Visits and visitors

Library associations work at the national level. For them it is important to develop the use of national library statistics.

Many countries do not yet collect public library statistics. Other countries provide only scattered numbers. In such situations, advocacy depends on the development of national systems for collecting, processing and publishing the data. Library associations can encourage and support such efforts. But developing and maintaining large systems must normally be left to the government.

Fortunately we have an alternative to national library statistics. Single libraries need not wait for national systems. If they are eager to use statistical data for management and advocacy, they can develop their own statistical systems.

I believe that local efforts will help development at the national level as well. Libraries that establish local systems may well function as test-beds for future national systems. The librarians involved will gain expertise in operating small statistical systems.  They will learn how to argue with statistics in their local communities. At the same time they will learn how to use systematic data from their services to manage their libraries. Such a pool of expertise will be very useful when countries decide to set up national systems.

Start small

I believe in a gradual approach to statistical development. Start small – not big. Look at what you are doing already – and go on from there.

I say this from experience. I have worked with many types of social and economic statistics – in many different countries – for more years than I care to remember. I have seen some successes – and many failures. It is surprisingly hard to design statistical systems that work well. But mistakes and failures are not a problem if organizations are ready to learn from them.

Governments tend to believe that statistics can be created by administrative decisions. That is not the case. The government can to some extent force libraries – or schools or businesses – to provide data about their activities. But they cannot force libraries to apply statistics in their daily work. Librarians will only employ statistics actively if they believe in their value.

Producing reliable data is hard work. If library statistics are only used for administrative control from above, libraries suffer. They must do a lot of work collecting and reporting data, but get very little back in terms of interesting analyzes and effective arguments. So why bother?

Conflicts of interest between data producers (libraries) and data collectors (government) slows down statistical development. Statistics are created to be useful. But they will only be employed by the people that produce them if they suit the actual needs of libraries.

Wise governments collaborate with libraries.  Wise libraries collaborate with governments – and with each other. Wise institutions consider their experiences- and learn from their mistakes.

A stepwise approach

Single libraries can develop their own statistical systems.

Such local initiatives work best as learning processes. They should preferably be shared with the wider library community from the start. That way everybody has a chance to learn from your experiences. Your library becomes more visible, and you often get useful feedback from other library people.

The best approach, I believe, is to do this step by step. A stepwise model may look like this.

• Describe your current situation – with regard to library statistics
• Publish this information on the open web
• Explore your current possibilities
• Take one step forward – by developing one of these possibilities
• Evaluate the results – after a suitable time (week, month, year, …)
• Repeat the steps above

I assume that every library can describe its current use – or non-use – of library statistics. I also assume that the libraries we discuss have workable internet connections.

Web publication is free. Many different publication systems are available. A combination of gmail, Google Docs and WordPress blogging is widely used, however. That is the one I will refer to when examples are needed – since it is the one I know from my own experience.

Libraries that lack web access can also develop local statistics, but doing so without the web would require rather more work.

User statistics

The current possibilities will of course vary from country to country and from library to library. But I assume that all public libraries will want to say something

• about their users
• about their lending
• about their data services
• about their staff

Since we want to do advocacy, it may be good to start with the users.

Users are concrete, living persons. They are easy to imagine inside the library – borrowing books, doing school work, reading newspapers, sending e-mail to friends and families. Thus, statistics about library users should be highly relevant to politicians and administrators.

Users are voters and future voters.  So let us count them.

A librarian that wants to collect data by observing users, can basically do it in two different ways: by sitting down or walking around. Libraries that lack electronic counters can do “manual counting”. This means that somebody from the staff sits down near the entrance and counts the number of people that enter the library.

Traditional head counts give very little information.But stationary observation can be improved, so that we get more interesting and useful data.

Another approach is to walk around at regular intervals and observe what people do inside the library. For some years we have been trying out a standardized method for “walking observations” in Norway. The method, which is called Count The Traffic (CTT), has been tested in about seventy libraries in Norway since 2006. Both public, sacademic and school libraries have been studied.

The results are encouraging. We have been able to produce new and interesting user statistics by methods that librarians and library students can apply on their own.

The method is fully documented on the web – in English as well as in Norwegian. Here you will also find links to several case studies as well as an article about the results from twenty public libraries.

