- of single statistical statements (data, “facts”),
- of statistical tables,
- of statistical databases and
- of visual presentations derived from numerical data.
The actual production of statistics takes place in several different settings. The main production system, which delivers what we call official statistics, is designed and operated by the Norwegian government. The main production work is still done by the libraries, which have to gather and report the information in the first place. But an increasing amount of statistical production takes place outside the official framework, facilitated by the widespread use of digital tools. The main producers of unofficial statistics are database vendors, library researchers and the libraries themselves. Research and development projects usually have a statistical component, Some librarians also use their library systems to produce specialized statistical reports, especially in connection with lending and downloads.
At the official level, the government decides on the kinds of information that should be reported. The Ministry of Education and Research is responsible for academic and school libraries. The Ministry of Culture is in charge of the public library sector. Since 2010 the actual data collection has been carried out by the National Library. Before that the Norwegian Archive, Library and Museum Authority (ABM) was in charge. The flow of data is substantial. Public libraries have to report values on 177 different variables. Norway has about 430 municipalities, each with its own public library system. The annual data set contains nearly eighty thousand different numbers (data points). The corresponding statistics for academic and special libraries cover about 310 different library units and 145 variables. The full data matrix contains about forty-five thousand values.
Thirteen KOSTRA indicators
The information is stored in databases operated by the National Library. Public libraries report their information directly to the base through web based forms. Fifteen of the 177 variables are also reported to KOSTRA, a special information system for municipal statistics. KOSTRA is a vast and sophisticated statistical system which covers all municipal sectors. In addition to its collection of variables, KOSTRA also define sets of indicators for various municipal sectors – including libraries. All KOSTRA data are available for consultation and re-use through a very advanced user interface. This makes it easy to compare indicator values from different libraries (snapshots) and different years (time series).
KOSTRA indicators for public libraries have been available since 2002 (data collected in 2001). In the early years of the decade, very few people used them or showed any interest in their use. In 2004 the present author wrote an article on KOSTRA in a Norwegian library magazine, predicting increased use of KOSTRA data – by budget cutting managers – in the future (REF). He also criticized KOSTRA for its choice of “loans per librarian” (actually loans per FTE) as its lead indicator. After a substantial amount of professional mobilization and lobbying, KOSTRA changed its indicator set in 2006 (REF). The current set consists of thirteen indicators:
- K1. Library visits per capita
- K2. Loans per capita
- K3. Book loans per capita
- K4. Children’s books: Loans per child 0-13
- K5. Adult books: Loans per adult 14+
- K6. Loans non-book media per capita
- K7. Operating expenditure as a percentage of total municipal expenditure
- K8. Operating expenditure per capita
- K9. Accessions (all media) per capita
- K10. Media and salary expenses per capita
- K11. Inhabitants per staff member (FTE)
- K12. Turnover rate of children’s books
- K13. Turnover rate of adult fiction
Thirty ABM indicators
The KOSTRA database offer unique opportunities to analyze trends and patterns and to follow individual libraries over time. To judge from published information, on web and on paper, very few libraries have chosen to do so. Central library authorities have also abstained. ABM was the organization in charge of library development from 2003 to 2010. Instead of engaging with KOSTRA’s “official” indicators, it started its own public library indicator project. In 2010, ABM launched an alternative set of thirty indicators, without presenting the KOSTRA alternative.
A. Economic indicators
- A1 Library share of municipal expenditure = K7
- A2 Cost per hour open
- A3 Media cost per capita
- A4 Share of (salary + media expenses) that go to media
- A5 Costs per downloaded document
- A6 Share of media expenses that go to electronic ressources
- B1 Physical visits per capita = K1
- B2 Virtual visits per capita
- B3 Loans (physical) per capita = K2
- B4 Downloaded documents per capita
- B5 Share of visitors with a non-Norwegian language background
- B6 Share of borrowers from other municipalities
- B7 Visitors to library events per capita
- C1 Hours open (per year)
- C2 FTEs per 1000 inhabitants = K11 (inverse)
- C3 Accessions of physical media per 1000 inhabitants = K9 (multiplied by 1000)
- C4 Media donated by the Cultural Fund as share of total accessions
- C5 Stock of media in languages other than Scandinavia and English
- C6 Turnover of media in languages other than Scandinavia and English
- C7 Ratio between (distance) loans from and to other libraries
- C8 Share of total loans supplied by own collection
- C9 Reference queries per staff member
- C10 Share of working hours dedicated to work with the public
- C11 Share of working hours dedicated to national or regional fereference services
- C12 Share of working hours dedicated to work with schools
- D1 Share of working hours dedicated to development projects
- D2 Share of working hours dedicated to (own) professional training
- D3 Events per FTE
- D4 Cooperation projects per FTE
- D5 Number of interactive services
Only five of the thirty overlap with the thirteen indicators from the KOSTRA set.
KOSTRA and ABM
From a conceptual point of view both indicator sets have their strong and weak points. The great advantage of KOSTRA is the fact that all its data are available. Some of its time series actually go back til 2001. So far (May 2012) nobody has published a single data set for a single library covering the thirty indicators proposed by ABM. The KOSTRA set is modest and practical. It builds on variables that are included in the present production system. The ABM set is ambitious and idealistic. It wants to cover many new areas, such as immigrants, reference, school related work and project activities, and asks for information about new variables that will require new production routines. Notably
- The number of visitors with a non-Norwegian language background (very difficult to measure)
- The stock of media in languages other than Scandinavian and English (rather easy to measure)
- Loans of media in languages other than Scandinavian and English (rather easy to measure)
- The number of reference queries (reference statistics are collected, but need radical improvement)
- and also the number of working hours dedicated to
- work with the public (defined by work plans, but many grey areas)
- national or regional reference services (the national “Ask a Librarian” service closed down on March 31, 2012, however)
- work with schools (quite difficult to measure)
- development projects (quite difficult to measure)
- (own) professional training (fairly difficult to measure)
Nine familiar indicators
The indicators that public librarians tend to know, are the indicators they meet in the annual report fra ABM. In the main table, which lists every single municipality, each library is described by nine indicators. Eight of these overlap with KOSTRA. Four overlap with the proposed set of thirty indicators.
- Library visits per capita = K1 = B1
- Loans per capita = K2 = B3
- Children’s books: Loans per child 0-13. = K4
- Adult books: Loans per adult 14+. = K5
- Operating expenditure per capita = K8
- Accessions (all media) per 1000 inhabitants = K9 (multiplied by 1000) = C3
- Media and salary expenses per capita = K10
- Inhabitants per staff member (FTE) = K11 (inverse) = C2
- Medieutgifter pr. innbygger = A3
Three sets of indicators
In the public library sector we have identified three relevant indicator sets:
- The nine indicators actually used by ABM in its annual statistical publication – which I call ABM9
- Thirteen indicators from KOSTRA, which is managed by Statistics Norway (KOSTRA13)
- Thirty indicators proposed by ABM in 2010 (ABM30)
To what extent are these indicators actually used by their intended audience. Hardly at all. Loans and visits are of course observed and discussed in all libraries with any interest in statistics. But these numbers are traditional measures. They have always been presented and commented on by librarians. They are, furthermore, often published in their raw form, as observed variables (numbers, quantities, values) rather than as indicators.
In smaller libraries the reason is often the following. The librarians involved get the number of loans and visits from their own systems. At the end of the year they can easily compare this year and last year. If they want to calculate per capita figures, however, they have to consult other statistical sources. It only takes a couple of minutes to find the population figures if you know how. But they don’t know how. So they stick to what they know. The basic step from statistical variable to statistical indicator is not undertaken.
Indicators and variables
Indicators represent more sophisticated concepts than ordinary variables. Indicators are not the numbers we observe, but relationships between these numbers: rates, ratios, proportions, percentages. I must add that the tendency to report observed variables rather than indicators is widespread. The profession is not accustomed to discuss rates, ratios or indexes. The mathematics involved is trivial. Students learn the techniques in the early years of secondary school. But the students I meet in library school often stumble. Fractions and percentages are tools that decay if they are not used. The greater complexity of rates and relationships come on top of that. What do turnover rates mean?
Such questions must be adressed to the library profession rather than to statisticians. We can develop indicators that fit the concepts you want to study – as long as the concepts themselves are meaningful. The development of indicators goes hand in hand with the development of concepts.
As a small experiment I will publish the paper Indicators without customers, for our satellite conference in Turku, as a series of blog posts.