The transfer of theories and models from information behaviour research into other disciplines

Introduction

As Wilson (2018) noted in an earlier paper on this subject, the

interest in information behaviour on the part of disciplines other than information science is not new: even in the decade of the 1960s, the seventeen papers with the term information needs in the title were assigned to nine subject categories and were published in journals representing six different disciplines (Introduction, para. 3).

At that time, however, research in information science was relatively undeveloped, and there was little probability of theories from that field being transferred into other disciplines.

Today, however, as Wilson (2018) demonstrated, the situation is very different: information behaviour research is a dynamic area of research within information science, world wide. Wilson's data showed that more papers on information behaviour had been published in fields outside information science in the past decade, than in any previous decade. This paper intends to take the study of this phenomenon a little further by studying the impact of the research of three highly-cited scholars in the field of information behaviour: Carol Kuhlthau (USA), Reijo Savolainen (Finland), and Thomas Wilson (UK). These scholar's featured in Wilson's paper and Dervin was also chosen by Wilson, but her home discipline is communication studies, not information science, which led to confusion in categorising the citations, since her exports to other fields were not from information science; indeed she is an excellent example of research in communication studies being transferred into information science. Her work, therefore, has been excluded from this study.

This research attempts to answer the following research questions:

RQ1 – What style of citation is employed by researchers publishing in journals in fields other than information science, when citing the work of information behaviour scholars?

RQ2 – To what extent are citations to information behaviour in fields other than information science the result of researchers importing ideas from information behaviour research?

Related research

Interest in the diffusion of research is long-standing; for example Xhignesse and Osgood (1967) noted that their own research was based on earlier, unpublished work by Boll (1952). Xhigness and Osgood explored twenty-one journals in the field of psychology, examining the rates of inward and outward citations among the set, to determine the nature of the network created by these journals. This was a study of interactions within a single discipline and studies of this kind continue to be produced. Discovering very early studies of information transfer between disciplines, proves more difficult, and it seems that this had to await the emergence of information science as a research field.

Stigler et al. (1995), investigated the extent to which such trade took place within the network of economics journals, finding that journals with a focus on economic theory were exporters to applied economics journals. A set of forty-two economics journals was studied by Pieters and Baumgartner (2002) to determine the structure of the field in terms of the trade in citations among them. They also examined the exchange of citations between economics and a number of other disciplines in the social sciences and business, concluding:

Economics emerges as the primary source of knowledge in this network of social science and business disciplines. Six of the nine sister disciplines rely for twenty per cent or more of their interdisciplinary knowledge on economics. Yet, economics builds only slightly on knowledge from its sister disciplines. Merely ten per cent of the citations that five first-tier economics journals made between 1995 and 1997 are interdisciplinary and the majority of these citations go to finance. (p. 504)

Another single discipline study was carried out in the field of marketing research by Baumgartner and Pieters (2003). This was intended to reveal the structure of the field, as determined by citations, and did not consider the citation trading with other disciplines.

Earlier, Stigler (1994), had studied the balance of trade among statistics journals, concluding that, ‘those aspects of intellectual influence that are represented well by citations tend to flow on balance from more abstract work to more applied work'. (p. 107)

A study of more than one discipline was carried out by Locket and McWilliams (2005): they chose to explore the interactions among journals in economics, management, psychology, and sociology. Six journals were chosen for each of these disciplines and the results showed that journals in, ‘economics and psychology are the largest net exporters of knowledge, while the majority of management and sociology journals were net importers of knowledge'. (p. 144).

A multi-disciplinary study was carried out in Japan by Urata (1990), using nine journals from different disciplines: anthropology, economics, education, history, linguistics, philosophy, psychology, sociology, and statistics. He found that the citation flows resulted in a hierarchy of journals, based on an index of independence. The most independent disciplines, which imported few citations from other fields, were philosophy, history, linguistics, and psychology; and the least independent were education and sociology. Education, for example, imported from philosophy, psychology, anthropology, and linguistics, and had no exports.

