18
LISR 19,135-152(1997) Improving Reference Service Cost Studies Eileen G. Abels College of Library and Information Services, University of Maryland A thorough examination of the library and information science literature reveals problems with current reference service cost studies: they focus on individual reference services yet these services are difficult to differentiate; units of analysis used in these studies are variable; cost measures lack reliability and validity; and difficult allocation decisions are made arbitrar- ily. Reference service cost studies may be improved by developing detailed categorizations of units of analysis; increasing sample size; avoiding block of time measurement considering library services as a whole; determining alternative methods of allocating shared and indirect costs; and matching cost data with decision-making needs. Research is needed to determine if new approaches are feasible and cost effective. Lessons may be learned from previous reference service cost studies and from techniques employed in other service industries. Cost studies of reference services can provide managers with quantitative data for a variety of administrative decisions. These studies are meaningful only if the data gathered support the decision being made and if the studies are based on sound methodologies using reliable and valid measures. Without these, erroneous conclu- sions may be drawn that impair sound decision making. The cost studies reported in the literature seem to support the conclusion that the cost of reference services is extremely high and their future is in jeopardy (McClure, 1986; Mm-fin, 1993; Summerhill, 1994). This conclusion may not be warranted. A thorough examination of reference service cost studies reveals several problems that make the results of these studies suspect. This article summarizes current costing practices, critiques these practices, and suggests directions for future research to improve reference service cost studies. Direct all correspondence to: Eileen G. Abels, College of Library and Information Science, Hombake Library Building, Room 4105, College Park, Maryland 20742 <[email protected]>. Library & Information Science Research, Volume 19, Number 2, pages 135-152 Copyright 0 1997 Ablex Publishing Corporation All rights of reproduction in any form reserved. ISSN: 0740-8188 135

Improving Reference ServiceCost Studies

Embed Size (px)

DESCRIPTION

Improving Reference Service Cost Studies

Citation preview

Page 1: Improving Reference ServiceCost Studies

LISR 19,135-152(1997)

Improving Reference Service Cost Studies

Eileen G. Abels College of Library and Information Services, University of Maryland

A thorough examination of the library and information science literature reveals problems with current reference service cost studies: they focus on individual reference services yet these services are difficult to differentiate; units of analysis used in these studies are variable; cost measures lack reliability and validity; and difficult allocation decisions are made arbitrar- ily. Reference service cost studies may be improved by developing detailed categorizations of units of analysis; increasing sample size; avoiding block of time measurement considering library services as a whole; determining alternative methods of allocating shared and indirect costs; and matching cost data with decision-making needs. Research is needed to determine if new approaches are feasible and cost effective. Lessons may be learned from previous reference service cost studies and from techniques employed in other service industries.

Cost studies of reference services can provide managers with quantitative data for a variety of administrative decisions. These studies are meaningful only if the data gathered support the decision being made and if the studies are based on sound methodologies using reliable and valid measures. Without these, erroneous conclu- sions may be drawn that impair sound decision making. The cost studies reported in the literature seem to support the conclusion that the cost of reference services is extremely high and their future is in jeopardy (McClure, 1986; Mm-fin, 1993; Summerhill, 1994). This conclusion may not be warranted. A thorough examination of reference service cost studies reveals several problems that make the results of these studies suspect. This article summarizes current costing practices, critiques these practices, and suggests directions for future research to improve reference service cost studies.

Direct all correspondence to: Eileen G. Abels, College of Library and Information Science, Hombake Library Building, Room 4105, College Park, Maryland 20742 <[email protected]>.

Library & Information Science Research, Volume 19, Number 2, pages 135-152 Copyright 0 1997 Ablex Publishing Corporation All rights of reproduction in any form reserved. ISSN: 0740-8188

135

Page 2: Improving Reference ServiceCost Studies

136 Abels

CURRENT COSTING PRACTICES

Reference services are broadly defined as those services that provide access to information through direct or indirect inte~ediation. Most reference service cost studies determine the average unit cost of one or more specific services (Cable, 1980; Cochrane & Warmann, 1989; Kantor, 1986; Kantor, Saracevic & D’Espo- sito-Wachtmann, 1995; Nachlas & Pierce, 1979; Robinson & Robinson, 1994; Spencer, 1980). The average unit cost of a reference service is calculated by dividing the total cost of a reference service by the total volume of work. For service R, the unit cost is represented by:

U(R)= C(R) / V(R) (1)

where U(R) = average unit cost of service R, C(R) = total cost of service R, and V(R) = total volume of work of service R.

The total cost of an individual service is calculated by adding together all direct and indirect costs for that service. Direct costs are those labor and other costs that can be directly assigned to a service or output. Direct labor cost is the cost of time spent on a specific service. Other direct costs are costs of materials, supplies, and equipment used for a specific service. Indirect costs are those costs that are not easily associated with any one service, such as costs for space and utilities. Indirect costs may include fringe personnel costs, facilities costs, service costs, and managing costs allocated to a specific service.’

The total volume of work is dete~ined by identifying work the unit of work for a given service and measuring the total number of units provided during a given period of time, usually a calendar year. The following list identifies the unit(s) of analysis most commonly used in cost studies for three categories of reference services: Mediated reference request services; end-user services; and user education services.

1. Mediated reference request services A. Reference desk service -Reference question/transaction B. Mediated online search service anline search

2. End-user search services A. automated reference service -Search session B. in-house use of materials -Book use

3. User education services <lassroom instruction session -Session attendee.

The application of average unit costs to individual reference services assumes that reference services can be clearly differentiated, that each service has only one unit

‘Fringe personnel costs are often included as part of direct labor costs, but this practice various in the studies reported. Treatment of other indirect costs also varies.

