INFORMATION NEEDS AND USES: FIFTY YEARS OF PROGRESS?

Figure 1: a model of information behaviour

At that time I noted that more attention had been given to how people used information sources and services than to other areas of the diagram, such as information exchange, information use, and information transfer. There has been some improvement in that situation over the last thirteen years, but, in the period covered by this review, it would be true to say that most "user studies" have been about how people use systems, rather than about the users themselves and other aspects of their information-seeking behaviour.

With the field delineated in this way it would be possible to review developments over the past fifty years in several ways:

• according to the methods of investigation - interview surveys, mail surveys with self-completed questionnaires, case studies, etc.;
• according to the social role of the persons investigated - research scientists, teachers, social workers, Ph.D. students, and so on;
• according to discipline - science and technology, social sciences, humanities, etc.; or
• according to theoretical framework - the cognitive approach, the behavioural approach, the phenomenological approach.

All of these (and more) have been used by other reviewers and the last is, to my mind, the most useful but, unfortunately, so much work has been done without reference to any theoretical framework that it must either be ignored completely or the 'Miscellaneous' category would be very large indeed. Consequently, I have used Figure 1 to provide the basis for organization, grouping studies under the headings: Information-seeking behaviour (divided into Library Surveys and User-focused studies); Failure; Information use; Information transfer and exchange; User satisfaction; Methodological perspectives; and, finally, a consideration of related work in Other Disciplines.

A LITTLE HISTORY

Although, for convenience, the paper referred to above dated user studies from the Royal Society Scientific Information Conference of 1948 [7], the subject goes rather further back in time. For example, McDiarmid's The Library Survey [8] was produced in 1940 and that referred to various kinds of surveys - which were about how people used libraries and the needs they sought to satisfy (although, interestingly, there is no mention of information needs in the book) - dating back to 1916 [9] and with a spate of studies in the 1920s and 1930s.

It would be true to say, however, that the field broadened out from the study of library systems to the study of the behaviour and attitudes of information users in general with effect from the Royal Society Conference. For example, in a study led by Menzel at the University of Columbia [10] all of the references are from 1948 onwards, while Paisley dated his review [11] from 1948. Also, in a 1948 paper [12], Donald Urquhart presented:

"...an effort to discover, by means of a questionnaire addressed to readers of publications borrowed from the Science Museum Library, some information on such questions as how references to publications are obtained, what the expected information is required for, and whether, in fact, the publications contain the desired information."

and remarked that, "No earlier survey of this type has been traced."

The Royal Society conference was followed up ten years later by the International Conference on Scientific Information [13], Area 1 of which was devoted to, "Literature and Reference Needs of Scientists: Knowledge now available and methods of ascertaining requirements." Those presenting papers included some that were to become well known in this area, and who have been cited in this review, for example, Fishenden, Saul and Mary Herner, Herbert Menzel, and D.J. Urquhart. The studies reported were still largely system-focused, and included studies of medical scientists [14], forest scientists [15], and, a feature of the time, a number relating to scientists in the atomic energy industry and associated research units [16, 17, 18]. One of the most interesting papers was by J.D. Bernal, Professor of Physics at Birkbeck College [19], who provided a "List of enquiries in the field..." that ought to be undertaken. A number of topics on the list have been pursued in later years by various researchers, but it is still at least partially relevant. Bernal ended his paper with a statement that is worth recalling:

"...a knowledge of the requirements of the different users of scientific information and the uses to which they wish to put the information they secure should be the ultimate determining factor in the designing of methods of storage and retrieval of scientific information."

Another development of interest was the establishment of the Centre for Research in User Studies (CRUS) within the Postgraduate School of Librarianship and Information Studies at the University of Sheffield in 1976. The School at Sheffield (now the Department of Information Studies) had fostered user studies through student dissertations and occasional research projects since its foundation in 1963 and in 1976 the British Library R. & D. Department established the Centre as part of a national policy [20]. From 1985 to 1987 the Centre was partially funded by the BLRDD to undertake specific projects and it ceased operation in 1989, following the retirement of the then Director, Dr. Norman Roberts - although work in the area of information needs continues to be an important aspect of the work of the Department.

