The data that is available suggests that the pattern for virtual online reference questions is the opposite of that reported for in-person, telephone, and e-mail reference. With a lower percentage of Level I–II questions (less than 50–60 percent), it appears that more complex questions are received through virtual reference than through in-person, telephone, or e-mail reference. This experience suggests that the Warner model and categories could apply to newer forms of reference service (e.g., instant messaging and text messaging) and requires further study. DeGroote’s study on centralizing digital reference service for multiple locations and establishing benchmarks is an example of a possible future research direction.16
Data gathered from fall 2005 was compared with data from fall 2006 to determine trends in the following two areas: the recorded level of business and the effect of telephone questions. A reduction in the number of reference questions following the fall 2003 opening of the library was expected, since both Balas and DeGroote reported declining levels of questions since the 1990s while ARL libraries reported a 34 percent reduction.17 MLK Library reference data suggests that not only has the reduction leveled off, but that the number of in-person questions is rising during peak semester months. Table 6 shows that the total number of in-person and telephone reference questions (7460) is similar to the number recorded in fall 2005 (8068), with only a 7 percent reduction.
This is good news, since fall 2005 data had shown a 16 percent reduction from 2004 use figures (9712). Only December 2006 shows fewer overall questions, particularly in Level I questions, a reduction of over 20 percent when compared to December 2005. The number of in-person queries was up slightly in October and November 2006 (4 percent and 6 percent, respectively). Figure 3 shows that the ratio of Level I–II questions to Level III–IV questions remained consistent, but that the percentage of telephone reference questions shrank from 18–23 percent to 14–16 percent.
Discussion and Conclusions
The Warner model was devised to support and plan the transition from a dual-desk setup to a one-desk setup in a health sciences library. It clearly applies to other reference operations and types of libraries. The adaptation of the Warner model at the MLK Library confirms that the definition of questions by level, with each level requiring more sophisticated methods for answering them, is appropriate to the MLK Library and would be useful in almost any library. The examples used for assigning questions to a level can be customized to reflect the type of library or clientele.18
The Warner model provides an effective way of evaluating the effectiveness of services because it provides a method for counting questions by difficulty and determining who can best answer them. It has also led to the increased efficiency of the reference desk, a successful tiered integration of librarians, library staff, and student assistants, all of whom utilize the same method of evaluating and answering questions. And this has, in turn, informed adjustments in optimal staffing levels. Future research can take a look at the question evaluation and staffing data in the context of determining if the Warner model has also led to better service for patrons, but this is beyond the scope of this article.
Librarians and support staff may require additional training in differentiating Level I–II and Level III–IV questions. In August–September 2005, student assistants and support staff received special training in tiered reference and the recognition of Level I–II questions. This training resulted in greater accuracy and consistency in the coding. The training of librarians and staff in the question levels and their participation in the ongoing definition and coding of questions can only help to make the ongoing evaluation of services more effective.
The MLK Library in-service, during which personnel from General Collections participated along with Reference personnel in a coding exercise, demonstrates that what has been learned at the reference desk can be applied to other service points for a greater consistency and accuracy building-wide. The experience at the MLK Library and in the library literature shows that training, particularly of paraprofessionals, is the key to maintaining quality of service and to reducing the number of situations in which the paraprofessional has to decide the category of a query “in the absence of detailed guidelines and without a chance to consult a professional.”19
The general ratio of 80–20 between the numbers of simpler (Levels I–II) and more complicated (Levels III–IV) questions found by Warner holds true in other libraries. Further testing would be needed to determine if it remains a constant for all reference services. This validates the Warner study and indicates its overall appropriateness for libraries of various kinds.
The goal of tiered reference is to maximize the skills of librarians and support staff by using the appropriate level of staff to answer different levels of questions. In the MLK Library, the information desk is centrally located on the first floor of the library. It was originally hypothesized that the information desk would siphon off all the Level I–II questions, leaving the rest to be handled by the reference desk on the second floor or the general collections desk on the third floor. This did not happen. Given the complexity of a nine-story library building, library patrons go to the nearest service desk to ask questions. The library as a whole, however, is working to achieve greater coordination by creating cross-training modules involving these different units.
Within the context of the Reference unit, the assignment of librarians to answer phones as their primary duty was changed to answering the Level III–IV questions referred to them by paraprofessionals or student assistants. The change in staffing also provided more desk support for those times when there was greater demand from patrons for assistance. Closer analysis of data by hour of the day could further fine-tune staffing decisions. Such analysis could identify times and days within specific months when the demand for librarians to answer Level III–IV questions is highest.
In addition, other factors will have to be considered in the application of the Warner model—the time it takes to answer reference questions, the differing conditions in libraries that are highly multicultural, and the consistuency that each library serves (health science, university, K–12, public, etc.). Inclusion of such factors in future research will certainly lead to the refinement of the Warner model.
The Warner model was adopted in the early part of the planning process that led to the establishment of this joint public and university library reference service (one aspect of the merged MLK Library). It defined the ongoing development of the reference service and helped to create a context in which public and university librarians and staff could work together to respond effectively to the needs of all patrons. It showed that there is definite overlap in the kind of information demanded by university and public patrons and aided in the achievement of efficiencies in staffing and training within the merged reference unit. It will provide a basis for the ongoing evolution and improvement of reference service in the MLK joint-use library.
Harry C. Meserve is Associate Librarian, Dr. Martin Luther King Jr. Library, San Jose State University. Sandra E. Belanger is Librarian, Dr. Martin Luther King Jr. Library, San Jose State University (retired). Joan Bowlby is Senior Librarian, Dr. Martin Luther King Jr. Library, San Jose Public Library. Lisa Rosenblum is Director, Hayward (Calif.) Public Library. Submitted for review September 18, 2006; revised and accepted for publication October 15, 2007.
Reference & User Services Quarterly, vol. 48, no. 3, pp. 247–258 © 2009 American Library Association. All rights reserved. Permission granted to reproduce for
nonprofit, educational use.
[...] Developing a Model for Reference Research Statistics: Applying the “Warner Model” of Reference Q… [...]