Page 3: Improving Reference ServiceCost Studies

Reference Service Cost Studies 137

of work that is completely representative of it, and that these units of work are comparable.

CRITIQUE OF CURRENT COSTING PRACTICES

A critical review of reference service cost studies reveals a wide range of average costs (see Table 1). Such a large range is indicative of underlying problems. First, the lack of clear service de~nitions of reference services and the dif~cul~ of dl~fe~~?~~i~t~~~ these services complicates costing efforts. Second, the units of work, also referred to as units of analysis, associated with these services are variable. Third, the cost measures used in the studies lack both reliability and validity. Finally, average unit cost studies of individual reference services are plagued with difficult

TABLE 1 Unit Costs of Reference Services Derived from Selected Studies

(in chronologicaf order)

Study

Cable (1980)

Spencer (1980)

Kantor (1986) [data from 1982/3]’

Cochrane &War- mann (1989)

Robinson & Robin-

son (I 994)

Kantor, Saracevic, & D’Esposito-Wacht- mann (1995) [Data from 19941

Cost and Service as Reported

Average cost of search (excluding hidden

costs) Directional queries, 1 step Reference queries, subtotal

Two step, quick reference subtotal Ready reference Biographical Bibliography distribution Verification Citation retrieval Instructional Referral

Multiple step, extended reference Subject search

Consultation, training, tours Query

Cost Reported

Adjusted cost’

$5.18 $15.29

$0.44 $1.30 2.52 7.43 1.98 5.84 1.40 4.13 4.06 11.98 1.27 3.75 3.37 9.94 2.73 8.05 0.82 2.42 1.69 4.96

4.57 13.82 9.09 26.82

$14.00 (median) $33.74

Full cost directional

Futt cost reference Average Total Cost Per Reference

Question Handled Average Marginal Paid Labor Cost Per

Question Handled Average Marginal Labor-related Cost Per

Question Handled Cost per unit of reference service” low

high

$3.53

$9.22 $6.84

$1 .a3

$4.45

$l.t6 35.52

$6.49

16.95 $7.83

$2.10

$5.09

$1.32 (low) 40.67 (high)

Notes: ’ 1996 costs calculated on the basis of a 7% annual increase. 2Costs given only for types of services to assure confidentiality.

Page 4: Improving Reference ServiceCost Studies

138 Abels

cost allocation decisions that impact the costs derived. The discussion that follows considers each of these problems in turn.

To facilitate the discussion, the terms d~~~e~t~ated service, varia~ili~, ~e~~a~iZ- ity, and validity are defined as follows:

l Differentiated Service is a service that has a unique unit of analysis that is associated only with that service;

l ~ar~u~i~i~ is the lack of similarity between one individual unit and another; l ~e~~~~~lj~ is the ability of a measure to obtain the same value each time it is

used in like situations; and l Validity is the ability of a measure to obtain true cost values.

Inability to Differentiate Reference Services

Although service differentiation is theoretically possible (Whitson, 1995) in prac- tice it is very difficult to achieve. The overlap of library services seems unavoidable, particularly given that their inputs are often shared. In fact, the integration of services is often desirable since it results in cost savings; however, it is this integration that complicates determining the cost of any single service (Braunstein, 1979). The average unit cost approach calculates the costs of output units of individual services, such as the answering of a reference request or the conducting of an instruction session. It is impossible to cost an individual service accurately if that service cannot be differentiated from other services.

User education illustrates the problem of how lack of service differentiation prevents accurate costing. User education related activities include: (I) the reference desk service, where instmctional requests are often received; (2) individual user consultation sessions that, in effect, individualize user instruction sessions; and (3) the provision and updating of search guides and aids for the end-user. Moreover, many of the activities involved in formal user instruction have spin-off contributions to other services. Guides and aids prepared for distribution during a classroom session may later be dis~ibuted directly to end-users who do not attend inst~ction sessions. Since these aids are not used solely for formal classroom instruction, the cost of their preparation and maintenance should not be allocated entirely to this service.

Variable Units of Analysis

Determination of average unit costs requires the identification of comparable units of analysis. As mentioned above, reference service cost studies generally use some aspect of service output as their unit of analysis. Regardless of the specific service in question, the unit of analysis chosen for output measurement tends to be variable. The problem with using variable units of analysis in dete~ining the overall cost of a service is that summary statistics, such as averages, are not meaning~lly descrip- tive. The problem of reaching wrong conclusions on the basis of average activity costs is common to any service industry in which activity time varies by product line and customer segment (Kock, 1995). For libraries, it affects mediated reference request services, end-user search services, and user education services.

Page 5: Improving Reference ServiceCost Studies

Reference Service Cost Studies 139

Mediated Reference Request Services. Cost studies of mediated reference request services have focused on the cost per reference request or transaction, whether that request is a ready reference question or a mediated search request. Reference requests are not all the same; they are of varying complexity and are extremely variable in nature. On one end of a reference request continuum are simple questions that might be answered in very little time without the use of any specific library tool or resource. On the other end of this continuum are complex reference requests that require much time and many resources.

There is additional variability in determining the cost of answering a reference request that requires online searching because of the many different and constantly changing pricing structures for online services. A search conducted in a database provided through one vendor will not cost the same when conducted on the same database provided by another vendor (Biaget, 1982; Jack, 1990; Saffady, 1992). Furthermore, the online connect pricing structure still used by many vendors makes pricing dependent upon the skills and expertise of the individual searcher. Many factors influence online searching and, thus the search output (Fidel & Soergel, 1985). In essence, searching is an art; the same request will be transformed into different strategies and different outcomes depending upon the searcher (Saracevic & Kantor, 1988).