In information science, a study by Cronin and Davenport (1989) considered the citation profiles of six professors of librarianship or information science in the UK, and, although this did not include a study of the import and export of citations between disciplines, the authors noted in their conclusion, ‘We have the feeling that the library and information field is a net importer rather than exporter of ideas' (p. 20). Cronin, with Pearson, (1990) picked up this idea and investigated the extent to which the ideas of six grandees of information science, Bertram Brookes, Cyril Cleverdon, Robert Fairthorne, Jason Farradane, Maurice Line, and Brian Vickery, were exported to other disciplines. They found that, ‘During the period in question the grandees amassed a total of 1,650 citations to their work, of which 156 (9.5%) were exports'. (p. 383).

This study was followed by another, by Cronin and Meho (2008), in which the import/export metaphor was again employed. The authors report:

Of the 54,181 papers that cited the 275 IS [information science] periodicals included in the study, 28,363 (52%) came from outside the field. Exports from IS to other fields have increased significantly over time... the number of non-IS papers citing the IS literature has risen from 3,982 for the period 1977–1986 to 18,079 for the period 1997–2006, an increase of 354%. (p. 560)

and they offer an explanation for the increase in ‘foreign' citations:

We believe that the striking increase in foreign citations to the literature of IS can be explained in large measure by two developments: the growth of research domains influenced materially by advances in information technology and Internet applications (e.g., computer science, business and management, health/medical sciences, and engineering), and the expansion of ISI's [now Web of Science] coverage of domains relative to information studies. (p. 560)

Meho and Sugimoto (2009) found that an assessment of the impact of information science in other disciplines required the use of both Web of Science and Scopus, and

this study found that of the 7,497 papers that have cited the study sample, 73% (5,479) came from outside the field (this difference is largely due to differences in literature coverage between Scopus and Web of Science). These non-IS citing papers were found in 1,703 journals and conference proceedings; the remaining 2,018 citing papers came from 83 IS journals and conference proceedings. (p. 2503)

There has been an interest in classifying the types of citations found in research papers since at least the 1960s. For example, Lipetz (1965) produced a classification made up of four major categories and a total of twenty-nine sub-categories, which was rather too fine-grained for use in this research. Small (1982) identified a more manageable set of citation types as perfunctory, reviewed, negative, supported or affirmed, and applied. This typology was applied by Stremersch et al. (2015) in a study of citation in the field of marketing. They found that, ‘In our sample, 31.8% of all citations are perfunctory mentions and 52.9% are review citations. In other words, only 15% of all citations show irrefutable scientific merit from citing to cited article'. (p. 65).

Moravcsik and Murugesan (1975) adopted a different approach, developing a typology for exploring the literature of high-energy physics. Eight categories were devised, resulting from the questions:

Is the reference conceptual or operational? In other words, is the reference made in connection with a concept or theory that is used in the referring paper, or is it made in connection with a tool or physical technique used in the referring paper?
Is the reference organic or perfunctory? In other words, is the reference truly needed for the understanding of the referring paper..., or is it mainly an acknowledgment that some other work in the same general area has been performed?
Is the reference evolutionary or juxtapositional? In other words, is the referring paper built on the foundations provided by the reference, or is it an alternative to it?
Is the reference confirmative or negational? In other words, is it claimed by the referring paper that the reference is correct, or is its correctness disputed?

There is some similarity to Small's categorisation here; for example, perfunctory, confirmative and negational are equivalent to Small's perfunctory, affirmation, and negative. The categories conceptual, operational, organic, evolutionary and juxtapositional, however, have no direct equivalent in Small's typology, although it is likely that his affirmation category would relate to Moravcsik and Murugesan's organic and evolutionary categories.