Over its lifetime, CRUS contributed significantly to the development of user studies in the U.K., with programmes in the humanities [21], the use of books and libraries in schools [22], and business information [23], with individual investigations in a variety of other fields including public reference libraries [24], metals information [25] and community profiling [26]. The Centre also ran training courses in research methods and published a quarterly, CRUS News, which has been transformed into the Department's Information Research News.

INFORMATION-SEEKING BEHAVIOUR

Library Surveys

Dividing the field into system studies and user studies, we find that there has been a continuing interest in the use made of information services, which can be subsumed under the general heading of library surveys. Among these, the use made of university libraries constituted a major sub-field, so much so that it warranted a review of the literature by Ford in the Journal's Progress in Documentation series in 1973 [27]. Ten years earlier there had been a number of investigations, some of which have a very archaic ring to them today. For example, Bowyer began the conclusion to his paper by noting that:

"The doctrine that the provision of books for undergraduates is a proper function of a university library is becoming accepted by university librarians and teaching staff." [28]

Hardly a matter for debate these days! Bowyer's paper was a narrative piece based on studies undertaken at Leeds [29] in 1961 and LSE [30] in 1962 and also on interviews with "the heads (or a representative) of twenty departments in the Faculties of Science, Arts, and Commerce" at the University of Birmingham during 1962, to elucidate answers to questionnaires circulated earlier. A common concern of all of these studies was the extent to which undergraduates were buying books: Bowyer found that "in only a very few departments was the purchase of books by students, in any real sense, obligatory", whereas Oppenheim found more book buying than had been anticipated, with women students buying less than men.

The Oppenheim study was of particular interest, having been undertaken by a social researcher (and author of a well-known text on questionnaire design) rather than by a librarian and, perhaps for that reason, was the only one with a clear, social science oriented research design. The aim was to compare first, second and third year students, to compare men and women, and to compare the impact of the course being taken by the student. Apart from the result on book-buying, noted above, the study showed heavy reliance on the library, with women rating the attitude of staff and provision of seats lower than did men. Again:

"There was some evidence to suggest that those who disliked the LSE library would seem to dislike reading generally: they owned somewhat fewer books, read somewhat less, and were less prone to read a book from cover to cover. They also less often used the lending library, and in extra-curricular reading their taste more often ran to fiction, especially crime, detection, westerns, and mysteries."

Also in the 1960s, Line was also undertaking studies at Southampton University - replicating the Tucker and Oppenheim studies to some degree, but focusing more on the attitudes of students towards the library and its staff and the ease with which materials were found. The first study [31] produced findings similar to those of other studies - low book buying, difficulty in finding books, extensive use of the neighbouring public library. The second investigation [32] found very little difference and suggested that seminars on library use had little effect, that the use of Hall of Residence libraries had declined, as had that of the public library. The fashion for investigations of this kind was not restricted to the U.K. - Salvan [33] in a response to Bowyer's paper noted, essentially, that the same problems existed in France and that its centralized direction of university libraries enabled a common response to those problems.

It is difficult to establish what notice was taken of the results in terms of effecting changes in the systems under study. Some changes, of course, did take place: an increase in user-education programmes, the extension of short-loan collections, and, in some places, the building of undergraduate libraries. How far the diffusion of these ideas was the result of the research and how far the result of other factors is very difficult to determine: it may be simply that the change in universities towards undergraduate library services (indicated by Bowyer's comment) was already in train and that the research only described what was happening, rather than leading that change.

University libraries were not the only type of library to be surveyed in the period: both public and special libraries experienced an interest in this kind of research. perhaps the most significant piece of work was that carried out by Clements [34] in 1967. Clements reported on an OSTI-supported survey of 33 public reference libraries using two questionnaires - one for personal visitors, the other for telephone, telex and mail enquiries. This was another "system study", but rather more interesting than most because it was able to quantify the nature of demand (from the analysis of almost 30,000 response forms) and also provided some analysis of the degree of satisfaction with services. The main need was for specific facts (33.4%), followed by information on a subject (17.8%). Employed persons were generally seeking information (rather than facts) and 70% found what they wanted: almost 76% of telephone and related enquiries were fully satisfied.