End-User Search Services. There is a great deal of variability in end-user searching of both print and electronic resources. An automated reference search session might range from one query posed on one database to several queries posed on multiple databases. Equally variable are the number of items consulted per visit to the library. Moreover, while one user may browse several items, another may read several items completely. In addition to variable units of analysis, the costing of end-user search services raises other issues relating to variability. End-user searching of automated reference sources, as with mediated searching, varies with the many factors that influence searching in general. Also, end-users consulting items in-house varies greatly across disciplines and formats. For example, the assumption of average material costs falsely equates usage costs for Chemical Abstracts, an expensive abstracting and indexing service, and for the Wall Street Journal, a relatively inexpensive daily newspaper.

User Education Services. Despite the increased emphasis on user education in libraries, there is only one cost study in the literature that includes user instruction among the services considered (Cochrane & Warmann, 1989). It uses two units of analysis: classroom instruction sessions and individuals attending classroom in- struction sessions. Classroom instruction session as the unit of analysis is, again, variable. Sessions vary in scope, length, and amount of resources used. Somewhat less variable are individuals attending sessions, yet it is a stretch to assume all attendees at a given instruction session are comparable.

Reliability of Measures

In addition to the variability problem, reference service cost measures lack reliabil- ity. It is difficult to provide accurate counts of the units of work that are used for

Page 6: Improving Reference ServiceCost Studies

143 Abeis

most reference services. While the total labor cost of salaried reference staff can be reliably measured based on yearly salaries, measures of labor costs for individual reference service output tend to be extremely unreliable, as can be seen, again by looking at mediated reference request services, end-user search services, and user education services.

Mediated Reference Request Services. For this discussion, the unique charac- teristics of reference desk requests and mediated online search requests require separate treatment. It is difficult to count accurately the total number of reference requests received at a reference desk since such statistics rely on busy reference librarians who often find it necessary to base their counts on their recollection of the number and type of requests received at the end of the day. Moreover, assigning reference requests to a designated category sometimes is arbitrary since one refer- ence request may actually represent several sub-requests or span several categories of requests. For costing, the totai number of reference requests may be calculated on the basis of yearly reference request statistics, hourly statistics, or statistics for a selected period of time, usually several weeks. The latter approach, particularly if only a few weeks are involved may yield a non representative sample, since reference desk traffic is generally cyclica depending on time of day and time of year.

The most con~on approach to measuring labor cost per reference request is to determine the amount of staff time spent on an average request. This is done by dividing a “block of time” by the number of reference requests received during that block of time (Murfin, 1993). Two problems are associated with this practice. First, during a block of time, such as an hour or a reference desk coverage period, reference fibrarians are not usually dedicated entirely to answering questions; that is, there is often idle time between requests. This was docu~~ented by one study that reported that the amount of time spent with patrons while on the reference desk for a period of one hour ranged from 4.3 minutes to 32 minutes, with an average of 19 minutes per hour (Balay & Andrew, 1.978). Thus, dividing any block of time by the number of requests received during that period of time does not give a reliable measurement since it attributes the cost of idle time to the cost of a request. Moreover, reference librarians may use the idle time on the reference desk to accomplish other reference related activities, such as selecting items for the reference collection, learning to use a new reference tool, or preparing a search guide. When this happens, the cost of the “idle” time should be associated with the other reference activity and not the reference desk service. The second problem with this method is the treatment of alf requests as comparable units As discussed above, reference requests are variable and cannot be treated as uniform.

While the actual search time spent performing a mediated online search can be measured reliably, this time represents only part of an extremely complex and variable process. A task analysis of the online search process reported in the literature delineates the following stages, each involving many detailed activities: ana~yst/reques~er interactions pre-onfine search pfanning; terminal session; and post-search activities (Wanger, McDonald & Berger, 1980). The results of this task analysis suggest that there will always be a question of generalizability due to the variability of each step of the process: a typical reference interview ranged in time from two minutes to 60 minutes, with an average of 8.5 minutes; 72% of the

Page 7: Improving Reference ServiceCost Studies

Reference Service Cost Studies 141

searchers conducted a reference interview over 75% of the time; the time spent annotating and evaluating searches ranged from one minute to 1.5 hours, with a mean time of 14.9 minutes.

End-User Search Services. Cost per search session or book use generally is not a reliable measure of an end-user search service. One reason for this is that the number of search sessions conducted is often determined by consulting sign-up sheets for workstations. However, the number of actual sessions conducted is not accurately represented by the sign-up sheets, since there is no mechanism for accounting for “no shows” and the kind of spontaneous usage that occurs when an available workstation is used without signing-up. One solution is to determine cost based on the number of possible sessions that could be scheduled during a one week period (Eaton & Crane, 1988). Another solution is to substitute the number of sessions on sign-up sheets with the number of minutes per database logon. The use of automated search logs permits accurate accounting of both in-house and remote searching. Logons represent actual usage, thus eliminating the “no show” problem associated with sign-up sheets. Moreover, calculating the number of minutes per logon helps to differentiate between short and long sessions. However, even having available the number of minutes per logon, it is still difficult to obtain reliable cost measures for automated reference services, insofar as most resources available to end-users are subscription-based. While the overall cost of a subscription is predict- able and reliable, end-users as well as librarians use these resources. Determining how to allocate subscription costs to end-user services, as distinguished from instruction and reference services, would require expensive time and motion studies.

