Private lives and public libraries.
A quantitative approach to the study of user behavior

Purpose

The paper presents and demonstrates an effective way of gathering statistical information about user activities (traffic) inside libraries. We outline the method, describe the procedures in sufficient detail for practical use, present traffic data from about twenty Norwegian public libraries and indicate how such data can be used for advocacy and strategic planning.

Design and methodology

Librarians have, in general, very little systematic information about activities inside their libraries. Staff meets users every day, so there is no shortage of impressions, ideas and mental images. But this information is ad hoc and qualitative rather than systematic and quantitative.

Count the traffic (CTT) is a cheap and simple method to gather such data. It is based on regular and systematic “tours of observation” through the public areas of the library. It can be carried out by the library’s own staff rather than by hired consultants. The data gathered will tell you, in some detail, about the structure of activities in the various parts (zones) of the library throughout the day (daily cycle) and the week (weekly cycle).

The data analysis is not technically difficult: standard spreadsheets will do. Many libraries should be able to do their own data processing without external assistance. The generation and evaluation of relevant tables requires some skills in standard social science research.

Findings

The method has now been applied in more than fifty Norwegian libraries – including public, academic, special and school libraries. Additional CTT studies will be carried out by library students in 2009.

When we compare the data with our intuitive expectations, it is fair to say that in public libraries

1. the actual use of computers – including personal (mobile) computers – was higher than expected
2. the frequency of activities carried out in groups – involving children and students in particular – was higher than expected
3. purely social activities – not involving computers or media – were higher than expected

Research limitations

Gathering data by direct observation is generally time-consuming. CTT was designed to minimize the data collection effort by using time sampling. But we do lose some detail. In 2007 five of the largest public libraries in Norway carried out an observation-based study using a different methodology. Instead of “sweeping through” the building at regular intervals, the observers “shadowed” individual users from the moment they entered till the time they left the library. Since observation was combined with a brief exit interview, it was also possible to link background and behavioral data.

This project, which was called Storbyundersøkelsen (The Metropolitan Study), gives more information about individual behavior than the CTT approach. The two methods are highly compatible. They map and quantify the same types of activities. The shadowing method is much more labor intensive than the sweeps method, however. A combination of relatively frequent CTTs with occasional “shadow studies” may therefore be the best overall data collection strategy.

Practical implications

CTT was designed for practical use. It show how – and to what extent – the various parts of the library are used, throughout the day and through a typical week. Using traffic counts, libraries are able to document the type and intensity of use. This is useful for allocating resources and – not least – for reorganizing the library space.

In a municipal setting, CTTs generates new types of data that are likely to be of interest to politicians and senior administrators. Some data will also be useful in contacts with parents and other stakeholders among the general public.

At the regional and national level, reliable data on user behavior make “life in the library” visible in a new way. Loans and visits are rather abstract categories. Reading, talking, browsing and using computers provide more vivid images of what the physical library actually provides.

Originality

The method provides important new data on user behavior in the physical library. The method can be applied in all types of libraries – and indeed in all types of visitor-oriented institutions.

It is cheap enough to be repeated on a regular basis, and simple enough to be carried out by library staff or young students.  The observation categories are standardized, but easy to use, since they are based on normal social concepts.  All libraries that choose the CTT approach will generate comparable data. We publish our data sets on the open web – and encourage others  to do so as well.

Both the method and the data should be of interest to all persons that support evidence-based librarianship.

---

Introduction

Count the traffic (CTT) is a method for systematic mapping of user activities in libraries and other visitor oriented institutions. For many years libraries have mainly measured their success by counting the number of loans, or by the number of loans per capita – based on the population served. This statistic is still important. But usage patterns are changing. In many Western countries, the amount of lending is going down.

The number of library visits captures another important dimension of library use. The average Norwegian visits a public library five times a year – and borrows slightly more than five books or other media. But it is not the case that every visit results in a loan. About fifty percent of the visitors do not borrow anything at all. They visit the library for other purposes - to read, to play, to study or to use the internet. To understand what is happening in the public library sector, we must explore a wider range of indicators.

That is easier said done. Researchers may play around with many different types of data in their own projects. But meaningful library indicators require repetition and standardization. Today we need new types of data to understand new types of behavior - on a regular basis. But official library statistics are highly resistant to change. Once a set of procedures and publication patterns have been established, they tend to be contiued by their own momentum. Why create additional work for everybody involved?

CTT tries to develop something new in the space between the individual project and the statistical bureaucracy. Like LibQual, CTT is a standardized instrument designed to be used for comparative purposes. Standard methods of data gathering and analysis are common in most scientific and technical fields. We find scales of intelligence and physical agility - for individuals, of corruption and quality of life - for countries; and of windiness and hardness in nature - the Beaufort and the Moh scale respectively. The most developed instrument in the library field is probably the German library index BIX, which combines seventeen statistical variables into a single additive index.

CTT is not a scale or an indicator as such, but a lightweight method of data collection that can be used to develop useful indicators. To realize its knowledge potential, libraries that use it must provide open access to their data sets.

Libraries are more than collections – they are places where people participate in a wide range of activities. Standard library statistics reveals very little about activities inside libraries. In this paper we show how to gather and analyze such data. The method has been tested by library staff in several Norwegian libraries, with guidance from the author, and also tried out by more than fifty second-year library students as part of their second-year practice periods (five weeks in late spring).

