This document examines the relationship between reference transactions at ARL libraries from 2006-2011 and factors relating to access (service points and hours) versus awareness (instruction sessions and participants). Using regression analysis on data from 91 ARL libraries, the study found that the numbers of instruction sessions and participants had a statistically significant relationship to reference transactions in most years, while service points and hours did not. Each additional instruction session correlated to about 30 more reference transactions, and each participant to 1-2 more transactions. So instruction and awareness, not just access, better predicted reference demand over this period.
Access or Awareness: Identifying Relationships between Reference and Other Dimensions of Public Services
1. ACCESS OR AWARENESS?
IDENTIFYING RELATIONSHIPS BETWEEN
REFERENCE AND OTHER DIMENSIONS OF
PUBLIC SERVICES
#ACRLPUBSERV
Amanda L. Folk
University of Pittsburgh
alfolk@pitt.edu
http://pitt.libguides.com/ama
ndafolk
2. THE BIG QUESTION
Do the number of instruction sessions, participants in
instruction sessions, service points, and service hours
predict the number of yearly reference transactions
between 2006 and 2011 at ARL libraries?
Which is more important in terms of yearly reference
transactions…
access (service points, service hours)?
awareness (instruction sessions, instruction session participants)?
3. COMPETING HYPOTHESES
C. Paul Vincent, “Bibliographic
Instruction and the Reference
Desk: A Symbiotic Relationship,”
1984.
Library instruction is not enough
to create an independent, self-
reliant library user in light of the
increasingly complex information
landscape.
MORE INSTRUCTION=MORE
REFERENCE
E. Stewart Saunders, “The Effect
of Bibliographic Instruction on
the Demand for Reference
Services,” 2003.
Library instruction should create
an independent, self-reliant
library user.
MORE INSTRUCTION=LESS
REFERENCE
WHO’S RIGHT?
4. METHOD
The sample includes data collected between 2006 and 2011 for 91
ARL Libraries…
university-affiliated,
in the United States,
with complete data for all variables examined.
Ordinary Least Squares (OLS) regression analysis for each year
examinedDependent Variable
Reference
transactions
Independent Variables
Instruction sessions
Instruction session
participants
Service points
Service hours
Control Variables
Total library
expenditures
Full-time student
enrollment
Number of faculty
5. WHAT IS OLS?
OLS is used to…
Understand the relationship between a dependent variable (e.g. reference
transactions) and independent variables (e.g. instruction sessions, service points)
and/or control variables (e.g. enrollment)
Make predictions about the dependent variable based on the best-fit line determined
in the regression analysis
Limitations of OLS include…
Only works well for linear relationships
Sensitive to outliers and multicollinearity
Cannot determine causality
7. FINDINGS
AWARENESS!
In both the “regular” (Models 1a and b) and log-transformed (Models 2a and b)
models, the numbers of instruction sessions and participants in those sessions
were statistically significant in almost every year examined.
The numbers of service points and service hours were not statistically significant in
any of the four models.
Based on 2011 data,
Each additional library instruction session could result in just over 30 additional
reference transactions.
Each additional library instruction participant could result in 1 to 2 additional
reference transactions.
The overarching question driving this research study is… In other words, do we see a statistically significant relationship between reference transactions and these other aspects of public services?
A secondary question of this study is…
This study is not meant to defend traditional reference services; innovation and change are imperative, as studies have shown that the traditional reference model is not cost effective, especially in light of declining numbers of transactions (Ryan, 2008). Despite this, educating our patrons, both formally and informally, remains a critical component of what all academic libraries strive to do. Understanding if and how various aspects of public services are related provides some evidence when making critical decision about changes to public services models.
I do provide a brief literature review in my papers, and towards the end of that literature review I highlight two competing hypotheses regarding the relationship between reference transactions and other aspects of public services.
BRIEFLY STATE HYPOTHESES
Saunders study—ARL data from 83 ARL libraries between 1995 and 2000. IV—participants, CVs—numbers of service points and service hours, full-time students, and faculty
Who’s right? So far, empirical research, including Sander’s study, has indicated that Vincent’s hypothesis holds.s
Canadian universities were not include (because of the use of expenditures) and libraries like LOC, NYPL, BPL, and NLM were not included. Any libraries that were missing any data for the variables included in the models for any of the years of investigation were not included (8 libraries were dropped)
LOOK UP FULL-TIME STUDENT ENROLLMENT—WHAT DOES ARL ASK FOR?
Sample size
Independent variables
Control variables are not usually included in write-up, because they weren’t the focus of the study. We include them because we know or we think they are important to understanding the dependent variable.s
R-squared
This is also called the coefficient of determination. This tells us how much of the variance in reference transactions is explained by our regression model. For example, this model explained 33% of the variance in reference transactions in 2011. This means that there are other variables that contribute to the variance in reference transactions that were not included in the model because the data wasn’t available or because these variables can’t be quantified.
Coefficient and standard error
The coefficient tells us how much this variable (the number of instruction sessions) affects the number of reference transactions, after all of the other variables have been taken into account. If we were to run a regression with only instruction sessions, this coefficient will likely be much bigger (thus making it seem as if instruction sessions have a much bigger relationship with the number of reference transactions. The coefficient is what we use to make predictions. The absence of a minus sign indicates that this is a positive relationship—as instruction sessions go, reference transactions go up.
The standard error is the standard deviation of the coefficient and indicates the precision of the coefficient. The smaller the standard error, the better in terms of prediction. In terms of prediction, we would see that an increase in one instruction session would lead to an additional 33 reference transactions plus or minus 11. The coefficient and the standard error are used in determining statistical significance.
Significance levels
My quant training came from an education econometrician, so I use t values instead of p values. They can generally be interpreted in the same way. The smaller the percentage level, the more confident we can be that there is a significant relationship between the dependent and independent variables. So what we really like to see are those two asterisks.
Rough predictions…not precise! Remember the standard error!
As we move forward and imagine what public services models should look like in the future, more nuanced data is crucial for making evidence-based transformations. While annual tallies of non-directional reference transactions can help us gauge the extent of informal teaching and learning opportunities, this data is not good enough. Susan M. Ryan’s study of the cost-effectiveness of a traditional reference desk, as well as other studies, indicate that most reference questions do not need to be answered by a librarian. Academic libraries, including both ARL and non-ARL libraries, need to start tracking and reporting meaningful teaching and learning opportunities through reference services in order to have valuable discussions about the future of reference services and the value they add to the learning experience.
Debra G. Warner, in consultation with her colleagues at Eastern Carolina University, devised a new classification system that they deemed to be more appropriate to the provision of reference services in the twenty-first century. Two of the four categories Warner proposes are related to the idea of opportunity teaching for librarians—strategy-based and consultations—and do not include the more mundane, though important, non-directional questions like “How do I make the header different on the first page?” or “How do I access JSTOR from off-campus?” Rather than bemoaning the general decline of reference transactions in general, academic librarians should be thinking of ways to maintain or even increase strategy-based and consultation reference transactions, since these provide valuable, one-on-one, point-of-need teaching and learning opportunities. If ARL and ACRL were to collect data related to strategy-based questions and consultations, it would provide a more accurate picture of the informal teaching that librarians provide and the value that they add to the academic experience.