Measures that attempt to capture the number of consultations of in-house mate- rials are also unreliable. Some libraries base usage statistics on the number of items left for reshelving. This assumes that the patron actually uses the item removed from the shelf. Also it does not account for users who consult books and journals while standing at the shelves reshelving them as they go along. In other libraries, patrons are asked to indicate usage on slips of papers attached to books and journals; this method is unreliable to the extent it relies on the cooperation of users, who may be uncooperative. Still another method is attaching a thread around a book or journal and determining usage by counting broken threads. In all methods of measuring end-user search service usage, reliability is affected by unreliable user behavior.

User Education Services. Number of classroom sessions and number of indi- viduals attending sessions can be counted accurately and both are reliable measures of formal user instruction, one component of a user education program. The labor time spent teaching a course can also be measured reliably. However, the time spent teaching a session does not include labor costs involved in preparation and this is what threatens reliability. It would be extremely difficult to obtain a reliable measure of the time spent preparing for classroom instruction sessions. For instance, a course being taught the first time requires a great deal more preparation than the same course taught a second or third time. Moreover, in some institutions, committees may be involved in conceptualizing a course and developing presentation materials. Such efforts represent large labor costs that must somehow be included in the costing of instruction sessions.

Page 8: Improving Reference ServiceCost Studies

142 Abe&

Validity of Measures

Perhaps the most serious problem with methods of determining average unit costs of reference services is the lack of validity. Each of the three types of reference services illustrate the problem.

Mediated Reference Request Services. The vaIidity of measures that cost mediated reference request services generally is questionable. To respond to a mediated reference request of any type requires expertise on the part of the reference librarian; reference librarians must develop their search skills and monitor new sources and system updates. Essential activities such as these are never represented by measures used in cost studies of mediated reference request services.

It has already been mentioned that mediated reference requests received while at the reference desk do not occupy all of the time spent on the reference desk service. Particularly problematic is the fairly common practice of con&ding that reference services in general are costly on the basis of the unit cost of requests received at the reference desk. This conclusion inaccurately equates reference desk service with reference sewice in general. The fallacy of using requests received as a measure of reference service has been noted in the literature (Rothstein, 1984). Time spent at the reference desk is actually only a percentage of a reference librarian’s time. Reference librarians meet their overall objective of providing access to information to patrons through a wide variety of activities and services. Costing studies that focus on only reference desk services and report average costs of reference requests mediated at the reference desk give the picture that one output from this one service is represen~tive of reference services generally.

End-User Search Services. In the case of end-user use of automated reference services, a validity problem inherent in costing relates to using number of search sessions as the unit of analysis. Number of search sessions, based on sign-up sheets, does not provide a valid measurement of the cost of automated reference services. Session data based on sign-up sheets do not distinguish between a small number of frequent users and a high number of infrequent users.

But even more generally, cost studies that focus on users use of either in-house electronic reference sources or print sources do not provide a valid measurement of total library resource use. The distinction normally made between the two types of services in cost studies reflects an outdated model of the use of library resources for three reasons:

l The use of various formats is now integrated; l Users are no longer restricted to use of in-house resources; and l Remote users also access library resources.

When first introduced, electronic resources were treated as separate from print resources, Now that electronic resources are prevalent in libraries and information centers, cost studies need to reflect their usage accurately. It seems likely that, as time goes on, more and more patrons will consult a variety of formats when they visit a library. The problem with measuring in-house book or material usage is that

Page 9: Improving Reference ServiceCost Studies

Reference Service Cost Studies 143

it reflects only a subset of library resource use. It is equally true that in-house users represent only a subset of all library users; in fact, remote access library users are becoming increasingly common. The chances of obtaining valid statistics might be improved by costing information access in general rather than trying to differentiate services by format, such as print or electronic.

User Education Services. Measuring the cost of user instruction on the basis of a cost per class session or cost per student attending a session also lacks validity, because classroom instruction is not representative of all of the activities that are involved in user education. Almost all reference activities involve some level of user education. User education generally consists of a program, not a single activity, involving, for instance, individual ready-reference instruction, individual user in- struction sessions, formal group instruction, and the development and maintenance of print and electronic aids. Focusing on formal classroom instruction as solely representative of a user education program generally fails to portray the true extent of the service. Measuring the cost of user education using instruction sessions as units of analysis is valid only if the definition of user education is restricted narrowly to include formal classroom instruction and nothing else. The same holds true for measuring the cost of user education in terms of students attending classroom sessions. The total number of people who attend classroom instruction sessions is relatively small in relation to the entire user population. Many users who do not attend such sessions also benefit from a user education program, insofar as they utilize search aids and ask for individualized instruction at the reference desk.

Difficult Allocation Decisions

The costing of reference services also poses difficult decisions regarding the allocation of shared resources. Details concerning how allocation decisions are made are often lacking in reports of reference desk cost studies; those studies that do mention allocation of resources expose the arbitrary fashion in which these decisions are made. Difficult allocation decisions arise when dealing with shared resource costs, shared labor costs, and indirect costs.

The collection is particularly problematic with respect to allocation since all reference services use at least some portion ofthe collection. One important question that has not been resolved regarding the collection is how “collection” should be defined in reference service cost studies. That is, should the entire collection be costed or only the reference collection? This question is further exacerbated by the transitional state of libraries in terms of its resources, its patrons, and its competitors (Abels, Kantor & Saracevic, 1996). Furthermore, costing efforts must take into account lost materials and aging of materials. While some studies exclude the costs of library resources in reference service cost studies, others attempt to allocate total collection costs across services.

