Wednesday, August 22, 2012

PL 35/12: What to do when nothing happens?

Filed under: iflastat, statistics, traffic — plinius @ 11:22 am

I take it for granted that international standards should be more than suggestions and proposals. They should express and codify best practices in their subject areas.

If the standards are not based on existing professional practice, but represent a call for innovation, their effect ought to be monitored. Standards that are recommended to – but not applied by – the library community represent a problem that the designers need to address.

This means that I am concerned about the gap between proposals and practices in the field of library indicators. At the Statistics and Evaluation Satellite meeting in Turku this August, I presented a paper documenting this gap, with Norway as an example.

  • Picture: Turku organizers with friends from Moomin Island

In this post I show, with a small example, that ISO standards may face similar problems.

Public seating

ISO Standard 11620:2008 proposes a set of performance indicators for libraries. Two of these are Library visits per capita (B.2.2.1) and Public seating occupancy rate (B.2.3.1).

A simple Google search for the exact string (“library visits per capita”) demonstrates (what we already know) that the first indicator is widely used for planning, comparisons and advocacy. More than twenty thousand documents were found.   Eight of the top ten contained or referred to empirical data about the indicator.

A similar search for information about the second indicator shows no sign of empirical use. Only thirty web documents indexed by Google (checked August 21, 2012). None of these provided or referred to empirical data about the indicator.

I am not saying that nobody calculates occupancy rates. I have done a bit of empirical work myself, as part of the TTT project (see below). I hope to do more. I have also found a few articles that discussed such rates. But the amount of published work (on the web) is so limited that it does not show up in a Google search.

I interpret this as a lack of demand, or interest, in the second indicator among the professionals we are trying to support. I see this as an issue that ought to be adressed by the “library indicator community”. Proposals without practice are paper tigers.

Library visits per capita

Documents with empirical data

  • Public Libraries in the United States: Fiscal Year 2003
    • This report includes national and state summary data on public libraries in the 50 states, the District of Columbia, and the outlying areas, with an introduction, findings, and numerous tables. 
    • The report is based on data from the Public Libraries Survey, Fiscal Year 2003, and includes information on population of legal service area, service outlets, public service hours, library materials, total circulation, circulation of children’s materials, reference transactions, library visits, children’s program attendance, interlibrary loans, electronic services and information, full-time-equivalent staff, operating revenue and expenditures, and capital expenditures.
    • The report includes several key findings: Nationwide, library visits to public libraries totaled 1.3 billion, or 4.6 library visits per capita. The average number of Internet terminals available for public use per stationary outlet was 9.5.
  • Data Shows Library Visits at Historic High. National Humanities Alliance, 2011. Quote:
    • December 2, 2011. Data analyzed from the FY 2009 Public Library Survey (PLS), a census of public libraries in the 50 states, DC, and the territories, shows a 24.4% increase in library visits per capita in the last ten years, with total visits increasing by nearly 40%. In 2009 (the most recent data available) libraries were visited a record-breaking 1.59 billion times, reports the Institute of Museum and Library Services.
  • Romanian Public Libraries Situation –A Panoramic View of the Year 2001. Director Doina Popa, Cluj County Library Octavian Goga.
    • Library visits per capita: 0.75
    • Library visits per active user: 8.45
  • Education > Statistics > Public Libraries > Library Visits per Capita (most recent) by state. Published by Contains latest available data for all American states.
  • Library Visits per Capita – Library of Virgina. Contains tabulated data for public libraries in Virginia 2005-2010 (Financial years). PDF file.
  • FY 09-10 Data Tables Library Visits Per Capita. Contains data for public libraries in Florida, grouped by the population of the service area. XLS file
  • Library Visits Per Capita. Bar diagram and table with values for public libraries in the population range 5,000-9,999, in Colorado in 1998 (!)
  • America’s Star Libraries: Score Calculation Algorithm. Methodological note about the LJ Index, with numerical examples, in Library Journal, 2011.