In the same year, Chubin and Moitra (1975) took issue with the claim by Moravcsik and Murgesan, that the large proportion of perfunctory citations found in their analysis of the high-energy physics papers cast doubt on the use of citations as a quality measure (Moravcsik and Murgesan, 1975, p. 91). They noted that the categories used by Moravcsik and Murgesan were not mutually exclusive (for example, a citation might be both conceptual and organic, or evolutionary and confirmative), and they evolved a typology with six, rather than eight, categories: four categories were affirmative and, within that, either essential or supplementary; two categories were negational, either partially or totally. The typology can be represented (with the six categories in italic) as:

• Affirmative
  • Essential
    • Basic
    • Subsidiary
  • Supplementary
    • Additional information
    • Perfunctory
• Negational
  • Partial
  • Total

More recently, Zhao et al. (2017) referred to Small but adopted the categories evolved by Tabatabaei (2013). Tabatabaei's categories are similar to Small's (1982): they are, applied, contrastive, supportive, reviewed, and perfunctory. Of these, contrastive and supportive appear to differ from Small, but contrastive would include negative, and supportive would include affirmation, so there is some similarity.

Another recent development is the arrival on the scene of scite a search engine designed to identify the types of citations in a designated document. Scite is described as,

a Brooklyn-based startup that helps researchers better discover and evaluate scientific articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contradicting evidence. (scite, 2020)

However, scite uses only three categories: supporting, mentioning, and disputing, with no further analysis. It also takes statements in the citing paper as separate instances of the categories; for example, a search for Wilson (1999) resulted in ten supporting references, 593 mentioning, and three disputing, but two of the three disputing references were in the same document. Also, the authors of the paper did not dispute the theoretical point made by Wilson but simply found a low correlation between self-efficacy and information seeking in the particular context they studied. Clearly, automation still has a long way to go before some of the work noted above can be replicated by artificial intelligence.

Given the range of typologies available, and given that no single typology dominated in the literature, it was decided to use Small's (1982) categorisation, which had also been used by Stremersch et al. (2015), and which appeared to share many elements with other typologies. It was also relatively straightforward to apply, requiring no deep analysis of the referring documents.

Data collection

Citing documents for the three authors were harvested by selecting from the relevant Google Scholar profiles for Savolainen and Wilson the most frequently cited single-authored paper (Savolainen, 1995; Wilson, 1999). Kuhlthau does not have a Scholar Profile and the most frequently cited paper was determined by carrying out a Scholar search and locating the relevant paper (Kuhlthau, 1991). Kuhlthau's book Seeking meaning has more citations than the selected paper, but, as papers had been selected for Savolainen and Wilson, it seemed appropriate to use the same document type for Kuhlthau.

The items in Google Scholar output show the number of citations received by a paper and it was a straightforward matter to click on the number and retrieve the list of citing items.

Working with the outputs from Google Scholar profiles presents some difficulties: the output is initially organized by ‘relevance' and one can have it presented arranged by date, which, in this case means most recent date first. My initial intention was to select a random sample of fifty pages (providing 500 citations) from the whole set, but Google Scholar only presents the first 100 pages of output (1000 citations) and, thus, in the case of the 2,834 citations of Inside the search process (Kuhlthau, 1991) it would not have been possible to select pages between number 100 and 284, meaning that more than half of the citations could not have been selected. In all three cases, therefore, it was decided to use the first 500 citations.

Nor is Google Scholar totally accurate: I found a small number of ‘false drops', that is, papers that did not include any citation to the author concerned. In the case of Wilson, for example, the list included The rusty tools of peace, by B. Urquhart in the World Policy Journal for Winter 2000/2001, which looked rather an unlikely citation. Examination of the actual item revealed no reference to his paper and, indeed, no notes or references at all. Curiously, the same item appeared in the list for Kuhlthau!

Even more curious was the case of The tribute of the allies, by A. French, which was published in 1972 in Historia: Zeitschrift für Alte Geschichte, and which, according to Google, had cited Wilson's paper. Of course, the date was the give-away, as that paper was not published until 1999 and, unless the author had been gifted some variety of time travel, it could not have been cited in 1972. In any event, French's paper was on the contributions made to the war chest of the Delian league by the allies of Athens, with no mention of information-seeking, either then or now, so quite how it got into the first fifty pages of the output on the basis of ‘relevance' cannot even be guessed at.