In an effort to normalize material usage across reference services, Kantor ( 1986) introduced a book use equivalency (BUE) measure. Through interviews with librarians and limited reshelving studies, he determined, for large academic librar- ies, a BUE of one for circulation, two for reference queries, and three for in-house use. A similar measure, referred to as a circulation equivalency (CE), was developed

Page 10: Improving Reference ServiceCost Studies

144 Abeis

by Robinson and Robinson (1994). Again, based on assumptions and available statistics from public libraries, they determined a CE of two for reference questions and one-half for in-house use. Book use equivalency is a narrow represen~tion of resource usage, and, as such is reason for concern. An additional concern in using book use equivalency is its basis on the assumption that usage of resources can be represented by an average. However, because the BUE is an approach to allocating the costs of the collection across services, it is worth looking to see if these concerns can be addressed.

Tools for accessing resources, such as the online public access catalog (OPAC) and the Internet also are difficult to allocate across services. To begin with, the costing of these tools is in itself very complex (Matthews, 1985; McClure, Bertot, & Beachboard, 1995). Then, these tools are changing at a very rapid pace. Yet, both the OPAC and the Internet are essential to all reference services to a varying degree and their inclusion or excIusion from cost studies impact the outcome. Reference service cost studies conducted since the inception of the Internet have not clearly stated if the costs of these tools have been included or excluded. Furthermore, as for the costs of the collections, decisions are needed regarding the level of costing: Should the cost of the Internet include part of the cost of installation of the Internet at the institutional level, or only IocalIy incurred costs?

Shared labor costs, like shared resource costs, pose dif~cu1ties in costing studies. As in other service industries, libraries have employees who work on multiple activities or services making it difficult to isolate the time spent ona specific service. Some methods used in reference cost studies to estimate time spent on various activities are based on employee estimation and recollection. Such estimations have been found to be inaccurate (Wanger et al., 1980) and may lead to erroneous conclusions (Kock, 1995).

In addition to shared direct costs, indirect costs also pose substantial problems to reference service cost studies. As is the case for other service industries, overhead costs for reference are extensive (Kock, 1995). While some reference service studies exclude all indirect costs and include only labor costs, other studies attempt to allocate all indirect costs across services. The unequal treatment of overhead costs is probably responsible for a great deal of the wide range of reference service costs reported in the literature.

Summary of Current Costing Problems

The review of current costing practices used for mediated reference request services, end-user search services, and user education services shows that they are problem- atic because of the inability to differentiate the services, the variability of the units of analysis used, and the lack of reliable and valid measures. They are problematic also because of difficult decisions regarding the treatment of shared costs, both direct costs, such as those for the collection, online catalogs, and the Internet, and indirect costs. Thus far, such decisions have been made on a fairly ad hoc basis. Attempting to obtain unit costs of individual services requires a great deal of conjecture and best guessing, which undermine the quality of the results. This is unfortunate in that average unit cost studies of individual reference services have the potential of providing managers with a great deal of specific cost data that can

Page 11: Improving Reference ServiceCost Studies

Reference Service Cost Studies 145

be used for service planning and cost benefit analysis. It would be useful, therefore, to attempt to identify ways of improving cost studies.

SUGGESTIONS FOR FUTURE RESEARCH IN COSTING

Some of the problems described in the previous section are endemic to the reference services being measured. While solutions to some problems might exist, they may not be cost effective. For example, while exacting time and motion studies might resolve the problem of reliability, they are extremely costly to carry out. Still it is possible that some problems could be remedied in cost-effective ways and that additional research might provide insight to gathering meaningful cost data.

Beyond Single Categories of Units of Analysis

As was noted above, in the average unit cost model, the variability of units of analysis frustrates meaningful costing efforts. The development of detailed catego- rizations of analysis can improve current costing efforts. While categorization does not remedy all of the problems associated with average unit costing, categorization helps reduce the problem of variability to a certain extent because rather than focusing on a single broad service unit, the focus is on several groupings of similar sub-units. Each category has its own unit of analysis and unit cost is calculated per category of service output. Thus, for example, rather than calculating the unit cost of reference desk requests in general, categorization would permit the calculation of the cost of specific types of reference desk requests. In one notable study, Spencer (I 980) provided time and cost for various levels of categories of reference questions: queries were first divided into directional and reference queries; reference queries were further categorized as two-step quick reference or multiple-step extended reference; two step quick reference queries were further categorized as ready reference, biographical, bibliography distribution, verification, citation retrieval, instructional, and referral.

Spencer’s detailed categorization of reference requests is but one of many existing categorizations of reference requests. There are many ways to categorize reference desk questions (Cable, 1980; Cochrane & Warmann, 1989; Kantor, 1986; Robinson & Robinson, 1994; Spencer, 1980) and mediated online search requests (Elchesen, 1978; Havener, 1990; Saffady, 1992). Unfortunately, while none of the categorizations is suitable for costing studies, lacking either mutually exclusive categories or clear-cut criteria for the assigning of requests to one category rather than another, still they are suggestive of a framework for the development of a usable categorization.

Cost studies oriented toward categories of reference services will require exten- sive and detailed data gathering efforts. The cost of gathering such detailed statistics may be high and burdonsome to staff, if gathered on a regular basis. While categorical data may be important for certain decisions, it may not be necessary for others. For example, if a manager is trying to determine if a particular category of reference question, such as multiple-step extended reference, should be eliminated from reference desk service, then it would be important to know the cost of this type of reference question. On the other hand, distinguishing between one-step bio-

Page 12: Improving Reference ServiceCost Studies

146 Abels

graphical questions and one-step verification questions would not provide useful data for this particular decision. When gathering specific category data, proper sampling techniques and methodologies are needed.

