A. Topics‎ > ‎TTT bibliography‎ > ‎

Literature survey

 Tracking the traffic in modern libraries

The current version (June 21, 2014) has been published in the Journal of Library Administration

Libraries that change need to monitor the patterns of library use. This essay presents a practical tool for the collection of systematic data on what users actually do inside the physical library.

Track The Traffic (TTT) is an instrument for studying user behavior in libraries and in other public spaces. It is based on tours of observation through the public areas of the library. The observers follow a fixed route and note down what the users are doing as they pass them. They only record activities that are easy to observe: is the user sitting or standing? is she alone or in a group? is she engaged with digital or printed media (or no media at all), etc. The tours are repeated at regular intervals through the day and through the week. This means that the universe of behavior is sampled rather than continuously observed.

Systematic observation

The use of systematic observation to gather quantitative data on user behavior in libraries is relatively new. It was first developed in the late nineties, when Leckie and Hopkins (2002) used what they called “seating sweeps” to collect data from the main public libraries in Toronto and Vancouver. The “seating sweep” approach inspired a fair number of follow-up studies, in academic as well as in public libraries. In 2004 a Norwegian library teacher and researcher (the author) started to use the same basic approach to study user behavior in Norwegian libraries. At the time he was not familiar with the Canadian work. After several pilot studies (2004-2007), the TTT method was standardized and used as a teaching tool in practical statistics for about one hundred library students during the period 2007-2010. Several Norwegian libraries have started to use TTT for management purposes. The TTT instrument has also been applied by libraries outside Norway, with projects in Sweden, Finland, Denmark, Germany, Lithuania, the United States and Uganda. Not all of these have been published, however.

Systematic observation is both a research and a management method. The “seating sweeps” approach has a research orientation. It was developed within a research context and involves a fairly large amount of data collection, processing and analysis. TTT was specifically designed as a management tool. A typical TTT study is shorter and cheaper than a typical “seating sweeps” project. But the two are not substantially different. The sweeps may also be used by managers, while TTT can be, and has been, used by researchers. TTT is a simplified and standardized sweep. Both may be called “traffic studies”.

During the last decade, there has been a slow, but steady growth, in the use of systematic observation in libraries. I expect the growth to continue, geographically and numerically. Both approaches allow library management to study usage patterns and trends that are hard to document by other types of data. In most cases they confirm “what we already know”. But that is true of most library research. Humans tend to recognize new patterns at an early stage. Proof is something else. Systematic data identify the intuitions that happen to be correct and prove them to be correct.  Below we map the history of traffic studies during the last fifteen years. Readers that would like quick introductions to the two approaches may consult Given & Leckie (2003) and Høivik (2012), respectively.

Seating sweeps

The first library researchers to use "walk-through" as an observation tool were Canadian. Leckie and Hopkins (2002) explored "the role of large central libraries as a type of public space within a changing urban context".  In 1999 they collected data on the main public libraries in Toronto and Vancouver. Four different methods were used: a written patron survey; face-to-face interviews with a smaller sample of patrons; in-depth interviews with staff; and finally "non-obtrusive patron observational surveys called 'seating sweeps'”. The sweeps consisted of three daily walk-throughs. The observers made "systematic and detailed observations of sixty different variables"  regarding "who (was) present in specific locations and what activities they (were) conducting at specific times of day" (p. 333-335). The full data set consisted of twenty-eight rounds of observation.

Since they worked with a high number of variables, the sweeps generated vast amounts of data. To reduce the workload, only Mondays, Wednesdays and Saturdays were used in the analysis, however. Their three day sample represented about seven thousand observations. In a second article, Given and Leckie (2003) emphasized the way spatial analysis techniques could be used to map "the physical layout of libraries and information centers".

"Walking and recording" is a well-known method in sciences like archeology and biology. An archaeologist may walk through a classical site in Greece, with a team of students, and let them note the number, and possibly the type, of pot sherds and other remains that they can observe on the surface (Snodgrass, 1987). A biologist may follow a series of parallel paths through a forest and note the number of plant species, or count the birds of different species he sees (or hears). Their observations are quantitative. Most anthropologists and quite a number of sociologists employ systematic observation in their research, but quantitative observation based on systematic sampling in space or time is rare. In market research, such methods are common, however. Intensive sample-based studies of consumer behavior are widely used.