It proved fairly straightforward to employ the categorisation of citation types (Small 1982; Stremersch et al. 2015): only one category was assigned to a paper, although, if ideas were applied, the cited paper generally appeared in the review. Occasional difficulties were experienced in determining whether ideas had actually been applied, and the borderline between perfunctory and review was sometimes a little fuzzy, generally, however, a perfunctory citation was the only citation of the paper, while a review often led to multiple citations.

Data analysis

The citing documents were first analysed to determine the discipline served by the journal in which the citation appeared. No reference was made to existing classifications of disciplines, but decisions were made by the author on the basis of the journal title and the field indicated by the journal contents. This is clearly subjective, and decisions had to be made for example, whether to include a journal that dealt with technology use in education in computer science or in education. In that case it was decided to allocate the journal to the field served by computer science, since it was clear that the intended audience was in the field of education. Health informatics, however, was joined with information systems, since it appeared that the intended audience consisted of informatics staff working in the health field, rather than the medical, nursing or health care staff.

That initial analysis resulted in a list of forty disciplines, which was reduced by combining some closely related fields, such as the range of disciplines in medicine and health care, and communication studies with media studies. This reduced the list of disciplines to thirty-seven. For further analysis only the top four disciplines were chosen, namely, computer science, health and nursing, information systems, and education.

Further analysis of the citations was guided by Small’s (1982) categorisation of citation types as relating to: application, affirmation, negation, review, and perfunctory, which was also used by Stremersch et al. (2015). The concepts related to the terms are probably intuitively understandable: application implies that the cited source, or some element of it, has been applied in the research associated with the citing document; affirmation, that the theory, model, or hypotheses proposed in the cited source have been confirmed by the citing paper; negation is the opposite of affirmation, i.e., the proposals in the cited paper have been rejected by the citing research; review means simply that the cited research has been covered in the citing document’s review of related research; and perfunctory that the citation is merely token and that the cited paper plays no particularly constructive role in the citing article. Occasional difficulties were experienced in determining whether ideas had actually been applied, and the borderline between perfunctory and review was sometimes a little fuzzy, generally, however, a perfunctory citation was the only citation of the paper, while a review often led to multiple citations.

The citations were also analysed according to whether the author(s) of a citing paper were from within the receiving discipline and importing the cited items, or from the information science sector and exporting the ideas into the receiving discipline (Cronin and Pearson, 1990).

Results

The combined database of citing documents contained a total of 267 items, selected from 1,500, i.e., 17.8% of the citing documents had been published in sources representative of fields outside of information science, the remainder being citations in information science sources. Of these 267 items, twenty-one were duplicates, that is, citing two or three of the cited authors: thus, thirteen cited Kuhlthau and Wilson, three cited Savolainen and Wilson, one cited Kuhlthau and Savolainen, and two cited all three, giving a total of 246 unique items.

The distribution over the first four disciplines of citations to the three selected papers is shown in Table 1:

A chi-squared test on these data revealed that the association between researcher and citing discipline was significant (𝜒² = 21.7286, p = 0.001356). An examination of the data suggests that this association may result from a) the proportion of Kuhlthau's and Savolainen's citations in computer science; b) the proportion of Wilson's citations in health-related disciplines; and c) the smallest proportion of Savolainen's citations being in education. We can also see that a considerably higher proportion of Savolainen's citations is distributed over disciplines other than the four selected for further examination.

In the next sections we discuss the distribution and use of the cited material in the four selected disciplines, beginning with computer science.

Use of information behaviour research in computer science

The papers by all three information behaviour researchers were cited in computer science sources, as shown in Table 1 above. The citations in computer science sources were analysed further, by citation type (Small, 1982; Stremersch et al., 2015; Baumgartner and Pieters, 2003), with the result shown in Table 2:

It will be noted that no occurrences of affirmation or negation were identified. The analysis by importer and exporter revealed that a total of 14 (18%) papers were authored by exporters and 66 (82%) by importers. The distribution by researcher is: Kuhlthau 7 (21%) exporters 26 (79%) importers; Savolainen 5 (29%) exporters 12 (71%) importers and Wilson 2 (7%) exporters 28 (93%) importers. Wilson's data, in this particular respect, appears to be very different from the other two researchers, but the distribution by researcher is not statistically significant (𝜒² = 3.6248, p = 0.456506).