Beyond Small Samples

Rather than drawing small samples during limited time periods, the use of larger samples drawn over longer periods of time will improve the reliabili~ of reference cost studies. The design of reference cost studies must take into account the cyclical nature of many reference services covering various times of the day, week, month, and year. Increasing the sample size and selecting times to be representative will permit generalizations to a service generally and not just to that service during the specific time period under study. The increased sample size will not greatly increase the cost of the study; it will however, extend the time to complete the study. The trade off is increased reliability.

Beyond Block of Time Measurement

The notion of abandoning “block of time” studies and replacing them with the “cost per transaction” has already been suggested (Murfin, 1993). Rather than dividing a block of time by the number of requests received, the “cost per transaction” method measures the exact time spent on each individual request. Although this approach to measurement was developed for reference transactions, there is no reason why other studies could not adopt a similar methodology. The improved methodology requires time and motion measurements.

While time and motion studies can be expensive to execute, Spencer (1980) demonstrated that random time and motion studies used in conjunction with self- observation can yield reliable results in a cost-effective manner. She further notes that such studies interfere minimally with day-to-day operations and can be executed easily by library managers. Other research has found that random alarm mechanisms are useful for conducting large scale random time and motion studies (Beecher, Self, Stinson & Anderson, 1982). Again, the increased time required for measurement has increased reliability as a payoff.

Beyond Individual Services

Library services cannot be viewed in isolation from other library services. As was noted above, it is difficult, if not impossible, to differentiate library services. For this reason, cost studies are more meaningful when cost data are gathered for all services offered, rather than for only a particular service at a given moment. Furthermore, the allocation of indirect and shared costs requires consideration of all services offered. Library cost studies could learn from the experiences of those who have applied activity based costing and manag~rne)~t (ABCM) to service organizations. ABCM provides a holistic approach by focusing on organization- wide activities. It strives to improve the utilization of strategic resources, and considers the expectations of stakeholders (Lambert & Whitworth, 1996). To apply activity based costing to a library, one would first determine all activities carried

Page 13: Improving Reference ServiceCost Studies

Reference Service Cost Studies 147

out within each department of the library. Then, for each activity, the following questions are addressed: why is each activity performed? how often is it performed? for whom it is performed? what resources are consumed when it is performed? and what factors influence the activity? There appears to be only one article in the library and information science literature that explored the use of activity based costing in libraries (Danilenko, 1985).

Beyond Arbitrary Allocation of Shared and Indirect Costs

As indicated earlier, the method of allocating shared costs to multiple services can have a significant impact on the results of cost studies. Shared costs include general library resources, labor costs, and indirect costs. The difficulty of allocating shared and indirect costs across library services in a nonarbitrary way is, in part, due to the lack of inventories of products on which to base the allocation (Lambert & Whit- worth, 1996).

Particularly difficult is the allocation of general library resources. These re- sources, including the collection and database access provided by the library, fall into the shared costs category since these resources are fundamental to a variety of services and can be accessed directly by the user without direct or indirect reference assistance. The costs of general library resources include the actual costs of the resources and the costs of acquiring, processing, and providing access to them through the catalog. An alternative to allocating the cost ofgeneral library resources is to expand the Book Use Equivalency (BUE) concept introduced by Kantor ( 1986) and discussed above: Rather than basing the measure on books alone, other re- sources, such as videos and CD-ROMs, can be included. The extension of a BUE to a resource use equivalency (RUE) would require a great deal of study to ascertain equivalency values. However, research into the use of the RUE could provide needed answers to questions about the feasibility of equivalencies and appropriate numbers to use in cost studies. Thus, cost studies could be improved by substituting educated guesses and estimates with numbers derived from research.

Shared labor costs for reference services are another example of costs that must be allocated. Careful analysis is needed of the shared labor time, to ascertain if, in fact, the time spent is associated with more than one service, or if the time is spent primarily for one service but indirectly contributes to another. This type of analysis has the potential to be subjective and arbitrary. To avoid this it is necessary to apply a percentage of the costs of the task to the various services based on usage data. In any case, further research is needed to determine the impact on cost data of various methods of allocating shared costs.

In addition to shared costs, research is needed to explore alternative methods of allocating indirect costs. As was noted above, indirect costs in service industries are high and arbitrary treatment of these costs can negatively impact the results of cost studies. Two types of indirect costs need to be recognized in a total cost model of library services: library indirect costs and insti~tional indirect costs. Library indirect costs can be defined as all library costs that are not categorized as direct costs, including both labor and resources costs. Library indirect labor costs may include staff time that is not identified with one discrete reference service, as well as managerial labor costs, that is, planning and supervision.

Page 14: Improving Reference ServiceCost Studies

148 Abels

The second category of indirect costs is institutional indirect costs. Included here are costs for library facilities, such as office space, heat, electricity, telephone, ~i~re, equipment, and fixtures, that are generally not included in a library’s operating budget. Also included as institutional costs are service costs, such as accounting, purchasing, and personnel services as well as managing costs, such as planning, management, and marketing. Institutional accounting systems vary in the way in which institutional indirect costs are calculated and usually are beyond the control of the library. In general in cost studies, it is desirable to allocate to direct costs as many costs is possible.

One method for allocating indirect or overhead costs is activity based costing. First, indirect costs are assigned to homogeneous cost pools, that is to activities that have the same consumption rate across products or services. Other indirect costs are assigned to cost pools, for each of which a rate of consumption is calculated (Kock, 199.5). Applying this approach to allocating indirect costs to reference services will improve reliability. However, research is needed to accurately determine the rates of consumption of indirect costs of various services.