Till now we have concentrated on gaining experience and getting reliable data. More work is definitely needed to interpret and apply the results. But interpretation and use is primarily a task for practitioners.

Acting on the results

‘The data may show that some parts of the library are too crowded, while other parts are underutilized.  In Norway many libraries find it hard to atttract teenagers. Reading rooms behind doors are seldom popular. people like to feel the flow of life, it seems. Computer queues are common – and many libraries still lack Wi-fi so people can bring their own portables.

Researchers and library teachers can participate in change processes. But only librarians can carry them out – and draw the right conclusions from their experiences.

What does it cost?

Observation takes time. A medium sized library that wants to collect statistical data by observation must be prepared to invest at least one full week – and preferably two – on the data collection itself. By medium sized I mean a library that has a few hundred visitors every day.

Full-time observation is rather concentrated work.The observers will need breaks. Observation should therefore be shared among two or more persons . You must also expect to spend two or three days on

• coding the data
• entering the coded data into spreadsheets
• calculating tables and diagrams
• discussing and interpreting tables and diagrams
• comparing your results with other libraries
• publishing your results on the web

Doing it for the first time, things will move slowly. After the first year, the process can be speeded up. Data processing will usually go much faster the second time.  Once the libary has a first data set, it may reduce data collection to one weekly cycle rather than two. The total time cost will probably be about three person weeks during the first observation year. For updates in later years, six to eight days may be sufficient.

A small practical test

If you like the idea of such a project, I recommend doing a very brief pilot study before you decide whether to “jump into the water”.  That can be done very simply:

• select one hour to carry out the pilot test
• find a place where you can sit down and observe the entrance while making notes (without being too conspicuous)
• bring a small stack of paper
• whenever a person or a small group (friends, parent + child, etc) enters the library, make a numbered note that shows
  • the time of arrival
  • sex and age (roughly)
  • something that can identify the person
• do the same whenever a a person or a small group leaves the library

If you have enough time and can observe most of the library from your seat, you may also make some notes on what the persons do inside the library.

Manual counting is expensive – since a person has to watch the entrance all the time. Counting every day is therefore impossible (except for very small libraries). Instead, libraries must select a small number of counting days, during which everybody is counted. These data are used to calculate the average number of visits per day. If the library keeps open 250 days per year (say), we multiply the average by 250 to get the annual total.

In small libraries it may be possible for the librarian on duty to do the counting in addition to her or his regular work. But it takes an extra effort. Medium-sized libraries must probably assign a staff person (on rotation) to sit near the entrance throughout the day.

Library authorities often recommend to do one full week of counting in the spring and one in the autumn. From a statistical point of view it is better to disperse the counting days, however. The library could for instance select one day in January, one day in February, and so on. Dispersed days provide more accurate data, since they are less influenced by external events and seasonal fluctuations.

The cost of data collection

If the library keeps open (say) five days a week, this means – in any case – devoting ten full work days to data collection. If the observer only counts the number of vistors, the library spends ten days of work to calculate a single figure – the annual number of visitors. That is not a very productive process. The cost is high and the benefit low.

Some libraries reduce the work load by counting for two days rather than two weeks. But this approach makes the figures very uncertain.  Your annual visitor numbers will go up and down for no particular reason. Two single days cannot “represent” a full year in a faithful way. It is better, I believe, to spend two full weeks, but to use those weeks to get broader and deeper data that can be used for advocacy and planning. The cost remains high – but we improve productivity by increasing the statistical benefit.

Counting and observing

Manual counting is based on direct observation. Electronic counters can only tell you how many people visit the library. If you want to know who these people are, what they do and how long they stay, you need to observe them. Such data are very useful for advocacy. It is also easy to combine them with stories about individuals and their benefits from using the library. Thus, manual counting is more than an alternative to electronic counting. It is the first step towards systematic observation.

ll over the world, libraries tend to gather data about users by counting the number of library visits. Visits are either registerede automatically, by electronic counters at the entrance, or manually, by direct observation.

Let us look more closely at the concept of library visits.

Concepts and measurements

Concepts and measurements are not the same thing.

Take electronic counters as an example. They do not measure visits as such – they count how many times an electronic beam has been interrupted. The counter does not distinguish between an ordinary visitor, a staff member, a big dog – or a child who runs in and out ten times in a row.