The citation type of most interest in these data is application, since it implies that the ideas from information behaviour research have found application in another field and the ten papers allocated to this category were investigated further to identify the nature of the applications.

In his doctoral thesis White (2004) pays particular attention to Kuhlthau's model of the information search process (Kuhlthau's work is cited 28 times in the dissertation) and notes that the model formed the basis for task selection in his own research. He goes on to say: ‘I do not choose six task categories that correspond with the six stages in the ISP, but instead to the three types of searcher interaction that the model predicts; background seeking, relevant seeking and relevant and focused seeking' [the author's emphasis] (p. 151). Another clear example of application is found in the paper by Bozzon et al. (2013) in which Kuhlthau's model was used as the basis for designing a prototype search system. A very specific statement of the use of Kuhlthau's model is given by Mills et al. (2014): ‘The model adopted for this research is based, in part, on the Information Search Process model established by Kuhlthau (1991)'.

Savolainen's (1995) model of everyday life information seeking is employed by Hsu and Walter (2015), who note that, ‘our model uses CMIS [comprehensive model of information seeking] as the framework and integrates it with constructs in TAM [technology acceptance model], ELIS [everyday life information seeking], and consumer behavior literatures that are relevant to local information seeking on the web‘ (p. 265).

An application is mentioned by D'Ambra and Wilson (2004), who first note: ‘This model by Wilson (1999)… supports one of the fundamental propositions of this paper: that the goal of uncertainty reduction is fundamental to the use of any information resource within a problem resolution context' (p. 295), and then note that the scale for measuring uncertainty was also derived from Wilson's work (p. 300). O'Brien and Buckley (2005) consider not only Wilson's models but also those of Kuhlthau (1988), Marchionini (1995), and Ellis et al. (1993), ultimately proposing, ‘a five-stage model, which has at its foundation, the stages associated with afore-mentioned models of information seeking behaviour‘ (p. 130). In their study of the information needs of medical doctors, Kourouthanassis et al. (2015) also combine models; in this case Wilson's (1999) and Bhattacherjee's (2001) adaptation of expectation-confirmation theory.

These examples of application suggest that researchers in computer science are drawing upon theories and models from information behaviour research, but often combining these with theories and models from other fields, or, as in the case of O'Brien and Buckly (2005), combining different models of information behaviour.

Use of information behaviour research in health-related disciplines

In the health-related disciplines, De Rouck and Leys (2011) affirm the relevance of Wilson's models at several points in their paper, noting particularly that they, ‘encompasses both active and passive information behaviour. The distinction between active and passive information acquisition is useful to classify information sources and channels and valuable for distinguishing content of the different information sources/channels' (p. 71). Pluye et al. (2007) affirm the usefulness of Wilson's models, noting: ‘in line with Wilson… the application of information might in turn influence the acquisition, and lead professionals to reiterate their search for information' (p.45)

Forsetlund and Bjørndal (2002) adapt Kuhlthau's interpretation of the idea of the bibliographic paradigm to argue that the ‘cognitive gap between the system and the user's natural processes of information use' results in the information sources being ‘not easily accessible, understandable or easily applicable' (p. 16). Consequently, they treat this, in their research, as a characteristic of information sources.

As Table 3 shows, in this area of research, there were a relatively large number of applications of Wilson's theory. It was thought desirable, therefore, to study these in some detail, to determine exactly what kind of application was involved. Two broad types of application emerge: the first takes an entire model as the basis for the research, while the second uses some part of the model or theory to aid the development of research instruments or to provide hypotheses for testing.