Libraries and users

But let us start with the basics. Why do we need to monitor change? What is the value of traffic data?

Most people see libraries as collections of published documents such as books, scientific journals, newspapers and popular magazines.  And it is true that libraries have collections at their core. But collections are established to serve communities of users. Librarianship is not book collecting. A collection without users is a document archive, not a library. The heart of library work deals with the interactions between users and documents.

In the past, most users had to visit the physical library to get access. The documents could either be used in the physical library or, if borrowed, in the user’s personal space at home, at work or, for that matter, anywhere he or she could bring a book.  I remember a picture of my old philosophy professor maybe forty years ago. He was also a famous rock climber and dangled on the ropes in the Himalayas with a book in his hand. I even recognized the volume, a bilingual edition of Spinoza’s Ethica that belonged to the philosophy library. Taking library books abroad was forbidden at the time, but nobody brought a case.

Today, the pattern of use is changing, particularly in academic libraries. Undergraduates remain heavy users of the physical library. People doing research or post-graduate studies tend to work elsewhere, since they can access most of the documents they need on the web. At the same time, many libraries expand their range of activities. The way we do academic work, whether we call it learning, teaching, study or research, is changing. In the past, information was a scarce good. Today, information is basically a free good. The production of new knowledge is becoming more collective, more interdisciplinary and more market-oriented.

Academics prefer to work upstream, producing research articles for their peers, hoping that others would take care of dissemination, public debate and practical applications. In the past that sometimes worked. Today we are too many. There is nobody “down there” that will turn the river of research into policy proposals, action plans, news bites and slogans. To reach society, we must connect upstream and downstream.

Advanced students and researchers work under a double pressure. While they struggle for degrees, project funds and tenure, they are also expected to serve a range of user communities through other channels: lectures and interviews; popular summaries; social media and, increasingly, pedagogical resources on the web. To achieve social impact, they often have to work in groups that combine many different skills.

Academic libraries are well placed to support these changes. Some of the skills the new academics need are available in libraries. Libraries have always offered training in information retrieval. In a digital environment, practical information management has become an essential part of study and research.  ICT skills, multimedia skills and general study skills may be added. When institutions start their own digital journals and set up their own academic repositories, libraries may become de facto publishers. Libraries have always offered space for individual study. In the new world, they also support group work. When student and research groups engage more deeply with the web, they need flexible and supportive physical environments. The old computer labs are too rigid.  Many libraries take the opportunity to expand their field of action.

Sweep projects

The data collection approach developed by Given and Leckie has been used by a fair number of North American investigators. Some studies were carried out by students as part of their degree work while others were initiated by libraries that needed data for planning purposes. The first follow-up study, by Young (2003) had a very practical goal in mind. The researcher directed the library at a small liberal arts college in Virginia. She “wanted to purchase new furniture that students wanted to use” and at the same time “encouraged students in behaviors that should be promoted in an academic library”). Reducing the noise level was also a goal. Clark University in Massachusetts carried out seating sweeps in 2004, in connection with the redesign of their Goddard Library (Mott, 2013).

The University of Massachusetts at Amherst has integrated observation studies into their library planning since 2005. Every year the library collects hourly data on the number of patrons that are present in different areas of the building. The Amherst studies do not register activities. They do, however, publish extensive spreadsheets, which makes re-analysis possible.

Howard Silver, who published his Ph.D. thesis on the Use of collaborative space in an academic library in 2007, did his fieldwork at the Bryant University Krupp Library in Rhode Island in 2005. Silver used the sweeps method combined with brief interviews. His interest was also practical, but with a longer time perspective than the universities mentioned above. Silver had noted the decline in the number of visitors and was concerned about the future of academic libraries. Would physical libraries survive at all in a digital environment?

