Grand Valley State University is a public liberal arts university in Michigan with its main campus in Allendale, additional campuses in Grand Rapids and Holland, and regional centers in Muskegon and Traverse City. The university serves over twenty thousand undergraduate students and more than three thousand graduate students in more than two hundred areas of study. The GVSU libraries consist of three libraries: the Zumberge Library located on the main campus in Allendale, the Frey Foundation Learning Center located in the Cook-DeVos Center for Health Sciences in Grand Rapids, and the Steelcase Library located on the Pew Campus, also in downtown Grand Rapids. The Zumberge Library serves the majority of GVSU’s undergraduate programs, particularly liberal arts and sciences; the Frey Foundation Learning Center serves the health sciences programs; and the Steelcase Library serves primarily professional programs, including criminal justice, social work, business, and education.

The Steelcase Library, built in 2000, is part of a larger building complex called the DeVos Center located on the Pew Campus in downtown Grand Rapids. The one-story library consists of a reading room for quiet study, a reference area, an instruction lab, a service desk, an automated retrieval system, and a staff area, including librarians’ offices. The Steelcase Library was built to accommodate the traditional library service model of separate service and reference desks. Librarians staffed the reference desk located in the reference area, and library staff and student workers staffed the service desk.

In addition to face-to-face reference, GVSU’s libraries offer virtual reference services and consultation services. At the time of this study, the university libraries offered virtual chat reference, but librarians no longer staffed this service. Instead, service desk staff answered these inquiries. The change in this staffing model was the result of a detailed study of chat reference at the university libraries that showed that use of chat reference was consistently low.⁶ In addition, because of the liaison librarian model used by GVSU libraries, librarians frequently receive email questions from student and faculty members in their liaison areas.

Since it opened in 2000, the librarians staffing the reference desk at the Steelcase library recorded reference transactions on a paper form using hash marks. This method of keeping reference statistics, however, only allowed for the collection of basic quantitative data, not qualitative data. Librarians were able to record only the date, timeframe the question was asked, (e.g., between 8 a.m.–9 a.m.), and whether the question was “directional” (specifically asking for directions) or “desk” (all other types of questions).

Beginning in fall semester 2008, GVSU libraries abandoned the hash mark system and began to record reference data in LibStats, an online tool that was developed to enable librarians to collect both quantitative and qualitative reference statistics.⁷ Using LibStats, librarians were able to record both the questions asked and answers given at the reference desk, providing a richer level of detail than the hash mark system. The software also recorded the exact date and time this data was entered. Librarians now could evaluate and assess not only the number but also the nature of and the exact time queries were made at the reference desk.

Some librarians believe that (for their learning communities, at least) the value of in-person, professional help at the reference desk is great enough to justify staffing desks with librarians. This position is supported by Gratz and Gilbert, who asked students where they went for help in their academic library. Nearly two-thirds of the responding students reported that they approach the reference desk or a reference librarian for help, while less than a quarter ask for help at the service desk. Based on the results of this study the authors concluded that these usage rates justified maintaining a traditional reference desk.⁸

Other librarians have opted to allocate their time to instruction and other endeavors, relying on front-line staff to refer students and faculty with complex reference questions to librarians. For example, Sonntag and Palsson provide a background of the arguments for removal of the reference desk at California State University, San Marcos, which freed up librarians to provide more in-class instruction.⁹ Kennedy describes the rationale for and the mechanics of the discontinuation of the reference desk at the Homer Babbidge Library where an information commons model replaced what the author referred to as the “antiquated reference model.”¹⁰ In a presentation provocatively titled “Redesign Your Reference Desk: Get Rid of It!” Meldrem, Mardis, and Johnson chronicle Northwest Missouri State University’s library’s transition to a tiered reference system where librarians are on call instead of at a desk.¹¹

Among those academic libraries that have removed their reference desks, few have contributed in-depth assessments of those changes to the body of library science literature. Sonntag and Palsson describe the statistics that student “information assistants” and on-call librarians kept regarding reference desk questions. All questions asked at the desk were recorded and tracked to help understand student need and the level of service that was being provided.¹² The authors do not, however, discuss the number of questions asked and whether those numbers changed with the evolution of reference service. Kennedy, while including a rationale and process for discontinuing a reference desk, does not include an assessment of the service change.¹³ Meldrem, Mardis, and Johnson’s assessment of the change to a tiered reference system was limited to satisfaction surveys of library users and surveys of staff.¹⁴ While useful and vindicating in their case, this is not quantitative evidence of success.