Examples of the former include Cramer's (2018) application of Wilson's model: ‘The study draws from a large sample of fathers across the health continuum and grounds its findings in a comprehensive model of health information behaviour (Wilson, 1997)' (p. 321). Cramer also notes that Shieh et al. (2009) have ‘developed scales measuring pregnancy health information seeking, pregnancy health information needs, and pregnancy health information barriers that connect more consistently to elements of Wilson's (1997) theory' (p. 323). Shieh et al. comment that,

‘Findings of this study support the proposition in Wilson's model of information behavior that information needs and information barriers predict the degree of information‐seeking. Our study in particular indicates that low‐income pregnant women with a higher level of need for information and a lower level of barriers to seeking information were more likely to engage in more seeking for pregnancy health information than low‐income pregnant women with a lower level of need for information and a higher level of barriers' (Shieh et al., 2009, p. 369).

Cao et al. (2016) based their research on Wilson's model of 1999, commenting: ‘Guided by Wilson's… second model of information behavior, it intends to identify the relationships among source characteristics, activating mechanisms, and OHISB [online health information-seeking behavior]' (p. 1106), and the research hypotheses are based, to a significant extent, on this model. All of the hypotheses were supported by the research; for example, ‘hypothesis 3b posited that higher levels of Internet self-efficacy predict higher levels of OHISB' and the research found a direct relationship between self-efficacy and the level of online information seeking. Thus, the research affirms the theoretical propositions underlying the model.

In his PhD thesis, Rhebergen (2012) notes that, ‘In total, 14 questions [were] asked about factors that may motivate information seeking. These factors were based on Wilson's general theory of information-seeking behaviour‘ (p. 27), and the same relationship to Wilson's model is found in an associated paper by Rhebergen et al. (2012).

Finally, Halkett and colleagues employ the general model in research reported in three papers, Halkett and Kristjanson (2007), Halkett et al. (2009), and Halkett et al. (2010). In addition to the application of Wilson's general model, some authors applied the problem-solving model, which is described in the same paper (Wilson, 1999). For example, Ekberg et al. (2010) employed the model in their design of a prototype information system: ‘In the following analysis, Wilson's model of information seeking behaviour was applied to the design of a prototype information system used in a community-based management of diabetes Type 1' (p.110); and, ‘A guideline based on information behaviour theory for design of Web 2.0 systems supporting chronic disease communities was developed, and found to help maintain a holistic view of the system' (p. 112).

Instances of partial use include that of Stonbraker et al. (2017), who note:

The first dependent variable was information seeking which, guided by Wilson's model, was dichotomized as active or passive, where active seekers are more engaged in the pursuit of heath information and passive seekers may obtain information that is relevant to them while engaging in another behavior or without looking for it' (p. 1592).

Zhang and Zhou (2019) take the affective element of Wilson's model and propose fear as a major motivator in one of their hypotheses:

H3: Fear aroused by risk messages is positively related to ITC [intention to click] the messages (p. 1361).

Use of information behaviour research in information systems research

The instance of affirmation of Kuhlthau's work occurs in Stavri (2001): ‘Uncertainty, (an affective component identified by Kuhlthau… appears to play a role in PHIS [personal health information seeking] as well as general information-seeking' (p. 1486). An example of affirmation of Wilson's work appears in Ivaturi et al. (2017): ‘Users in this mode [non-directed monitoring] are usually not aware of the need for information until they encounter it. This behavior corresponds to Wilson's… passive attention…' (p. 141). Another example from Kotzyba et al. (2018) states: ‘The activity of verifying is an important part of information seeking behavior and is e.g., described in Wilson's aggregation of Kuhlthau's and Ellis's Models' (p. 2607).

The instances of applications in the information systems field are half those in computer science, and many fewer than in the health-related field. Examples of application include Järvelin et al., (2015), who make extensive use of Kuhlthau's information search process model to explore ‘task-based information interaction'. Papadopoulou et al. (2013) employ Savolainen's model of everyday-life information seeking as part of the basis for their model of online, dyadic relationships and the associated information behaviour.

In common with the health-related fields, the largest category of citations is the review category. This is understandable, given that researchers are obliged to review research related to their own investigation and, inevitably, whatever their home discipline, they are going to discover the related research in information behaviour.