His objective was ... to determine how and why the collaborative spaces in an academic library were used, and how well the observed use matched the intent of the people who designed and managed the spaces. Silver concluded that the future of academic libraries was brighter than expected. The transition from print to digital media implied a shift from traditional teaching to self- and group-directed learning. University managers were willing to support libraries in a new role as learning spaces. With the design of new libraries increasingly emphasizing support for collaborative activity, librarians need to understand how and why their users are working together in library spaces.  But libraries had to change: they needed to redesign their physical space and to document their contribution to student learning. No published studies quantify the impact of collaborative spaces in academic libraries on student learning behaviors.

In Singapore, Pin Pin and Mokhtar bin Ramli (2008) used seating sweeps as one of their methods to study student behavior at the Li Ka Shing library at the Singapore Management University. The authors conducted two sweeps a day for six days in April 2008. Like many others they emphasized the changing functions of the library. They paid special attention to activities in the new Collaborative Study Area. The whole study was planned as an input to the “design, creation and continual improvements of its spaces to meet the needs of its community.“  The authors noted that the library is now “used as an event and training space”.

Most (2009) investigated a US public library in a rural, small-town environment (Gadsden County, Florida). Like the Canadians, Most wanted to understand the role played by the library in the community. She combined quantitative and qualitative data collected via surveys and interviews of adult library users, interviews of library public service staff members, structured observations – seating sweeps – of people using the libraries, and analysis of selected documents. Most concluded that libraries serve their communities as informational places and as familiarized locales rather than as third places, and that the libraries support the generation of social capital for their users.

In a Peruvian study carried out in 2006, Condori Soncco (no publication date, but after 2009) used seating sweeps to study the Central Library of the National University of Santa.  Only a few data from this study are available on the web, however. Soncco emphasized that the library had not recognized the need for group space. “In the reading room …  users always modified the arrangement of tables and chairs so that they could sit in groups, usually in groups of 2 to 4 persons”. The library needed “to adapt the spaces and to develop new services directed at users that work together in groups” (my translation).

May and Black (2010) studied three urban and three small town public libraries in Nova Scotia, Canada, using the same mix of methods  and arrived at similar results. The authors emphasized the success and value of libraries: libraries are vibrant places ...  they are community spaces used in a multitude of ways ...  provider of books and information, provider of access to technology and provider of a social space where members of the public are welcome. At the same time, each library was also revealed to be a unique place, reflecting the particular qualities of the community and the physical space of the library building itself. The conclusion is correspondingly broad: public libraries are complex institutions which play a variety of valuable roles in the community.  

In 2011 a team of interns carried out seating sweeps at sixteen branches of the Edmonton Public Library in Alberta to answer four questions: What are customers doing in EPL’s spaces? How would customers like to be using EPL’s spaces? What are current and future trends in library spaces and customer activities? How could EPL’s spaces best meet the needs of EPL’s customers? The thirty page report contained a very detailed analysis of empirical data and made extensive use of presentation graphics.  It also included and discussed photographs. In a more recent study, May and Swabey (2012) used seating sweeps to study user behavior at five smaller institutions in Alberta, Canada: Lethbridge College, Red Deer College, Grant MacEwan University, Mount Royal University and Southern Alberta Institute of Technology.

Management and practice

Track The Traffic was developed as a management tool, in cooperation with several public libraries, at Oslo University College (now Oslo and Akershus University College). Tests were carried out in a large metropolitan library (Oslo), in two medium-sized urban libraries (Lillehammer and Drammen) and in a small rural & residential municipality near Oslo (Gjerdrum). After this development period, TTT was also introduced as a possible choice for field work projects by students. As part of their three year BA program, library students at OUC do five weeks of fieldwork at the end of their second year. Students could choose between six different topic and TTT turned out to be quite a popular choice. Between 2007 and 2010 about ninety traffic studies were completed, covering all types of libraries.

All participants went through a brief training workshop and could contact their supervisor (TH) during field work. The final reports, which were graded, varied in quality, of course, but as a whole this material provided a good baseline for studies of user behavior in Norwegian libraries. The written reports are in Norwegian and had to be completed quickly, towards the end of the field period. Most of them are not suitable for publication. This would, in any case, require the permission of the student in question.