A genuine library visit, I would say, starts when a person enters the library premises in order to use one or more of the library services.  A library visit ends when the person leaves the premises.

A single visit corresponds to two “hits” or steps on the counter: one when the user enters and one when the user leaves.

But if the person only leaves for a few minutes, to buy a soft drink or make a phone call, she (or he) would not say that she visited the library twice. In social terms, a single library visit may include brief excursions outside.

A person who pops into the library to use the bathroom, would not speak of this as a library visit either. A real library visit is defined by its purpose.

Social concepts

A visit to the library is a social concept.

It is defined by the way we understand and use it in daily life. An electronic counter is an instrument for counting the number of library visits. But the instrument is not perfect. Very few instruments are.

An electronic counter will normally give values that are a bit too high – because the counter catches some events – entrances and exits – that do not correspond to  library visits in the ordinary social sense.

If people enter in groups, it may also miss some visits. Two or more people walking close together will often be counted as just one person.

To get more accurate data from a counter, we need to calibrate the counter. That means comparing electronic and manual data – and to calculate a correction factor.

APPENDIX

Observation sheets

An observation sheet may look like this.

Heading

• Place: XX public library
• Date: June 2, 2010
• Time: 0930-1030
• Observer: NN

Observation notes

• 1. 0932. Young girl (10?) arrives. Blue dress.  Returns two books. Goes to the children’s area (CA)
• 2. 0935. Adult woman (brown glasses) with small boy (blue shorts) arrive.   Mother leaves boy in CA. She reads newspaper.
• 3. 0940.  Two teenage boys come together (blue sneakers; Spiderman T-shirt).  Use computer for gaming.
• 4. 0945. The blue dress girl (1) goes out
• 5. 0947. The blue dress girl (1) returns
• 6. 0948. Older woman arrives. Black dress. Returns a book. Browses fiction shelves.
• 7. 0954. Older woman (6) borrows two books and leaves.
• 8. 1001. Older man arrives. Beige jacket. Returns several books.
• 9. 1010. Three girls around twelve arrive together (ponytail; green skirt; red satchel). Work together at a table.
• 10. 1011. Blue sneaker boy (3) leaves
• 12. 1012. Man in beige jacket (8) leaves without borrowing anything.
• 11. 1015. Blue dress girl (1) borrows some books and leaves.
• 12. 1017. Adult male (light blue shirt) arrives. Goes to computer.
• 13. 1024. Spiderman (3) leaves
• 14. 1025. Older man (8) leaves
• 15. 1029. Woman and boy (2) leave.

Processed observations

Observation notes are raw materials. They have to be processed to be of value. First we must extract the basic results about each visitor:

• Young girl (1) spent 43 minutes in the library.
  • Returned books. Read in CA. Borrowed books.
• Adult woman (2) spent 54 minutes in the library.
  • Came with somebody. Read newspapers
• Young boy (2) spent 54 minutes in the library.
  • Came with somebody. Used CA.
• Teenage boy (3) spent 31 minutes.
  • Came with somebody. Played computer games.
• Teenage boy (3) spent 44 minutes.
  • Came with somebody. Played computer games.
• Older woman (6) spent 6 minutes.
  • Returned book. Borrowed two books.
• Adult man (8) spent 11 minutes

Three young girls (9) and adult man (21) and were still in the library when observation stopped at 1030.

Afterwards we code the data and enter them into spreadsheets. These operations will be described elsewhere – and are not necessary for the test itself.

Note on privacy

What people do openly in public spaces is seldom very sensitive information. But systematic observation is still a bit different from occasional and random observation. So I would like to add that librarians are professionally obliged to respect the privacy of users. Data about single individuals should not be shared, whether they come from the books they borrow or from observation studies.

Time series

Electronic counters are very effective, since they register visits automatically.

To get the most from your data, I recommend noting the number of visitors daily – preferably in a spreadsheet – and to use these figures to present and evaluate the amount of traffic – day by day, week by week, month by month, and year by year. Such data – which statisticians call time series – help us see the fluctuations in traffic. The number of visitors often follow typical patterns through the week and through the year.

Such data also make it easier to observe the relationship between visits and external events: a local festival,  a national election, a big sports event, a sudden heat wave, and so on. Very little additional work is required to increase our understanding of visitor flows.