The basic method is simple: at fixed intervals - usually once an hour - an observer should walk through the public areas and note down the number of visitors, and what they are doing, in various parts of the library. This should be repeated for at least one weekly cycle. The counting days should preferably be spread out during several weeks or months, to reduce the impact of random or seasonal events. All the information should collected in spreadsheets and analyzed.

In the English-speaking world the method is known as seating sweeps. Two Canadian researchers, Lisa Given and Gloria Leckie, first used the sweeps method in 1999, to study user behavior at the Toronto Reference Library and the Vancouver Public Library. The Norwegian version has been developed in close collaboration with several Norwegian libraries and has now been tested out in more than fifty institutions. We have labelled the method CTT - or Count The Traffic -  in English, from the expression  In Norway it is known as TTT – from the expression Tverrgående TrafikkTelling (transversal traffic counting)

In Norway, the method is introduced to library students as follows:

The purpose of this study is to investigate how library visitors actually use library services – inside the library itself. The method has been tested in Lillehammer and Drammen [two medium-sized cities], and you may therefore compare your results with those of other libraries. CTT can also be applied ito academic and special libraries and to other institutions with extensive visitor areas and acitivities.

What we observe in this study is not the individual user, but the different activities that take place inside the library.

All library workers know that the level of activity follows certain patterns or cycluses. The traffic varies throughout the day, the week and the year. More accidental variations come on top of this, for instance due to the weather, special events and moveable holidays (Easter, Whitsun). We do not try to observe on a continual basis. Our observation schedule depends on a sample of observation times. We must see to it that this sample is representative. Observations should therefore be spread evenly throughout the day, while the actual counting days ought to be distributed over several weeks.

To undertake this study, you will need the following resources:

• A list of activities – with explanations
• A floor plan covering all the public areas of the library.
  
• The plan should be divided into functional zones: reception area, newspaper section, children’s books, etc.
• A time plan, with dates and times for all observation rounds (”sweeps”).
• A stack of observation forms.
  
• The form combines the zones, on one axis, with the activities, on the other.
  
• Observers fill in one form per observation round.
• A set of guidelines for observing activities and entering data.

The list of activities is standardized. Additional categories and sub-categories may be added, but the basic structure should be retained, to facilitate comparisons between libraries. CTT includes fourteen basic categories - with a fifteenth for anything not covered by the basic set.

Table A. Fourteen basic categories

The actual floor plan and the division into zones will of course differ greatly from library to library. An architectural plan may be available – otherwise the students were asked to make their own sketch of all areas open to the public.

The data collected allow us to calculate the number of hours library users spent on each of the fifteen activities, in each designated zone, for each hour of the day and each day of the week. This means that the level of detail is high. The many factual details may be of great interest to the staff in that particular library. But for comparative purposes - between libraries and years - we need to summarize the data.

Sandvika Public Library

The most basic summary looks at the activities only - disregarding time and space, so to speak. As an example I take a data set from Bærum, a large municipality just west of Oslo. Sandvika Public Library is a large branch library in an urban area, serving more than twenty thousand people in the western part of Bærum.

Table B. The distribution of time between activities, by rank

 

Data source

• Place: Sandvika Public Library, Bærum, Norway
• Time: May 2008 (one weekly cycle) and February 2009 (one weekly cycle)
• Author: Sydsæter-Knudsen (2009), p. 41
  

I would not use the fifteen observational categories for comparisons, however. They provide too much detail. The pattern of activities becomes clearer if we aggregate categories into larger sets. One approach is to distinguish between traditional and modern library activities. Here I take modern to cover computer based and group activities, while everything else - except other - is categorized as traditional

Table C. The distribution of time between aggregated activities


Data source: Table B

Indicators are statistical variables that express or represent interesting properties of the libraries - or entities - that we study. The CTT approach allows us to study a number of properties that are not covered by official library statistics. The most important ones, in my view, are the following:

1. To what extent does library use have a modern character?
2. To what extent do activities involve groups rather than individuals?
3. To what extent are computers being used?
4. How much of the time at the library is spent interacting with staff?

Some additional questions are also of interest:


The corresponding indicators are easily defined by grouping the appropriate categories. The four main indicators - let me call them A as a group, covers:

1. Modernity: 2 + 4 + 6 + 8 + 9 + 10 + 11 + 12
   
2. Group use:  2 + 4 + 6 + 8 + 10 + 12
   
3. Computer use: 9 + 10 + 11 + 12
4. Staff contact: 9

The three in the second rank (B) follow:

1. Own computer ratio:  (9 + 10)/(9 + 10 + 11 + 12)
2. Browsing time: 5 + 6
3. Non-media time: 1 + 2 + 3 + 4 + 14
   
• does not include contact with staff, which often involves media
4. Stand-up time: 1 + 2 + 5 + 6 + 14
   
• does not include contact with staff, which might happen from either position

It would be more correct to remove category 15 - other activities - from the percentage basis. But since it is generally quite low, I have not bothered to do so in this paper.

Sandvika and Høvik

To see the indicators at work, we compare the branch library in Sandvika with Høvik,  a small branch library in the same municipality. The time use patterns differ substantially:

Table D. Eight observation based indicators, Høvik and Sandvika

The differences between Sandvika and Høvik are clearly related to demographic as well as material factor