Høivik has, however, made summaries, in English as well as in Norwegian, of some of the best reports. He has also started to publish results from specific library categories such as public (Høivik, 2010) and academic libraries (Høivik, 2012), The TTT projects had a double function. For the students they served as a hands-on introduction to empirical library research on a small scale (about ninety hours of work). For the libraries in question they offered a chance to see their activities from a new and unfamiliar angle. For many librarians, this was their first concrete exposure to evidence-based librarianship.

Some students went deeper. A team of three Erasmus students did a fairly large study of the main library at Oslo University College. A detailed presentation of their methods and results are available on the web, see Arango et. al. (2009). Two bachelor students used their second year fieldwork as an inspiration for their final third-year papers, which require three hundred rather than ninety hours of work. Both studied public libraries. For details see Høivik (2010)

Shadowing the visitor

In Norway, as in many other countries, systematic observation represents a new approach to user studies. During the last decade interest in observational methods has increased, however. A fair number of libraries have initiated their own traffic counts. In 2007 five of the largest public libraries in Norway carried out an observation-based study using a different methodology. Instead of “sweeping” the building at regular intervals, the observers shadowed individual visitors from the moment they entered till the time they left the library. Observation was discontinued if the users settled down at one place, however. Since observation was combined with a brief exit interview, it was also possible to link background and behavioral data.

A similar, but less time-consuming approach was used in a study by van Beynen et al. (2009) in the Nelson Poynter Memorial Library at the University of South Florida. “For three semesters during 2007 and 2008, random samples of visitors were observed as they moved around the first floor of the building. Researchers were equipped with a map ..., a list of possible visitor activities, space to track visitor stops, and a section for comments. … Visitors were observed from the time they entered the library until they a) settled at their intended destination; b) left the library; c) left the first floor; or d) five minutes of observation had lapsed.” Breaking off after five minutes can save lots of time, but makes the data less compatible with the sweeps and the TTT method.

The Norwegian project, which was called The Metropolitan Study (Storbyundersøkelsen), is the only public library “shadow study” I am aware of. The results have only been published in Norwegian (ABM-utvikling, 2008). Shadowing the user provides the same type of data as the other two: what is the user doing when and where? But these observations are sequential: they follow each user through space and time. They are pearls on a string rather than pearls in a bag. Shadowing therefore gives much more information about individual behavior than the TTT approach. But complexity has a cost. Shadowing is even more time-consuming than a full-fledged seating sweep. Empirical studies need balance. When we invest in complex and detailed data collection, we must balance that with a more complex and balanced analysis.

But the two methods are highly compatible. They map and quantify the same types of activities. In Norway the full study covered about six hundred visitors in each library. This corresponds to between 150 and 300 hours of observation - a substantial amount of time just to collect the data. The libraries used their own staff as observers, but had to engage a commercial firm to train the staff and to analyse and present the data. The consultants produced a fat report with many tables and diagrams, but with no substantial analysis of the findings. At the end of the day, most of the data were left unprocessed. The 2007 study was never repeated. The moral is clear: without processing, data remain dead. The more data you gather, the more time you have to spend on data analysis.

Trends and comparisons

This experience suggests a weakness at the strategic level. Evidence-based librarianship can not rely on single studies. We need to compare results between libraries and over time. Valid knowledge comes from repetition. We must repeat in order to recognize patterns, to study systematic variations and to discover deviations and inconsistencies. Far too much library research take the form of ad hoc projects. Tools are often invented on the spot. There is a widespread lack of methodological continuity. To overcome this fragmentation the scholarly community needs to do two things. The first is to develop a small number of standardized, validated and not too expensive instruments. Low cost design is important. Costly studies will not be repeated. The second is to select and support such instruments whenever possible. Such a process must be consensual. Standardized instruments can not be imposed. They must be developed through discussion and attraction rather than administrative decisions.    

In retrospect we see TTT as a simplified version of the sweeps approach. Track The Traffic was created without knowledge of the Canadian studies. But since both depend on observations made while walking through the library, the categories used turn out to be similar. The sweeps are broader and more detailed, however. They are clearly designed to gather large amounts of data including the following: who was using the library (i.e., gender and approximate age), the activities in which those individuals were engaged (e.g., reading, writing, talking, eating, sleeping, and using library computers), the library location in which those activities occurred (e.g., book stacks, computer terminal, printer, and public telephones), and the personal belongings that those individuals had with them (e.g., briefcases, cell phones, laptop computers, food and drink, and baby carriages). The sweeps were also meant to be used together with other methods such as interviews, surveys, and document analysis.

