Library visitors, collection management and simple sampling procedures

Citation

Ole Pors, N. (2000), "Library visitors, collection management and simple sampling procedures", The Bottom Line, Vol. 13 No. 4. https://doi.org/10.1108/bl.2000.17013daf.002

Library visitors, collection management and simple sampling procedures

Library visitors, collection management and simple sampling procedures

Keywords Collection management, Library users, Sampling frame, Statistical forecasting

Introduction

In this column, I will discuss two very simple sampling procedures that can be used with benefit in library settings. One of the sampling procedures relates to the number of visitors. The other relates to patterns of lending. You could argue that presenting sampling procedures like these in an electronic environment are unnecessary, but I hope that readers of this column will grasp that a manual and focussed sampling can be less expensive, time-consuming and more precise than using automated tools. I will refer to research I have undertaken using the procedures.

Number of library visitors

First, I will delimit the problem to customers that come to the library. This is not about virtual access, but physical library visits we are here examining. The number of visitors or customers coming to the library for different purposes is a very important indicator of market penetration. It is becoming more important, because an increasing segment of our library patrons come to the library with purposes other than lending. All of this investigation is connected to the supply of more services that we finance in our libraries.

Counting visitors is often done with help of different turnstiles or other automatic counting tools. Such tools are becoming commonplace. I have not yet seen any study about the accuracy of these counting devices. The accuracy depends on the pattern of people visiting the library. How many of the people are staff or other personnel like delivery persons visiting the library? Is it possible that a group of people is counted as just one visit? These tools may be good enough if you just want a picture of the access to the library. But still, there are purposes where the employment of other tools could be useful.

If you have to conduct a user survey, I have often found that a traffic count is a very effective instrument in relation to the sampling of respondents. You will very often see that sampling or respondents or visitors or users are conducted in a statistically very doubtful way. You have to try to control your sample in a way that increases the trustworthiness of the sample in relation to the unknown population. Different groups of people use the library at different times of the day and the week. There are idle times and busy times. It is important that a sample of users is drawn in relation to the pattern of visitors. It is time-consuming to count all visitors. Sampling can be used.

In user surveys, I have often used a sampling method to count visitors. Let me give an example. A public library is open 50 hours per week. We choose randomly two consecutive weeks. The total opening hours is 100. In each of the hours, we pick a five-minute interval, for example 12.45-12.50. We count, in that interval, the number of people either entering or leaving the library. We do this for each of the 100 hours, which constitute our sample.

With a rather small effort, we have now discovered our pattern of visits (see Table I).

On the basis of such data, it is very easy to plan a sampling procedure. If we want to conduct a user survey consisting of 400 respondents, we simply use the proportions and distribute the 400 questionnaires according to the proportion.

• •

  Monday, in the period 12.00-13.00, we have to distribute 0.009*400 = four questionnaires.

• •

  Monday in the period 13.00-14.00 we have to distribute 0,016*400 = six questionnaires.

We have now controlled our sample in relation to the patterns of visitors and their possible different needs. We can further use our sample data to estimate a confidence interval for the average numbers of visitors. We can see that there is quite a deviation in the average number of people visiting the library per hour.

Calculating averages

The average number of visitors per time interval is 5.5. We calculate the standard deviation and it is in this example 1.9. The sample is 100. We simply put these figures into the formula for calculating the confidence interval. The confidence interval is 5.13 to 5.87. We can state with 95 percent confidence that the true mean in a five-minute interval is placed inside the confidence interval. It is then rather easy to calculate the minimum and the maximum number of visitors per hour, per week or whatever interval you want to use. The estimate will, of course, be more precise if we increase the sample size.

I have tried to demonstrate the usefulness in relation to user surveys and to estimating the number of visitors using a simple and easy sampling frame.

The collection issues: what do users borrow?

We know quite a lot about users information search behavior at the moment. In relation to literature, we know that browsing is the most common method of information seeking. We also know that people come to the library with a multitude of information needs. Some of them are concerned with subjects, other with categories of fiction and some are more precise and stated as a request for a specific title or literature by a specific author. The request for specific titles is less extensive than most librarians would think and it tends to be concentrated on a rather small amount of titles. The relationship between the holdings of the single library and the users reading or borrowing pattern are then of tremendous interest. There is substantial literature about this subject.

To investigate the distribution of the collection in relation to subjects through the means of automated systems it is rather easy. It is also easy to get information about use of non-fiction through the system. It is much more difficult to get reliable information about users' use of fiction, simply because you normally do not employ a classification system in relation to this kind of literature. Collection management librarians would appreciate indicators of interest in relation to different categories of their given collections. In the official and national library statistics, at least in my country, you cannot see the lending figures for the libraries split up between fiction and non-fiction. It is only split between adult and children.

So, in such an example, if you want to investigate the pattern of borrowing a simple method is to take consecutive 1,000-1,500 returned books (and journals) and take them to a room and inspect them, tally their Dewey class number for the non-fiction and categorise the fiction. An example of fiction could be the following:

• •

  lyrics and poems;

• •

  thrillers;

• •

  crime novels;

• •

  science fiction;

• •

  historical novels;

• •

  family sagas;

• •

  problem-oriented novels;

• •

  romantic novels;

• •

  experimental literature;

• •

  classics; and

• •

  mainstream novels etc.

This classification is just one of many possible. The important factor is that it covers the acquisitions and it is in accordance with the collection policy. Give a couple of librarians the task to classify the fiction. This is done extremely quickly. On the basis of this simple sample, you have a very good picture of your users' reading pattern. Conducting the same kind of sampling from the circulation system can be done, but the categorization of fiction becomes much easier with the document in hand.

We have used this method in a collection development investigation at a Danish library. We found it extremely fast and easy and we coupled it with collection data we obtained by systematic sampling of every fiftieth document in the collection. On the basis of these two types of data, it was very easy to identify the so-called overused and underused areas of the collection.

Is one sample of a random 1,500 consecutively returned books a valid sample? Possibly, if you look for patterns and not titles. You would normally repeat the exercise once a year simply to judge and evaluate use patterns.

This kind of sampling in relation to collection management has been used extensively by the British librarian A.W. McClellan and by the weeding expert S. Slote. The principles in this kind of sampling are sound and simple. They are focussed and the results will nearly always be surprising. It is due to the simple fact that most of the activities in relation to the collection are hidden, done by browsing without any interference or interaction by librarians. The librarians will tend to focus on the rather small proportion of requests concerning authors and titles.

Conclusion

In this column, I have demonstrated the strengths of very simple and basic sampling frames to be employed in investigation of the patterns of visits and the patterns of reading. The methods and they provide information they will give a more user focused library and allow library managers in all units to eye how well their library budget is spent on sections of their given collections.

Niels Ole Pors Associate Professor at The Royal School of Library and Information Science, Denmark

Further reading

McClellan, A.W. (1978), The Logistics of a Public Library Bookstock, Association of Assistant Librarians, London.

Pors, N.O. (1991), "Users, collection use and on-line searching", International Journal of Library and Information Research, Vol. 2 No. 3, pp. 63-76.

Slote, S. (1998), Weeding Library Collections, Library Unlimited, Littleton, CO.