Documents without empirical data

  • Looking-Around-Inside-the-Library. Chapter 6 in CAMEO Handbook. Methodological comments about a series of indicators. CAMEO =Community Analysis Methods and Evaluative Options. Based on discussions in the 1990s. Quotes
    • Library Visits per Capita relates the number of visits made to the library to the population total the library is established to serve. The raw number of total library visits can be ascertained by installing a door counter that counts all traffic or by counting library visits during a typical week multiplied by 52 (the number of weeks in a year).
    • A typical week is one during which the library is open its regular hours (no holidays) and which is neither extraordinarily busy nor slow. Libraries that have strong seasonal variation in library use, as in resort or academic communities, might consider two or more sampling weeks.
    • The annual use of the library can be an impressive figure and very useful for public relations purposes. For example, “More people use the library than attend the Baltimore Orioles baseball games.” Or, “More people come to the library than travel through National Airport in a year.” Or, “More people use the library than participate in Richmond’s parks and recreation programs.”
    • The information can also be used to gauge the wear and tear on the building.
    • Seasonal variations in library visits can be used to alter staffing patterns.
  • Output Measures for Public Libraries. Short comment on methods and interpretation by Moya K. Mason, 2012.
    • If the analysis shows that numbers of visits per capita are lower than expected, a library can take measures to increase patron use by reviewing such things as the collection development policy, library hours, and staffing levels. 
    • … it would be in the best interest of libraries not owning a turnstile to rent one for the length of the measuring because there are too many reliability issues involved with using a person to do it. Even if he or she is competent, they could get tired or become distracted.
    • But there is not any way to account for the number of homeless people who enter a library but do not actually use library services, unless a regular member of the staff is on hand to identify particular people who are well known, and they are seen going to a specific area of the library (i.e. the bathroom). 

From a statistical point of view, the “non-empirical” documents seem analytically rather weak.

Systematic observation

The Public Seats Occupancy Rate can only be calculated through systematic observation inside the libraries in question. The objective of PSOR, the ISO committee states, is

  • to assess the overall use rate of public seats provided for reading and working in the library,
  • by estimating the proportion of the public seating in use at any given time. 

The number of public seats tends to be constant. The occupancy rate therefore depends

  • primarily on the number of visitors inside the library at any one time,
  • and secondarily on the activity pattern of the users

Many users visit the library only briefly, in order to borrow or return books. They spend most of their time standing or walking around, without using the seating provided. Users that spend longer periods at the library will tend to sit down most of the time.

The committee writes:

  • Measurement may be conducted at specified time of the day, the week or the year, e.g. peak times or off-peak times.
  • This should be stated explicitly when using the performance indicator. 
  • Due to the inherent variability of the performance indicator, a more accurate performance indicator may be attained by measuring the Public Seating Occupancy Rate at random intervals over a period of time and then calculating the mean occupancy rate

In the TTT project (CounT The Traffic) we have now carried out systematic observations of user activities in nearly one hundred libraries. Both public, academic, school and special libraries have been included. Since most of the studies have been done by second year library students, the quality of the reports vary quite a bit. But since the students were trained beforehand, and since the methodology is concrete, fully documented and illustrated by case studies, the basic data look valid.

Theological University case study

When we started (with students) in 2008,  TTT did not include calculation of occupancy rates, but this has been changed in the last revision of the protocol. We now recommend systematic mapping of public seating by functional zone.

The independent Theological University of Norway (“Menighetsfakultetet”) is located in Oslo. It is a small institution, with about 900 students and 90 staff. Here I summarize the results from a Count The Traffic study (TTT) conducted in May 2010. The university used the following zones

  • A. Main service desk
  • B. PC area near desk (6 PCs)
  • C. Seating area near desk. Couch and chairs with about 8 seats.
  • D. PC room 1: 15 PC seats
  • E. PC room 2: 26 PC seats – inside the library
  • F. Group rooms: 2 small rooms with PC and 3 small rooms w/o PC.
  • G. Non-fiction shelves
  • H. Periodicals recent issues + books. 2 PCs and 8 regular seats
  • I. Seating area 1: 8 seats + non-fiction
  • J. Seating area 2: 8 seats + non-fiction
  • K. Seating area 3: 2 PCs and 22 regular seats
  • L. Non-fiction shelves. 1 seat
  • M. Study area 1 for master students. 40 seats
  • N.Study area 2 for master students. 20 seats
  • O. General study area. 70 seats.
  • P. Walk-through area. Mediatek (photocopier, microfilm reader, PC w/o web). Lockers.