TTT projects

Seating sweeps as originally defined are fairly labor intensive. They require people who can devote large chunks of time to the project. But the Canadian approach was standardized enough to develop a real following. TTT is even more focused. It was designed to be useful for library staff and students with limited time on their hands. The method generates a smaller and highly standardized data set. The instrument maps sixteen activities by location (functional zones). Only one activity is registered for each person. Track The Traffic includes demographic data (gender, age group, ethnicity) as an option, but not as the standard. Group size and personal belongings are not part of the TTT instrument at all. Since the instrument is standardized, data can easily be compared between libraries. Since the cost is low, TTT can be applied several times in order to observe trends in user behavior.

Since 2010, TTT studies have been carried out at the polytechnic colleges in Gjøvik, Østfold, Akershus (twice) and Oslo (thrice). A private business oriented university has carried out a study for internal planning (personal communication). The library in Drammen did a repeat study. In 2011, several public libraries in the county of Buskerud organized a regional statistical network. Two of these libraries (Nedre Eiker, Øvre Eiker) completed traffic studies in 2012. Since 2012, TTT has been part of the curriculum at LATINA, a digital training program for librarians and college teachers. LATINA stands for Learning and Teaching in a Digital World and is based at the Oslo and Akershus University College Learning Centre.

Recent LATINA studies using the TTT instrument come from Makerere University Library, Uganda (2012, unpubl.); from Oslo and Akershus University College (2013, unpubl.); and from Mykolas Romeris University, Lithuania (Olechnovičius, 2013). TTT has also been tried out in two medium-sized public libraries in Germany (Ulm) and Switzerland (Wintherthür) (Heinz et al., 2011), at Tampere University in Finland (Lehto et al.,  2012), and at East Carolina University in the United States (James, 2013).


The main purpose of this essay is to make the scattered literature on traffic studies easily available for library managers. I have included studies on public as well as on academic libraries, since the methodology is identical and many of the topics similar. But after a decade of (intermittent) work in this field I am willing to make some general comments as well.

Observation studies generate commitment as well as data. Reliable statistics help library staff focus on the same issues.They frame and stabilize discussions inside the organization. Traffic studies have the additional benefit of involving the staff directly, since they can, and ought to be, engaged as observers.

But there are costs involved. When we study libraries, it is very tempting to concentrate on data that are easily available because they are collected as part of our business processes. Our digital systems generate more data, in fact, than we can comfortably analyze. Libraries generally find it easier to count the number of loans than the number of visitors. All loans are registered, by manual or electronic methods. Electronic systems can show the number of loans day by day, or even hour by hour. We can study loans by subject and, if the law permits, by the type of student.

Studying users is much more tricky.  Most academic libraries now have electronic counters at the entrance. Visitors cross the beam twice – going in and going out, and libraries often calculate the number of visitors by dividing the total by two. But that tells us nothing about who the users are, about what they are doing, and about the time they spend inside the library. Our systems allow us to track the flow of books in great detail, but only scratches the surface when it comes to user data.  Traffic studies is a response to the scarcity of data on user activities.

Collecting, processing and analyzing data takes time, however. I believe TTT gives a decent return on investment (ROI), but only if it is well done. Statistics is not the same as research, but systematic data collection is a central component in all empirical research. The methods we use to collect management data must be judged by the same criteria as ordinary academic research.  A small and well-designed pilot is more valuable than a slapdash “big study”. When we use statistics in decision making, data quality may in fact be important than in the average research article. Research findings go through long institutional processes before they are translated into action (if ever): peer-review, replication studies, meta-studies, small-scale testing. Decision data guide action then and there.

My final point is strategic. I take it for granted that people turn to statistics because they want to learn something new about their world. Statistical data are not needed in a stable world. If you know the past, you can predict the future. No further input is required. In a changing world, statistics are useful if you want to study and prepare for change. But people and institutions who cling to the status quo, should not collect data. If they collect data, they should not analyze them. If they analyze them, they should not act on them.