THE "BRANDEIS MODEL": DOES TIERED REFERENCE WORK?

James H. Sweetland, North American Convenor

A number of U. S. universities have experimented over the years with the idea of "tiered" reference service, where the initial user contact person is not a librarian, but is expected to refer the more difficult questions to one. While there are a number of reasons for setting up such a system, in essence the primary one appears to be a combination of the librarians� desire for more "professional" work, and the administrators� need to provide more staff without much more funding. In any event, in recent years, this has become known as the "Brandeis Model", after Brandeis University, although that institution does not seem to have invented the model.

As set up at Brandeis, there is an "information desk" staffed by graduate assistants (Brandeis, by the way, does not have a school of librarianship or information science), trained to screen questions, and answer directional and simple look-op factual questions. The professional librarians staff a "consultation office" to deal with the more difficult questions. It is telling, by the way, that the professional librarians spend only 9 hours on call in this office, and that policy is that they are to spend up to 20 minutes per question while on duty. It is also important to know that there is a formal training program for the graduate assistants, which includes regular reference to a 100 page-long manual.

While the performance of the assistants was not studied, John Staler and Marjorie Murfin (the latter one of the creators of the "WOREP"-Wisconsin-Ohio Reference Evaluation Program), found that Brandeis� librarians� overall success rate was 72%, compared to overall results for WOREP to date (for 74 libraries studies) of 57%, and a general rule-of-thumb in the U.S. of a 55% success rate for all types of reference work. An earlier study, by Douglas Herman, head of the department at Brandeis, found a similar percentage for the librarians. However, applying the now-common method of "unobtrusive observation" (persons unknown to the staff ask questions and record the results), Herman found only a 63% success rate overall, but 69% when there was a librarian on duty, and only 44% when there was not.

The causes of the results can be summarized as follows:

Librarian success appears based on having the time to work on the questions, along with having plenty of time not scheduled with patrons to work on more difficult questions. This is combined with a higher than typical use (at least in medium-sized academic libraries studied in WOREP) of the universities� own online catalog, of OCLC and RLIN (both general union catalog databases) and of computerized databases in general. Studies have also commented on the good planning, experience, and general motivation of the Brandeis librarians.

Student assistant performance is less successful mainly because, as the Brandeis study suggested, when librarians were not present, students tended to attempt to provide answers, even when policy stated they should have not done so, and the students tended to rely too heavily on the online catalog, which like most, is not complete for all library holdings.

In short, from the user�s point of view, even having a professional librarian somewhere in the reference room increases the odds of getting a correct answer by almost 50% (i.e. from 44% to 69% in the Herman study, or 72% in the WOREP study).

A number of researchers over the past thirty years have suggested that only somewhere between 10% and 20% of the questions asked in a typical library reference department are actually "difficult", or "reference" questions, the rest being simple interpretation (what does this abbreviation mean; how do I print from this OPAC), location (where is this call number), or simple factual questions-all readily answered by qualified, trained staff, who are not librarians.

With the Brandeis model, and with the university�s willingness to allow studies of its value, we now have a better idea of whether or not the theoretical two-tier system, which should work, actually does. Ironically, it would appear that the model does work, but only when there is a librarian present, and especially when the librarian answers the question.

For further reading see:

On the Brandeis model:

• John C. Stalker and Marjorie E. Murfin. "Quality Reference Service: A Preliminary Case Study." Journal of Academic Librarianship 22 (November, 1996): 423-429.
• Douglas Herman. "But Does it Work? Evaluating the Brandeis Reference Model." Reference Services Review 22 (4), 1994: 17-28.

• Loriene Roy. "Reference Accuracy." Reference Librarian No.. 49-50 (1995): 217-227.
  Summarizes the studies and comments on factors which might influence the accuracy rate.
• Matthew L. Saxton. "Reference Service Evaluation and Meta-Analysis: Findings and Methodological Issues." Library Quarterly 67 (July, 1997): 267-289.
  Performs meta-analysis on 59 studies, with comments on the overall results.

Back to Career Viewpoints