Arndt, however, does provide more in-depth data on a reference transformation in her 2010 case study of reference in the Dickinson College Library. The study chronicles how the library staff responded to internal statistics indicating low reference use (less than one question per hour), with only 66 percent being actual reference questions.¹⁵ Librarians at Dickinson decided to remove the reference desk completely and to instead schedule on-call librarians to be available for reference queries. To ensure that reference continued to be available to faculty and students, they sought to reframe reference as a service, unshackling it conceptually from the physical presence of the desk. Their effort to maintain reference availability included marketing their reference services and thoroughly and consistently training access services staff for effective and collaborative referrals to subject librarians. Dickinson’s librarians assessed the change one year later, using reference use statistics, number of visits to the “Ask a Librarian” website, and a student survey. All measures indicated success. For the purposes of this literature review, the reference use statistics are most notable. According to Arndt, the number of true reference questions (combined appointments and walk-in consultations) dramatically increased by 37 percent while nonreference questions answered by librarians (not surprisingly) decreased by 66 percent.¹⁶ Arndt is careful to point out that the success is due in large part to characteristics specific to that library. Nonetheless, the case study presents strong evidence that removing the reference desk, accompanied by the right measures, can actually lead to an increase in actual reference questions.

Dinkins and Ryan studied the questions that a paraprofessional referred to an on-call librarian at Stetson University to assess whether more extensive staffing of the reference desk by paraprofessionals could be successful.¹⁷ After analyzing the reference questions referred by the paraprofessional to an on-call librarian and comparing the numbers to the total number of questions the paraprofessional handled, they conclude that “the reference desk could be staffed by a trained paraprofessional,” and “librarians and the library administration are pleased with the results”and believe that the referral system is working.¹⁸ Interestingly, the total number of reference questions during the paraprofessional’s shifts decreased dramatically and more markedly than the modest decreases seen in the other shifts. Over the year of the study, only about 7.4 percent of the total questions were referred to a librarian.¹⁹

Some scholarship has illustrated how libraries can use data to make informed decisions about reference staffing. Notably, Ryan analyzed 4,431 questions over four study periods to find, among other things, that 89 percent of the questions asked at the reference desk could be answered by students or trained staff.²⁰ Ryan used salary data to calculate the amount of each transaction cost; noninformational direction and machine questions (the most common transaction) cost the library $17,919 over 2,528 transactions.²¹ Dennison, of Winona State University, also approached the question of reference desk staffing using statistical evidence. There, librarians collected reference statistics over several years then analyzed each hour to determine whether to staff, double staff, or not staff the reference desk with librarians.²²

No research studies have examined whether or how patrons’ reference questions are answered when no one is staffing the reference desk in an academic library. This study analyzes the number and nature of questions answered by librarians at a reference desk and compares them to the number and nature of questions answered by librarians who are on call in order to determine which reference staffing model best meets the needs of its patrons: a staffed reference desk, an on-call reference system, or a combination of both.

The purpose of this research was to determine:

1. the nature and number of questions asked at the Steelcase Library reference desk and what percentage of those were reference questions according to the classification scheme the researchers developed to analyze them;
2. the nature and number of questions asked at the Steelcase Library service desk that prompted a staff member to summon the on-call librarian to answer; and
3. patterns of reference activity to inform reference scheduling in the Steelcase Library.

This study covers a period of seven academic semesters, beginning spring/summer semester 2009 through the end of winter semester 2011. During this time, librarians working at the Steelcase Library recorded all reference questions they were asked in LibStats. The LibStats interface allowed Steelcase librarians to record whether they were at the reference desk or on call when they answered a question. For purposes of this study, only the questions themselves were examined. The answers to these questions were not part of this study and were not evaluated.

The sample of questions for this study was the entire population of questions recorded by Steelcase librarians in LibStats for the study period (N = 1,567). All questions recorded during the study period were analyzed and rated according to the classification system described in table 1.²³

The researchers analyzed the questions recorded in LibStats using content analysis. Content analysis is based on the premise that groups of words that comprise a document or other written communication “can be reduced to, or organized into, categories in which words or word units…share the same central meaning or connotation.”²⁴ Based on this technique, the researchers developed a system to analyze the content of the questions asked at the Steelcase reference desk. They devised a four level classification system similar to that used by Bravender, Lyon, and Molaro to code the questions.²⁵Table 1 shows the classification system used to analyze the data.

Questions classified in the first category could be answered without reference to library resources, including those that were computer-skills based. Questions in this category could be answered by anyone working in the library or would be referred to the appropriate university department. The second category of questions were those seeking known items or brief information that could be found with minimal searching in the library’s catalog or databases or through a Google search, i.e., ready reference questions. The third category included questions that required strategy-based searching in the library’s resources. Such questions usually required appropriate source selection, development of search terms, and critical evaluation of results. A fourth category was created for questions that were related to citation formatting or the use of bibliographic management software.

Based on the classification system described in table 1, both researchers independently categorized the questions from one semester. They then compared their results and further refined their mutual understanding of the classification system. The researchers then independently coded all seven semesters of questions and compared their ratings. Differences in coding were resolved by the mutual agreement of the researchers.

Tables 2 and 3 show the staffing models used in Steelcase Library over the study period.²⁶

As shown in tables 2 and 3, during the first three study semesters, spring/summer 2009, fall 2009, and winter 2010, librarians extended on-call hours in the morning by one hour each semester. They were physically present at the reference desk for the rest of the day until 6 p.m. Prior to the start of the spring/summer 2010 semester, and based on statistics from the 2009 spring/summer semester which indicated very low reference activity, it was decided that the reference desk would not be staffed during the spring/summer 2010 semester. Instead, librarians were scheduled to be on call only.

