as Place
• I mostly use the library via online access…It's especially great when working at home or doing field research abroad
• Remote delivery of materials is a great service…as my duties make it challenging to get onto central campus to find a book.
• Your questions imply that we actually go to the library and want to interact with librarians
Effect of Online Library Resources by Status
Mode of Library Use by Status
Comments: Library as Place
• Person to person assistance can be key, especially when beginning new projects or trying to navigate through new search tools….
• It would be easier to find print journals and books if the library was open more hours…
• The system of departmental libraries makes it very difficult for an interdisciplinary researcher to find and collect books s/he needs.
Research into Practice
• ongoing decisions regarding the consolidation of library service points
• revealing areas of potential collaboration– e.g., the author collaborated with the
Applied Health Sciences Library (alx) to create a health information portal after seeing evidence of a strong shared constituency
Conclusions
Network maps and cluster trees:• reveal connections and patterns
in complex systems• provide a data visualization tool
that can help inform library decision-making
ConclusionsCould be applied to other types of
data:– co-locate/bundle suites of services
frequently used or highly valued by certain groups of library users
– create browsing collections (virtual or tangible) based on “people who checked this out, also checked out”
Questions?