Beyond Average Unit Costs

It is impo~ant to gather appropriate cost data for decision making. The cost studies listed in Table 1 all utilized average unit costing, yet the objectives of these studies were markedly different (see Table 2). For example, testing a methodology is different from performing a cost-benefit analysis. Some of the differences between objectives are more subtle; determining how much to charge for a service is different from identifying candidates for fee-based services. Different objectives require different data to be collected. Before cost studies are undertaken, library managers need to consider the decision to be made and determine what data will, in fact, support their decision making.

TABLE 2 Management Decisions or Purposes of Selected Cost Studies

Study Management Decision or Purpose of Study

Cable, 1980 How much does a service cost in order to determine how much to charge-- for that service?

Cochrane and War- How much do basic services cost and which are appropriate to offer in a

mann, 1989 fee-based service? Kantor, 1986 To determine unit costs at several libraries simultaneously. [Could be used

to determine fees for cost recovery.] Kantor, Saracevic, and To determine the cost and value of 18 library services at 6 institutions.

D’Esposito-Wacht- mann, 1995

Robinson and Robinson, Row are resources spent? 1994 What shares of the total budget are devoted to each service?

What is the composition of the resources allocated to each service? What are the costs for each unit of service delivered? What resources could we reallocate to support an existing service?

Spencer, 1980 Is random time sampling with self-observation a cost-effective method of

obtaining reliable cost data? -.

Page 15: Improving Reference ServiceCost Studies

Reference Service Cost Studies 149

While average unit costs may provide appropriate data for some decisions, other decisions may require different or additional data. In describing the use of econom- ics in libraries, Kingma (1996) discusses the impor~nce of marginal costs to economists in making cost-related decisions. Marginal cost is the additional cost associated with producing one additional unit of output. Only one of the reference cost studies reviewed (Robinson & Robinson, 1994) reported marginal costs. This study is illustrative of how library and information center managers may be able to learn from the experiences of others. But again, a first priority for future research is to provide insight into matching cost data and costing approaches to managerial decisions.

Beyond Costs of Services

All services must be viewed in the context of the contribution they make to the users of those services. The notion of the impact of a service is expressed in many ways in the literature: impact, value, and benefit. Value analysis of activities, as a part of activity based costing in service industries, is described in some detail by Lambert and Whitworth (1996). Of the reference service cost studies reviewed, only one studied costs and impacts of services (Kantor et al., 1995). While an impact analysis is clearly more expensive to execute than a cost study alone, it can be argued that cost studies of services are meaningless without consideration of the value of those services. The increased expense of a cost-benefit study would be particularly warranted in the case where a service needs to be justified.

Reliable and valid measures for the value of reference services are even more elusive than reliable and valid costing measures. Several authors have reviewed works on the value of info~ation (Turin, 1993; Repo, 1986) and several studies have addressed the value of services (Broadbent & Lofgren, 1993; Griffiths & King, 1993; Kantor et al., 1995). Yet, this area clearly merits additional research,

Summary of Recommendations for Future Research

Possible improvements to current costing practices of reference services require research and extensive data gathering efforts. Current reference service cost prac- tices can be improved by moving beyond: (1) single categories of generic units of analysis to more homogeneous subcategories; (2) small samples to larger samples that take into account various usage cycles; (3) costing individual services to holistic approaches that take into account organizational contexts; (4) arbitrary allocation of shared and indirect costs to the implementation of effective alternatives; (5) average unit costs to matching appropriate types of data to the decision being made; and (6) studies of the cost of reference services to studies that also consider the impact or value of these services.

CONCLUSION

Reference service cost studies can provide managers with data for decision making. Most cost studies of reference services calculate the average unit costs of reference service output. While this approach has the potential of providing managers with a

Page 16: Improving Reference ServiceCost Studies

150 Abeis

great deal of specific cost data, an analysis of the literature reveals that the approach is problematic as currently implemented, Current costing practices are based on a tm~tio~al model of reference service that requires service differentiation. They are based on variable units of analysis and the cost measures lack reliability and validity. Average unit cost studies are also problematic because of difficult decisions regard- ing the treatment of shared resources, such as the collection, and shared tools, such as online catalogs and the Internet; decisions which have thus far been made on an ad hoc basis.

Research in several areas can provide insights that will lead to improvements in current costing practices. Some cost related research questions are:

What reference service units are most appropriate for cost studies? What systematic ways can be developed for allocating shared and indirect costs across reference services? Row can the book use equivalency measure be expanded to a resource use equivalency measure? What cost data are most appropriate for a given cost study objective? What approaches, such as activity based costing and management, are transferable from other service industries to the costing ofreference services?

Reference service cost studies will continue to be important for library and information managers. The challenge of conducting meaningful cost studies based on reliable and valid measures has become greater in this time of transition.

This article has identified several ways of improving current reference service costing practices, First, some improvements can be achieved by im~Iementing findings of prior reference service cost studies. Second, approaches used in orher service industries may prove useful for costing reference services and third, research is needed to address still unanswered questions.

Ack~ow~edgmeRt: The author would like to acknowledge Elaine Svenonius, who patiently read many drafts of this article and provided feedback throu~out the process. Special thanks to Marilyn Domas White and the LISR reviewers for their valuable suggestions. The author would also like to acknowledge the assistance of Allen Brewer in the development of this manuscript.

REFERENCES

Abe&, Eileen G., Kantor, Paul B., & Saracevic, Tefko. (1996). Studying the cost and value of library services: Applying functional cost analysis to the library in transition. Journal of the American Society for Information Science, 47,2 17- 227.

Balay, Robert & Andrew, Christine. (197S). Use of the reference service in a Iarge academic library. CaEEege & Research Libraries, 36,946.