Throughout the fall 2010 and winter 2011 semesters the reference desk was again staffed from 11 a.m. until 6 p.m. During these semesters the researchers implemented a system to study and compare reference activity at the desk during what were the highest-use reference times as based on earlier recorded reference statistics. During fall semester 2010 from 11 a.m. to 2 p.m. and 4:30 p.m. to 6 p.m., the reference desk was staffed on alternate weeks, that is, it was staffed on week 1, week 3, week 5, etc. and on-call week 2, week 4, etc. During winter semester 2011, the reference desk was staffed on the opposite weeks from the fall semester, that is, it was staffed on week 2, week 4, week 6 and on-call week 1, week 3, etc. During the spring/summer 2011 semester, the reference desk was not staffed and all reference was on-call.

This study analyzed a total of 1,567 questions (N = 1,567) asked of librarians at the Steelcase Library and recorded in LibStats over seven academic semesters from the beginning of spring/summer semester 2009 through the end of the spring/summer semester 2011. The inter-coder reliability was found to be Kappa = 0.8487. According to Landis and Koch this is almost perfect agreement.²⁷

Table 4 shows the total number of questions in all categories that Steelcase librarians answered during the study period.

In the fall 2009 semester, the reference desk was staffed for a total of 592 hours. In the fall 2010 semester, the reference desk was staffed for 168 hours, about a quarter (28.4 percent) of the time it was staffed in fall 2009. A total of 460 questions were answered in fall 2009 and 257 in fall 2010. The numbers of category 1 and 2 questions, which were defined as not requiring the services of a librarian, fell during the fall 2010 semester, during times when the reference desk was not staffed and librarians were on call. Most significantly, however, the number of true reference questions (category 3) answered by librarians remained almost the same: 86 in fall 2009 and 91 in fall 2010.

Questions about citation formatting and citation software (category 4) continued to be asked of or referred to librarians at a steady but low rate regardless of whether the reference desk was staffed or not.

Table 5 shows the number of reference (category 3) questions that were asked during the fall and winter semesters of the study period by week. Shaded numbers represent times when no one was staffing the reference desk and a librarian was on call, and nonshaded areas represent times when the desk was staffed. During fall 2010 and winter 2011, when the desk was staffed on alternate weeks, the average number of reference questions answered by on-call librarians was the same as the number answered when the reference desk was staffed.²⁸

The researchers expected to find that overall reference activity at the Steelcase Library reference desk had declined since 2009, and data collected during this study confirms this. Furthermore, analyzing alternating weeks of librarians on desk and on call revealed to the researchers that when a staff member is visible at the reference desk, the librarian staffing the desk answers more directional and ready reference questions. However, the researchers were surprised to find that, at the Steelcase Library, librarians are answering the reference questions requiring the skills of a librarian, whether or not librarians are physically at the reference desk. The researchers had hypothesized that without someone present at the desk, questions would go unasked and thus unanswered. According to the results of this study, this is not the case. Students with reference questions are coming to the service desk to ask them, and service desk staff members are referring questions that require librarian expertise to the librarian on call.

So while the higher-level reference questions and citation related questions are being answered by librarians, whether they are staffing a reference desk or are on call, the other questions are no longer coming to librarians. This indicates that this single-service on-call reference model is working. The researchers hypothesize that ready reference and directional questions are being answered at the service desk.

At the Steelcase Library reference questions fell in the spring/summer semesters over the course of the study period as shown in table 4. There is no clear explanation for this, although there are several factors that may have caused this decline. It is probable that the number of reference questions is linked to the number of students on campus during the spring/summer semester, which varies over the course of the spring/summer semester. Also, because of the small number of reference questions asked during these semesters, even a small curricular change in one program (e.g., no longer requiring a research paper in one class) could significantly affect the number of questions asked.

The data collected during this study was illustrative in establishing actual reference question patterns over the day and the semester. This evidence-based knowledge impacts staffing decisions and schedules, allowing the library to be sure librarians are on call during the periods with the heaviest reference loads.

Librarians at the Steelcase Library were initially ambivalent about adopting an on-call model of reference. While they knew that overall activity at the reference desk was declining, and they saw the practicality of spending time in reaching students in other higher-impact ways, they also strongly felt the need to be available to help students at their time of need. The anxiety they felt about the reference change appears to be common among many academic librarians who are understandably afraid that, by not being physically present at a reference desk, they are somehow failing their students. At the Steelcase Library, this study allowed the researchers to compare reference models in an objective light and assess the changes about which the librarians felt uncomfortable. Further qualitative study would be useful to determine patron satisfaction. However, through quantitative assessment, the researchers discovered that the change in the reference model did, in fact, seem to meet the reference needs of their learning community.