Special libraries, too, provided a base for work. Among the more interesting examples was the study of users at Shell Research, [35] which, prompted by Allen's work on the "gatekeeper" concept, did more than simply provide statistical data on that use. The report noted that those who made use of the greater number of services tended to be the above-average users of each service, and that the above-average use was shared by a small number of users:

"This means that some of those research workers who read more journals than average also read more than the average number of reports, and at the same time ask more than the average number of questions. This therefore suggests that they also carry a proportion of the total work load which is above average."

Some studies, of course, were conducted across information agencies of different kinds - initially concerning the information-seeking behaviour of scientists. One of the earliest was carried out in 1965: this was a survey on behalf of the Advisory Council on Scientific Policy. The sample of 6,194 scientists produced a 48.77% response rate. Although the text refers to "information-seeking and -using" rather than information needs, I would describe it as a system study covering the use of sources, especially abstracts, patents, reviews, journal papers, library use, and the use of abstracting journals. The report [36] included suggestions for future work and suggestions for action, including more reviews, trial publications, a need for balance of abstracting and "titles" journals, the development of a formalized personal contact system, and more positive reader service from librarians.

There was also something of a fashion for what might be called a level of micro-exploration: the information the user brought to the catalogue and how he or she used the catalogue. Again, however, this was an old concern - McDiarmid, for example, cites the work of Akers in 1931, [37] which examined which elements of Library of Congress catalogue cards students actually used. In more recent times, Ayres et al. [38] showed that the user's information was generally accurate and suggested that title catalogues might aid the user, while Kenney in a study at the International Labour Office library [39], showed that the subject headings were the preferred form for subject searches. Of course, these studies (among others) related to the card catalogue, which has all but disappeared today and it was interesting to see papers by Akeroyd [40] and Hancock-Beaulieu [41, 42] on the impact of OPACs on user search behaviour, updating this kind of investigation.

Hancock-Beaulieu found that the use of an on-line catalogue does not appear to have increased the extent of subject searching (in fact, seems to have decreased it). The manual PRECIS index supported the contextual approach for broad and more interactive search formulations, whereas the OPAC encouraged a matching approach and narrow formulations with fewer, but user-generated, formulations. Success rate of the on-line catalogue was slightly better than that of the manual tools, but, curiously, fewer items were retrieved at the shelves. Non-users of the bibliographic tools seemed to be just as successful in finding what they wanted - a fact that raises other interesting research questions, such as; What search strategies do those non-users adopt, and how do they differ from those of users?

User-focused studies

As noted earlier, work on information users and their needs is much scarcer than that on the use of systems and services, although the situation has changed a good deal over the past ten years or so. However, a considerable number of investigations have dealt with all aspects of the use of information systems and sources and, frequently, with a focus upon particular disciplines or work roles. These constitute a group that lie midway between the system-oriented projects and the person-centred studies.

In The scientist's view of his information system[43], Rowland reported on surveys undertaken for Royal Society Committee on Scientific Information and concluded that:

"The overall impression given by these results is one of conservatism, with traditional attitudes and practices continuing to prevail.... Presented with a series of doomsday scenarios,... they welcomed only those possibilities that represented minimal change."

But:

"Contrasting with the general conservatism, a liking for the recently developed online computer-based methods of information retrieval was clear."

Unsatisfied demand in this area resulted not from the lack of funds, but from lack of access to terminals.

Somewhat earlier, in the public library sector, a major interview study (506 interviews) of reading habits in three London boroughs, [44] which covered reading in general but also looked at public library membership and use, revealed gender differences in information use with, for example, 49% of men choosing "Scientific and technical information to do with your job or hobbies", and 38% of women choosing "Running a house". 36.5% of women and 27% of men chose reading materials in the field of "Health and hygiene".

Some twenty years later a major study was carried out in Baltimore, U.S.A. into the information needs of ordinary citizens [45]. In terms of overall research design and development of the research instrument, this study stands as a bench-mark for large-scale investigations of this kind. The study addressed the following issues:

1. "What are the information needs of the urban community?
2. How are these information needs presently satisfied?
3. Could institutional forms be devised to better satisfy these needs (i.e., more effectively and economically from the public's viewpoint)?"

within a conceptual model developed by Dervin, which linked the urban resident to information needs, information solutions to problems, and information sources, and which identified the psychological, intellectual, institutional, and societal barriers to the satisfaction of needs. This framework structures Dervin's review of the literature in the report (Chapter 2) and, more than twenty years later, is still a reasonably comprehensive picture of the research situation.

Table 1: Ranked list of problem areas

Most frequently mentioned topics	Rank
Neighbourhood	1
Consumer	2
Housing/maintenance	3
Crime and safety	4
Education	5
Employment	6
Transportation	7
Finance/public assistance	8
Health	9
Miscellaneous	10
Recreation	11
Discrimination	12
Legal	13

The main methodological development in the Baltimore study was the approach to information needs from the direction of the ordinary life and work experiences of urban residents. They were not asked to state their information needs, but to identify the problems they experienced, and then how they went about solving them.

The rank order of major information problems or questions cited by residents would probably stand today, although issues relating to Crime might be somewhat higher up the list shown above.

The survey design was rigorous, with the interview schedule undergoing four pilot tests before being used. The sample of 1,615 households was drawn according to a strategy used by the U.S. Bureau of Census and resulted in a refusal rate of 16% and a vacancy rate (i.e., the household address was unoccupied) of 3.8%. The resulting data were analysed by a variety of statistical techniques and the results revealed not only wide variation in the ability of residents to find answers to their problems, but also the lack of knowledge of urban residents' problems on the part of information professionals.

Partly as a result of Dervin's involvement, the Baltimore study sparked off other studies [46], which were reviewed by Chen & Hernon in their work on the information needs of residents in six north-eastern states of the U.S.A. [47]: Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont.

In the special libraries sector, Mote (who had carried out the work at Shell Research, referred to above) had sought to characterize users in an attempt to understand their differences in information use [48]. Mote identified three groups of scientists at Shell Research Ltd. according to the character of the discipline within which they were working:

1. those working in fields with well-developed underlying principles, well organized literature, and well-defined "width" of subject (e.g., organic chemistry);
2. those concerned with a wider subject area with less well-organized information (e.g., an organic chemist who is now concerned with both the physics and the chemistry of lubricants); and
3. an "exaggerated form" of 2) above, a scientist who covers more subjects, with problems involving greater variation, and almost non-existent organization of the literature.

The hypothesis was formed that there would be increasing need for information through the three groups, with a maximum for group 3). Researchers were identified and assigned to the subject types, their enquiry records were checked, and support for the hypothesis was found. Mote concluded that library/information services might be planned accordingly - self-service libraries for category 1) users and more intensive, information-worker support for categories 2) and 3).

Categorization was also a feature of Brember & Leggate's [49] study of medical library users' use of different kinds of information sources, which adopted Checkland's Soft Systems Methodology [50]. Three user types were derived: the practitioner, the researcher, and the practitioner-researcher.

More recently, Palmer [51] used semi-structured, in-depth interviews to probe personality, discipline and organizational structure as related to the information behaviour of biochemists, entomologists and statisticians working in agricultural research. Cluster analysis was used to group respondents, and five groups emerged, which Palmer called:

• non-seekers (mainly statisticians) for whom information was either a problem, because of the difficulty of finding it, or "was almost disregarded since it was rarely sought for problem-solving."
• lone, wide rangers (five statisticians and three entomologists), who were longer-serving members of the organization, subscribed to more journals, and regularly scanned outside their own subject fields;
• unsettled, self-conscious seekers, who were either new to the organization or new to the problem area with which they were concerned - they used the library more heavily than any other group;
• confident collectors (mainly entomologists), who had abandoned regular information-gathering, but who kept personal files and used a variety of ways (including personal networks) to collect information; and
• hunters (all biochemists) who maintained regular routines to make sure that nothing relevant escaped them.

Discipline, work role, time spent in the subject field, and organization were the most important determinants of the extent of information behaviour and there were some indications of male/female differences.

As a corrective to the considerable attention devoted to science and technology, a study of the Information Requirements of the Social Sciences (INFROSS) was launched in 1969. This study might be described as the last of the large quantitative studies of information-seeking behaviour, and exhibited both the strengths and the weaknesses of that approach. The strengths were in the adoption of a sound survey methodology, with follow-up to improve the response rate, and a scientifically drawn sample of 2,602 social science researchers from the U.K. population of researchers. The low response rate (41.8%), however, marked one of the potential weaknesses of such studies - the inability to represent the views of more than 50% of the sample. Non-response is affected by many factors, but one of the contributing factors in the case of INFROSS, as Line acknowledges, could well have been the size (20 pages) of the questionnaire [52].

The questionnaire results were augmented by a small-scale interview survey (125 persons) and, as part of another project, by observation of the day-to-day behaviour of social science researchers making use of an experimental university information service.

Some of the findings supported those of other studies: for example, there were similarities in the distribution of usage over different sources of information, and the survey method allowed the correlation of variables, leading to such findings as the positive association between age and perceived satisfaction, and the negative association between satisfaction and the need for historical and descriptive material. Line and his colleagues concluded that the main deficiency was the under-developed state of information services for social scientists - hence the experimental project referred to above. In pressing this view of the need for such services, INFROSS had a successful outcome: many U.K. universities introduced social sciences information services to mirror those already established for science [53].

With the start of a study at Sheffield University into information needs and information-seeking behaviour in social services departments (the INISS Project) in 1975, research took a new direction. INISS was conceived as an "action research" programme, in which the fact-finding phases would be followed by a stage in which innovations suggested by the research would be implemented in social services departments. With funding from, first, the British Library R. & D. Department and then from the Department of Health and Social Security, this programme was carried through between 1975 and 1980 [54, 55].

In reacting to the relative sterility of survey based research, INISS was also conceived as a qualitative research programme, using observation and in-depth qualitative interviewing. These methods, coupled with the participative research mode in the action phase, allowed "triangulation" of the data, providing a rich picture of the nature of communication in social services departments and the role of information-seeking in relation to the performance of social work tasks and the management of departments. The focus upon communication, work roles, and related information in an area of applied social sciences allowed the researchers to focus on types of information, regardless of source or format, rather than on formally published documentation, which had tended to be the focus in previous investigations.

At the end of the investigation, the researchers concluded that:

"...the action research strategy gave considerable scope for the creation of condidtions in which the [users' information] requirements could be met. The collaborative approach which grew up in the course of the Project amountyed to a redefinition of the role of the information specialist in relation to this user group. Instead of the introduction of a range of centralized (and, to the fieldworkers, peripheral) information services based upon traditional library and information work, it proved possible to 'tailor make' simple services for people on the basis of their diagnosis of their information problems." [56]

For those involved in Project INISS, it is interesting to see recent paper on how information services might make use of the extensive literature on information seeking behaviour [57].

In a subsequent investigation, the tools devised in the course of Project INISS were adapted to other local government settings and the results were published as two "do-it-yourself" manuals [58, 59]: a means of disseminating research results in a way that could actually lead to innovation in information services.

FAILURE

The studies of university library use led to another line of research: the so-called shelf-availability studies undertaken by the Cambridge Library Management Research Unit [60, 61]. These studies relied upon self-selected respondents (i.e., those who chose to follow the desired procedures) and, as a result a relatively small proportion of "eager" users provided the larger proportion of responses. However, the findings were of interest, revealing that most instances of failure were not the result of users' incompetence, but simply due to the fact that the wanted materials were already being used by someone else.

Another investigation explored failure in another sense: Martyn [62] explored the costs to scientific research and development of failure to find relevant materials at the start of a project. The study covered 647 scientists engaged in research and had an 88% response rate. 23.3% of the respondents (144 scientists) reported finding information too late for it to be useful - 43 of those scientists reported that, as a result, they had unintentionally duplicated other work. In a brief report for the New Scientist [63] Martyn made the "admittedly crude" estimate that such unintentional duplication of work cost the country 0.9% of the total R & D funds - £6,000,000 in 1962. He noted that the estimate was extremely conservative and that the true costs could be double that figure. In 1987, Martyn replicated his 1964 research [64] and found the situation very similar. His sample was smaller and the response rate not as good, but 26.8% of respondents reported late discovery: Martyn suggested that the costs of failure to discover information were probably of the same proportion of total R & D funds as they had been in 1964.

INFORMATION USE

The main strategy for determining what information has actually been used over the past fifty years has been citation analysis. One of the earliest such studies in the Journal was that by Earle & Vickery [65] which analyzed a sample of citations in the 1965 U.K. social science literature by subject, bibliographic form, country, language and date, and made comparisons with citations from science and technology literature. The authors compared use by demand on the National Lending Library (now the B.L. Document Supply Centre) with citations, and concluded that "neither citation nor loan demand is an adequate measure of literature use by a large community. Each is only an indicator, illuminating some aspects of use but with its own inherent bias." It was found that books (measured by citation) accounted for about 70% of social science citations, while periodicals accounted for about 30% and that social science citing differed from citations to scientific and technical literature in having a lower number of journal titles, and a greater degree of self-citation.

However, there have been other investigations into how information is used - most notably in policy research. For example, Caplan et al. [66], in examining the use of social science knowledge in policy decisions, found, not surprisingly, that the political implications of research findings determined how far they were put to use.

Other studies have reported a division of use into two kinds, which are described by Cole [67] as "direct" and "indirect". Cole analysed the records of information enquiries from scientists working in industry and found that 48% related directly to "operating information", or such topics as plant design and operation, while 22% was gathered as background information or "briefing".

Cole's results present an interesting contrast to those reported by Wilson [68] in his study of the use made of journal articles acquired through scanning a current awareness bulletin in the field of social work. Respondents were presented with copies of documents they had requested through the system and asked what use they had made of the information. In 58% of cases the document had been used to provide background information of one kind or another: adding to the person's general knowledge of the field, confirming or clarifying ideas, and allowing comparison with the ideas or practice of others. A further 12% were use for training or personal development. Thus, these two categories covered 70% of the cases - very different from the 22% reported by Cole. In 5% of cases the document gave practical guidance on how to do something, and a further 17% contributed directly to a specific task, such as writing a report, providing the basis for a project, or being quoted to support a point made in a meeting.

It is, of course, time-consuming to follow up document use, since interviewing is absolutely necessary to secure the necessary information from a user. However, the methodology used by Wilson and Streatfield demonstrates that it is possible. The results of studies of information use could be used, of course, to aid either questioning by intermediaries, or the design of more effective interfaces for human/computer interaction.

INFORMATION TRANSFER AND EXCHANGE

Although this topic has rarely been the subject of specific investigations, the transfer of information (i.e., the delivery to others of information believed to be of relevance to them by the initial recipient of that information) and its exchange, that is, its use in a bartering or trading arrangement, have been touched upon in a wide variety of studies.

Allen's study of information transfer and use in research laboratories [69] is a well known example of this kind of investigation, and his Ph.D. thesis, almost thirty years after its first publication, is a classic. In it, Allen developed his concept of the "gatekeeper" - the member of a team who acts as the channel for communication of ideas from outside, and from within. This was an extension into the research laboratory of the "two-step flow of communication" concept developed by Lazerfeld, Berelson & Gaudet in 1948 [70]. Among other things, Allen showed that the communication behaviour of technologists differed from that of scientists. The latter made greater use of formal communication media (scientific journals, reviews, advanced textbooks) than of informal, interpersonal communication, while the situation was reversed for the technologists.

In recent years, electronic communication has become common in many fields of specialization with the growth of national networks, such as JANET in the U.K., and international networks such as INTERNET (originally based on the academic network in the U.S.A.) In an update to the Royal Society's 1982 review of the scientific communication system, Meadows & Buckle [71] note that:

"The most obvious changes have occurred in informal communication between scientists, reflecting the rapid expansion of electronic networking in recent years... At the same time, the formal communication system has posed an increasing number of problems (e.g., information overload, rapid increases in costs)."

There are, of course, network developments that will formalize communication, such as the creation of electronic journals with editorial boards and peer review, and it seems likely that these trends will intensify as the networks put more capacity in place and as tools for navigating the network and extracting the information needed become more user-friendly.

One investigation (Schrader [72]) dealt with this area from a business perspective. This was a study of informal technology transfer between companies. Schrader noted that information exchange between competing companies is usually regarded as disadvantageous and that it is usually described as "information leakage". Shrader, however, prefers the term "information trading" noting that it can create economic benefits for a company and that, "It is advantageous if employees are aware of when to exchange and when to hide information and if they act accordingly." And, further:

"Information trading creates incentives to innovate. Internally generated technical knowledge is used not only within a firm, but also bartered for further knowledge -- as long as the benefits outweigh the cost."