Use of information behaviour research in education

Given that Savolainen's paper deals with everyday-life information seeking, it is not surprising that there are few citations to his work in the education sector. What may be thought more surprising, however, is the relatively small number of citations to Kuhlthau's work. This can be explained by the fact that Kuhlthau has invested considerable effort in taking her work into the education sector through workshops and consultancy and her impact is likely to be in actual practice, rather than in research. There is also the fact that other work by Kuhlthau is highly cited; her book, Seeking meaning (2004), has more than 3,000 citations according to Google Scholar, and a second book, Guided inquiry (2015), has more than 800 citations. Consequently, using these additional sources could reveal a rather different picture.

The affirmation of Savolainen's everyday life theory is found in Branch (2003): ‘The findings of the study are consistent with Savolainen's... assertion that "everyday life information needs proved to be quite heterogeneous"...' (p. 12). That of Kuhlthau's work is provided by Hess (1999): ‘The findings of this pilot study do support prior research (Barry, 1995, 1997; Kuhlthau, 1991) and provided some interesting insights into information overload as a core dimension of cognition held by a graduate student during Web-based searching' (p. 12). Rad et al. (2009) affirm Wilson's model: ‘The findings are also in agreement with Wilson's general model of information behaviour of managers' (p. 2514).

Marshall et al. (2006), apply Kuhlthau's information search process model in the design of an experimental online learning platform, and in the evaluation of similar, commercial platforms. Repanovici (2008) applies several of Wilson's models in research linking information behaviour and information literacy to information technology.

The number of cases in education is rather too small to reach any general conclusion, but, as with the health field, some researchers are applying some part of a model or theory, while others are adopting a model in its entirety as the basis for their research.

Discussion

The three papers discussed here were different in scope and intention, but all had serious use in disciplines other than their home discipline of information science. It is not surprising, in the light of other investigations in other disciplines, that the use is often perfunctory, where a merely token reference is made to a paper, for example: ‘Based on ideas from cognitive psychology, the model identifies interactions between different actors during information search processes, while also integrating system design issues (Ingwersen 1996; Wilson 1999)', where there is no further reference to Wilson and no elaboration of how his work relates to system design. Another example relates to Kuhlthau: ‘Kuhlthau suggests an information search process (ISP) model with six stages to describe the behaviour of seekers', with no further reference to use of the model (Pang et al., 2015, p. 47).

As Table 6 shows, 36 per cent of the citations reviewed were perfunctory. A chi-squared test reveals that the distribution of citation types is statistically significant at the p < 0.05 level (𝜒² = 28.9193, p = 0.000669). The biggest single contributor to the difference is the number of citations identified as applications in the health-related journals.

These inter-disciplinary citation types are very similar to the intra-disciplinary citations studied by Stremersch et al. (2015). Table 7 shows the comparison with the results of this research: the proportions from Stremersch et al. and Zhao et al. have been rounded to make the similarities more obvious. The comparison with Zhao et al. (2017), who used slightly different categories, shows less similarlity, but the figures for application and review are rather similar. It must be borne in mind, however, that Zhao et al. used a slightly different categorisation, and fitting them into Small's categories, as here, may not result in an accurate reflection of the data.

Comparing the results of this work with the types of citation found in the other disciplines, it is evident that the health-related fields make most use of ideas found in information behaviour research. Whether this would continue to be the case if more cited papers by a greater variety of authors were studied, is a matter for further research. It is also clear that application may mean that a model or theory in its entirety guided the research, or that some element was extracted to guide the development of, for example, a questionnaire, interview schedulue, or measurement scale. This is supported by Chubin and Moitra (1975) who identified essential, affirmative citations as either basic, that is ‘central to the reported research' (p. 426) or subsidiary, involving a specific method or tool. The fact that all three cited papers are theoretical in character corresponds with Stigler's (1994) comment that the balance of trade favours the transfer of theory to applied research.