Beecher, John W., Self, Phyllis C., Stinson, E. Ray, & Anderson, Nancy D. (1982). Use of random alarm mechanisms for analyzing professional and support staff activities in science libraries, Part 1: Methodology. Library Research, $137-146.

Page 17: Improving Reference ServiceCost Studies

Reference Service Cost Studies 151

Biaget, T. (1982). Prices and online searching: The cheapest host is not always the one you think. In 6th International Online Information Meeting (pp. 27-37). London: Learned Info~ation.

Broadbent, Marianne, & Lofgren, Hans. (1993). Information delivery: Identifying priorities, performance, and value. Information Processing & Management, 29, 683-70 1.

Braunstein, Yale M. (1979). Costs and benefits of library information: The user point of view. Library Trends, 28,79-87.

Cable, Leslie G. (1980). Cost analysis of reference service to outside users. Ba~letin of the Medical Library Association, 68,247-248.

Cochrane, Lynn S., & Warmann, Carolyn. (1989). In J.C. Fennel1 (Ed.), Building on the first century: Proceedings of theJifth national conference of the associa- tion of college and research libraries (pp. 5562). Chicago, IL: Association of College and Research Libraries.

Danilenko, Gene. (1985). Activity based costing for services: The corporate infor- mation center. Special Libraries, 85,24-29.

Eaton, Nancy L., & Crane, Nancy B. (1988). Integrating electronic information systems into the reference services budget. Reference Librarian, 19, 161-177.

Elchesen, Dennis R. (1978). Cost-effectiveness comparison of manual and online retrospective bibliogmphic searching. Joumal of the American Society for Information Science, 29, 56-66.

Fidel, Raya, & Soergel, Dagobert. (1985). Factors affecting online bibliographic retrieval: A conceptual framework for research. Journal of the American Society for Information Science, 34, 163-l 80.

Griffiths, Jose-Marie, & King, Donald W. (1993). Special libraries: Increasing the information edge. Washington, DC: Special Libraries Association.

Havener, W. Michael. (1990). Answering ready reference questions: Print versus online. Online, 14,22-28.

Jack, Robert F. (1990). The new ESA/IRS pricing scheme: A comparison with DIALOG. Online, 14,35--39.

Kantor, Paul B. (1986). Three studies of the economics of academic libraries. In G.B. McCabe & B. Kreissman (Eds.), Advances in library administration and organization (pp. 22 1-286). Greenwich, CT: JAI Press.

Kantor, Paul B., Saracevic, Tefko, & D’Esposito-Wachtmann, Joanne. (1995). Studying the cost and value of library services: Final report. (Tech. Rep. No. APLAB/94/3/1,2,3,4). New B~swick, NJ: Rutgers State University of New Jersey, School of Co~~ication, Information and Library Studies, Alexandria Project Laboratory.

Kingma, Bruce R. (1996). The economics of information: A guide to economic and cost-benefit analysis for information professionals. Englewood, CO: Libraries Unlimited.

Kock, Soren. (1995). Implementation considemtions for activi~-based cost systems in service firms: The unavoidable challenge. Management Decision, 33(6), 57-63.

Lambert, Don, & Whitworth, John. (1996). How ABC can help service organiza- tions. Ch& Magazine, 70(4), 24-28.

Page 18: Improving Reference ServiceCost Studies

152 Abels

Matthews, Joseph R. (1995). Public access to online catalogs. New York: Neal- &human.

McClure, Charles R. (1986). A view from the trenches: Costing and perfo~ance measures for academic library public services. College & Research Libraries, 47,323-336.

McClure, Charles R., Bertot, John Carlo, & Beachboard, John C. (1995). lnternet costs and cost models forpublic libraries: Final report. Washington, DC: U.S. National Commission on Libraries and Information Science.

Mur~n, Marjorie E. (1993). Cost analysis of library reference services. In G.B. McCabe & B. Kreissman (Eds.), Advances in library administration and organization, Vol. 11 (pp. l-36). Greenwich, CT: JAI Press.

Nachlas, Joel A., & Pierce, Anton R. (1979). Determination of unit costs for library services. College & Research Libraries, 40,240-247.

Repo, Aatto J. (1989). The value of information: Approaches in economics, account- ing, and m~agement science. Journal of the American Society for Z~formation Science, 40,&J--85.

Robinson, Barbara M., & Robinson, Sherman. (1994). Strategic planning and program budgeting for libraries. Library Trends, 42,420-447.

Rothstein, Samuel. (1984). The hidden agenda in the measurement and evaluation of reference service, or, how to make a case for yourself. Reference Librarian, II, 45-52.

Saffady, William. (1992). The availability and cost of online search services. Library Technology Reports, 28, 177-263.

Saracevic, Tefko, & Kantor, Paul B. (1988). A study of information seeking and retrieving III. Searchers, searches, and overlap. Journal of the American Society for reformation Science, 39: 197-2 16.

Spencer, Carol, (1980). Random time sampling with self-observation for library cost studies: Unit costs of reference questions. Bulletin of the Medical Library Association, 68(l), 53-57.

Summerhill, Karen S. (1994). The high cost of reference: The need to reassess services and service delivery. Reference Librarian, 43,7 l-85.

Wanger, Judith, McDonald, Dennis, & Berger, Mary C. (1980). ~vaZ~ation o~the online process: Final report. (Available from NTIS PB8 1-132565). Santa Monica, CA: Cuadra Associates.

Whitson, William L. (1995). Differentiated service: A new reference model. Journal of Academic Librarianship, 21(2), 103-l 10.