USER SATISFACTION

Although many different kinds of studies have paid passing attention to how satisfied users are with the information they receive, there have been few studies that have dealt specifically with satisfaction. Most of what exists relates to the evaluation of information retrieval systems, including on-line information services - an area which, in general, I have excluded from this review. However, the principles are the same, whether the user is satisfied (or not) by the output of an information retrieval system, or by browsing in the library.

Fortunately, a review of this area [73] has appeared recently in another journal: in this, Applegate sets out three models of user satisfaction and relates the literature to these three models. They are, first, the material satisfaction model, where the performance of the information "product" in terms of relevance, pertinence, recall, and precision, is believed to lead to "use", which is a measure of "material satisfaction". Secondly, the single path, emotional satisfaction model, whereby material satisfaction (derived from product performance) is seen to lead to "emotional satisfaction", which, in turn leads either to repeat use and/or complaint about lack of satisfaction. Finally, the multiple path, emotional satisfaction model, the most complex of the three, in which emotional satisfaction depends upon material satisfaction, the "product setting" (i.e., price, inter-personal skills of the intermediary, etc.], and "disconfirmation" - the extent to which the user's expectations (derived from innate factors such as demographic and psychological characteristics, and acquired characteristics such as information about the product) are confirmed or disconfirmed. Again, emotional satisfaction leads to behaviour.

This is the most comprehensive account I have seen of models of user satisfaction and, clearly, offers many ideas for further research in this area.

METHODOLOGICAL PERSPECTIVES

Although I have taken a pragmatic approach to this survey of the past fifty years, it would be wrong to give the impression that there has been no progress in theoretical terms. Although it would be true to say that throughout most of the period the orientation of investigators was very practical - in that they were seeking guidelines for the development of information systems and services - it is the case that attempts to provide a theoretical framework, or, at least, to think in terms of causation, did occur from time to time. For example, the study by Mote, referred to earlier [74] an attempt was made to link information use to characteristics of the discipline of the user. Again, Oppenheim's study of student library use [75] sought to find distinctions according to sex, year of study and the nature of the course, and Stuart also identified male/female differences [76] in his study of reading in three London boroughs.

It was the reviews in the Annual Review of Information Science and Technology, however, that began to impose conceptual frameworks on the work that had been done. In the first of these, [77] Menzel evolved a typology of user studies and also singled out for comment two "methodological improvements": solution development records and paired research teams, both of which had been used by Allen [78] at M.I.T. Menzel's typology was not used again, but, two years later, Paisley [79] evolved a model of "the scientist within systems (within systems)", a set of ten "almost-concentric circles", which was slightly modified the following year by Allen [80], and both of these have been referred to frequently since. However, the majority of studies that reviewers have had available to them over the years have been (as noted earlier) studies of the use of systems - whether library systems, information systems, or the total system of scientific and technical communication. As a number of recent authors have noted, [81] the movement away from system-centred studies to person-centred studies did not begin until the 1980s, and the various models for organizing the literature were, consequently, system-oriented.

The move towards person-centred studies appears to have been initiated more or less independently by a number of researchers: Belkin [82], Dervin [83], and Wilson [84] are those most commonly credited [85] with initiating the move. These researchers are generally held to approach the problem of information needs and use from a cognitive perspective, but it is important to note a distinction between those who consider the cognitive state of the information user, and those who consider his or her cognitive style. The former is variable, dependent upon the context of the situation of information-seeking and information use, whereas the latter is held to be an innate characteristic of the individual.

This can be exemplified by reference to Belkin's work, which was designed to lead to the development of more effective information retrieval systems, and postulated an "anomalous state of knowledge" in the user, which the system must seek to match, if retrieval was to be effective. Thus, an anomalous state of knowledge would depend, not upon the cognitive style of the individual, but upon the circumstances that led to a gap in his or her understanding of a situation. Ford, on the other hand, although defining information need in much the same way as Belkin:

"The approach I have taken to describe an 'information need' is an awareness of a state of 'not knowing' - or some conceptual incongruity in which the learner's cognitive structure is not adequate to the task." [86]

is more interested in the extent to which different, innate cognitive styles will lead to different strategies of information-seeking.