The opening hours were

  • Monday 0800-2100
  • Tuesday 0800-2100
  • Wednesday 0800-2100
  • Thursday 0800-2100
  • Friday 0800-2100
  • Saturday 0900-1700
  • Total: 73 hours

Observations were carried out at thirty minute intervals:

  • Monday May 10 and May 31: first tour at 0830 and last tour at 1500
  • Tuesday May 11 and May 18: first tour at 0830 and last tour at 1530
  • Wednesday May 12: first tour at 0830 and last tour at 1530
  • Thursday May 27: first tour at 0830 and last tour at 1500
  • Friday May 7 and May 18:first tour at 0830 and last tour at 1530

The average attendance (occupancy) during the full weekly cycle was 72 persons. Average occupancy through the day was:

  • 0900: 34 persons
  • 1000: 47
  • 1100: 73
  • 1200: 58
  • 1300: 78
  • 1400: 80
  • 1500: 58 persons

Average occupancy through the week was:

  • Monday: 56 persons
  • Tuesday: 58
  • Wednesday: 60
  • Thursday: 73
  • Friday: 46 persons

Since the theological library had about fifty PC seats and about 185 ordinary seats (small group rooms at F not included), the highest seat occupancy rate observed (at 1400 = 2 PM) was less than thirty percent (85/235).

Humanities and Social Science Library, University of Oslo

A similar calculation at the library for humanities and social sciences at the University of Oslo gave much higher rates of occupancy. At this university the average number of visitors (occupancy) during the full weekly cycle was 351 persons. During the weekdays we found:

  • 0900: 279 persons on the average
  • 1100: 502
  • 1300: 490
  • 1500: 489
  • 1700: 357
  • 1900: 191

Average occupancy during the five first days of the week was 377 On Saturday the average occupancy was about forty percent lower (223). In the area under study, the library has a total of 438 seats for reading, 73 PC seats and 21 group rooms. If we assume that the average group consists of four students, the effective number of seats would be close to six hundred (587). The seat occupancy rates during the average day were as follows:

  • 0900: 48 percent of all seats were occupied
  • 1100: 85 percent
  • 1300: 83
  • 1500: 83
  • 1700: 61
  • 1900: 33 percent

The variation during the day throws light on the permanent struggle between students and university managers. The students claim that there are too few seats, and point to the lack of seating in the middle of the day. The administration points to the many available seats in the early morning and the late afternoon and ask students to get up early in the morning and work till late in the evening. This would lead to a better use of existing facilities.

In other words: a classic conflict of interest between work and play.

Interpreting occupancy data

The ISO committee says:

  • Due to the inherent variability of the performance indicator, a more accurate performance indicator may be attained by measuring the Public Seating Occupancy Rate at random intervals over a period of time and then calculating the mean occupancy rate.

In a country like Norway, the mean occupancy is not a very interesting indicator. It will always show a lot of spare capacity. It is not useful for planning, for trend analysis or for comparisons between libraries.

The mean occupancy must, by definition, be based on a series of observations through time and space. If we just calculate the mean – or the dispersion, for that matter – we “throw away”the most important information hidden in the data:

  • the time pattern of use
  • the space pattern of use

If we want indicators, the maximum occupancy, which shows the seating capacity when the traffic is highest, is more relevant. Users will only be hampered, and start complaining, when the load is close to 100 %.

That very rarely happens in public libraries, however. There is no shortage of seating as such – only a shortage of specific types of seating. Access to library PCs is still a problem in many libraries during peak hours.





  1. […] PL 35/12: What to do when nothing happens? Like this:LikeBe the first to like this. […]

    Pingback by ST 12/12: Strange visits « Global statistics for advocacy — Wednesday, August 22, 2012 @ 6:53 pm

  2. […] PL 35/12: What to do when nothing happens?. All chiefs and no Indians […]

    Pingback by P 51/12: Blogging i august « Plinius — Sunday, August 26, 2012 @ 11:28 am

  3. […] PL 35/12: What to do when nothing happens?. All chiefs and no Indians […]

    Pingback by PL 17/13: IFLA voices 2013 | Plinius — Wednesday, August 28, 2013 @ 11:16 am

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at

%d bloggers like this: