Guidelines for Electronic Media Statistics Frederick C. Lynden 08 Oct 1993 21:19 UTC
Please reply to Jan Bruusgaard at the address below; not to SERIALST. -ed. ----------------------------Original message---------------------------- Currently librarians are very interested in keeping statistical track of media, especially electronic media. At a recent meeting of the IFLA Standing Committee on Statistics, Mr. Jan Bruusgaard, Norway, presented "Remarks on compiling statistics on electronic media". It is an excellent document, and Mr. Bruusgaard would like comments from American librarians by December 15. His address is: Jan Bruusgaard Government Administration Services Documentation Services P.O. Box 8129 0033-OSLO NORWAY His paper will also be considered by the ARL Statistics Committee at its annual meeting. Once he was received all comments it will be published by the IFLA statistics section. Thank you for your assistance. Frederick C. Lynden Brown University AP010037 at Brownvm Remarks- Compiling Statistics on Electronic Media I. Purpose A. For libraries it should be a goal to document all types offered as well as possible. B. Electronic media may have a larger and larger part of the library's acquisition budget. Statistics will give grounds for the spending of these resources. This can also explain the use of resources to training and motivation. C. Electronic media would partly be a supplement to and partly a substitute for printed media. Probably the part that substitutes printed material would bring about a decline in the traditional statistics on printed material.For this reason it is important to document the corresponding increase in electronic media. D. Statistics on electronic media may make it possible to estimate the value and benefit of databases in relationship to: 1. printed material 2. other databases or electronic media 3. cost of the material E. Statistics would give management information to the library for planning and budgeting. II. Background and concepts. A.Parts 1. In the traditional statistics we only have one part - the library itself. Because of the complexity of the statistics of electronic media it is natural to include: a. database hosts, vendors and producers b. Software producers and developers 2. If we succeed to make standards which are acceptable to a. and b., with a view to which kind of information that is relevant to libraries, the compilation of data would be much easier. B. Products. Electronic media consists of different products that have to be handled in different ways. 1. Databases are the most important. These can be classified by medium: a. Online-databases b. CD-ROM databases 2. Databases can also be classified by type: a. Bibliographic or reference databases b. Fulltext databases c. Factual databases Reference databases may often be used as a tool in retrieving printed material, while fulltext databases would be used as a replacement of or supplement to printed material. 3. A third way to divide databases are into: a. textual databases b. picture databases c. (In the future) sound databases 4. Other types of electronic media that may be important are: a. software b. multimedia-packets C. Ownership. Electronic media differs from printed material in the way that there is a more complexstructure in the case of ownership. 1. Printed material would be the library's own property. Some of the electronic documents would also be. In addition electronic media may be: 2. rented 3. only the data owned by others are permitted to be used, e.g. databases owned by the producer. 4. Different ownership leads to different placement. a. The media can be placed inside the library b. or it can be placed outside the library, where only the terminal and telecommunications network give the library access to the data. III. Measurement A. How to measure electronic media? 1. Count a. Manual counting. This is done by many today and is the simplest way, e.g. number of CD- ROM-records, number online-database searches executed. b. Electronic counting. A major tool may be the development of software that count and give printed repots. This software has to work together with the other software that is used in the retrieval process. 2. Sample tests may be another possibility of measurement. B. What to measure. 1. Collection. a. Macrolevel. (1) Number of CD-ROMS records, number of diskettes (2) Number of databases available. This would give a picture of how a large part of the electronic world the library has access to On the other hand many database hosts offer many databases where as a single library only takes advantage of a few, e.g. Dialog who offer hundreds of adifferent databases. b. Microlevel (1) Number of bytes. One byte corresponds to one letter. This measure may tell something about the size of the databases. (In the computerworld the measures kilo-, mega-, giga-, and terabytes are used.) One problem here is that a database both consists of pure data and indexes (overhead) which may be 2-10 times as large as the pure data. (2) Number of records/documents in the database. Together with an average size of the records, in bytes, number will give a good picture. The average size of the records is important, because the size of one record would vary much since a pure bibliographic record is much smaller than a record with abstract or a fulltext. 2. Use a. Number of loggins to a database will give a measure of how many requests are solved by each single electronic source. b. Time. The spending of time is an easy measure.Here we count number of minutes which are used on connecting each database (online- bases), or number of minutes on a terminal i being active (CD-ROM). c. Number of bytes fetched from disk (database) would give a rather physical measure of use. d. Number of records fetched from disk would give the number of logical units that are retrieved. This measure has to be used together with the average size of the record. e. Number of bytes/records printed to a printer or file show what the user gets out of the session.This measure removes the material which is judged as irrelevant. This measure can be seen as a parallel to loan, because this is what the user can bring out of the library. f. If we presume that cost used on a single unit reflects benefit, spending is a suitable measure.Costs can be divided into: (1) Once only costs (2) Subscription costs (3) Cost for use or pr minute It is likely that a combination of the measures mentioned above is the best to get a suitable picture of the use and collection of electronic media. One possibility is to develop an abstract measure which represent the different factors with unequal weights. IV. Problematic areas A. The development of standard counting procedures require some standardizing work done by the software and database vendors. IFLA or ISO may take the initiative in setting standards or guidelines among these vendors. B. Even if the library's own catalogue is a database, and often the most used one, this should be separated from other electronic media. Union catalogues makes the border unclear. C. Protection of individuals. This kind of measures gives spin off effects that can be used to measure the effectiveness of the single employee, and maybe map the single person's use of the media. It is important that this kind of measurement is done in agreement with the users and employees. V.More advanced possibilities - research. There are two fields from where the knowledge can be utilized to get better measures of electronic media. A. Information theory. Using some information theory it maybe would be possible to develop measures about what is new knowledge and what only are duplication of existing knowledge that also are located in other sources. Many electronic media have a great deal of overlap, often in a much higher degree than printed material. B. Information retrieval. It should be possible to use knowledge and methods from this field to give a more precise answer to what is retrieved and used, e.g.compared to the principles of recall and precision. 19.8.93 Jan Bruusgaard, Government Documentation services POB 8129 Dep N-0033 Oslo Norway Member of IFLA's SC section of statistics.