The distinction is an important one, particularly as the cognitive styles perspective has been employed to a significant extent in research into the design of management information systems (MIS) and decision-support systems (DSS). Huber reviewed this literature some ten years ago and concluded that:

"(1) the study of cognitive style as a basis for deriving operational guidelines for MIS and DSS designs has not been fruitful and (2) such study is likely not to prove fruitful." [87]

Huber's case for his conclusions is closely argued and worth examination by information scientists. However, he does see a role for the concept of cognitive style in relation to MIS/DSS, noting that, if better instruments for the assessment of cognitive style were forthcoming, they could be used to identify the "natural propensities" of system users, and so aid training in the use of appropriate routines for decision-making.

This suggestion that cognitive style instruments could aid training has recently been the focus of investigation by a team in the Department of Information Studies, University of Sheffield . The team concluded:

"Learning and cognitive styles have been demonstrated to be a significant component of individual behaviour within the hypertext environment. This component is not, however, rigid and inflexible and does not necessarily enforce a particular style of usage upon a particular individual. Providing a variety of tools optimised for particular preferred modes of usage creates a rough equality of overall task-related performance, and allows the user to evolve an appropriate strategy for effective performance." [88]

Dervin approaches the problem of information-seeking behaviour as a communications researcher, rather than as an information scientist, and adopts a "sense-making" approach, which she defines as:

"...a label for a coherent set of concepts and methods used in a... programmatic effort to study how people construct sense of their worlds and, in particular, how they construct information needs and uses for information in the process of sense-making... In the most general sense, sense-making... is defined as behavior, both internal (i.e., cognitive) and external (i.e., procedural) which allows the individual to construct and design his/her movement through time-space. Sense-making behavior, thus, is communicating behavior." [89]

The central concepts of the sense-making approach are those of situations, gaps, and uses or helps. To this author's mind, the approach is more phenomenological than simply cognitive. Situation refers to the position in space and time where sense is being made; gaps are the points at which questions need to be asked to resolve barriers to understanding or to make progress in making sense; and uses/helps are the ways in which answers to the questions are put to use by the sense-maker. Dervin's aim is to apply these concepts as a general model, applicable in all communicative situations.

Although Wilson has been identified with the cognitive approach, his view of information needs and information-seeking behaviour is phenomenological in character, derived from the work of Alfred Schutz. From the phenomenological perspective, individuals construct their own social "world" from the world of appearances around them (see Schutz's Phenomenology of the Social World [90]). However, they can only do so with the help of others who have already placed a cognitive structure on the world of appearances. Thus, all of the devices we create to organize the cognitive structures of the world (libraries, retrieval systems, encyclopaedias, etc.) are socially constructed and help the individual to construct his or her own "meanings". We can see information needs, therefore, as derived from the individual's attempts to make sense of the world (as Dervin), and information-seeking behaviour as almost always frustrated in some degree by the division between the meanings embedded in information systems and the highly personal meaning of the information-seeker's problem.

Clearly, where a community of meanings has a high degree of consensus for a group (say, particle physics), it is likely that the search for information relative to problems can be tested against information resources that have been organized around the same consensus. Where the community of meanings is weaker, information systems are more likely to fail the user. Schutz describes the individual as needing to consult his "stock of knowledge on hand" and Wagner's [91] comment on this is illuminating for the information scientist:

"He cannot interpret his experiences and observations, he cannot define the situation in which he finds himself, and he cannot make any plans for even the next minutes without consulting his own stock of knowledge."

With Schutz in mind, Wilson expressed the view that:

• " our concern is with uncovering the facts of the everyday life of the people being investigated;
• by uncovering those facts we aim to understand the needs that exist which press the individual towards information-seeking behaviour;
• by better understanding of those needs we are able better to understand what meaning information has in the everyday life of people; and
• by all of the foregoing we should have a better understanding of the use and be able to design more effective information systems." [92]

In his 1981 paper Wilson presented a model of the information-seeking process (Figure 2) which has been amended from the original version to show how the work of another researcher, Ellis, [93] can be incorporated into the overall model. This turns out to be quite easy, since Ellis is concerned solely with the information-seeking process, rather than with the analysis of the needs that motivate search behaviour. Ellis has recently extended his work from the social sciences (his model was based on the search behaviour of social science researchers) to physics and chemistry [94] and has found that the original model fits behaviour in these fields with very little modification. He now identifies the stages in information-seeking behaviour as: Starting, Chaining (following citation linkages), Browsing, Differentiating, Monitoring, Extracting, Verifying, and Ending.