A study by González-Teruel and Pérez-Pulido (in press) appeared while this paper was in preparation. The authors studied the impact of Savolainen's everyday-life information seeking model in fields other than information science, finding, as in this study that the major impact was in computer science. However, the study used a different categorisation of citation types, and a different set of subject categories, making further comparison with the results here impossible.

Considering the second research question, the use of the notion of importers and exporters is problematic, simply because authors from within the field of information science are often carrying out research which has strong associations with another field, as in the case of information retrieval and computer science, or information seeking for learning and education. In the preceding analysis, therefore, this distinction must be seen as approximate at best.

Bearing this in mind, however, the distribution of exporters and importers by discipline is shown in Table 8:

Eighty-four per cent of the citing documents were authored by importers, i.e., researchers in the citing discipline importing ideas from information behaviour research. However, there was no statistically significant difference between the disciplines at the p < 0.05 level (𝜒2 = 4.3855, p = 0.222734). The fact that 83% of the citations were made by researchers in other disciplines importing the theories and model of information behaviour suggests that in this field at least, information science research is making an impact in other disciplines.

The same analysis was performed for distribution by researcher (Table 9) and this was statistically significant at p < 0.05 (𝜒2 = 13.3782, p = 0.001244). The significant difference is the result of Savolainen having about twice as many exporters as would occur by chance, and Wilson having about half as many.

Limitations

The most obvious limitation of this research is its restriction to exploring the impact of just three highly-cited papers in disciplines other than information science, and then the examination of only four such disciplines. It would be desirable, of course, to examine the highly-cited work of other researchers in the field of information behaviour, and to explore their impact in a wider range of disciplines. The work, however, cannot be automated: the allocation of citations to disciplines and the categorisation of citations requires human judgement and many hours of work. In spite of the limitations, however, the work does reveal something of the influence of information behaviour research in disciplines outside of information science, as well as confirming that an interest in information behaviour is not restricted to a single discipline.

Conclusions

The three papers studied in this investigation (Kuhlthau, 1991; Savolainen, 1995; and Wilson, 1999) differ from each other in readily discernable ways. Kuhlthau's paper deals with the information search process, which is one element in Wilson's model, which covers information behaviour generally, while Savolainen's paper deals with the area of everyday-life information seeking, thereby restricting the potential areas of application.

This paper is not intended as a contribution to the literature on citation theory; the reader expecting such a contribution will be disappointed. In the terms used by Leydesdorff (1998), this paper uses citation analysis as explanans, that is, it seeks to explore the relationship between information behaviour research and other disciplines, rather than to explain the function of citing in the information behaviour literature. However, the question of how theories and practices from one field are able to permeate other disciplines, is a question for citation theory.

Thus, the notion of a boundary and its permeability may be of use in respect of what factors affect the ability to transfer information across disciplines. The boundary of a discipline is constructed by its members through the formal or informal mechanisms that determine membership. For example, to be a researcher in cosmology, one must have the necessary academic qualifications, and the subject implies that these qualifications will require significant knowledge of physics and mathematics. Cosmological knowledge would be difficult to transfer into another discipline, unless similar criteria for membership were met. The research objects constitute another boundary element: we can hypothesise that only disciplines that deal with similar research objects are likely to be able, usefully, to transfer theories and ideas. The more difficult boundaries are to cross, the less permeable they will be.

Thus, when we examine the relationship between, say, health-related research and information behaviour, the boundaries between the two disciplines are quite permeable: it is quite possible (as the review above demonstrates) for a nursing researcher to import information behaviour theories, and quite possible for an information behaviour researcher to investigate information seeking by nurses. The research objects are the same for both, i.e., the person, who may be a patient or a member of the medical staff. The methods of research are also the same, usually, interviewing or some form of survey, and the information behaviour theories do not require any specialised vocabularly, but can be expressed in everyday language; no abstruse mathematical formulae are required.

Further research, aided by the concept of disciplinary boundaries might cast new light on the processes of knowledge transfer between disciplines.

Another factor is the nature of the cited research: as Stigler (1994) noted, and as this research appears to confirm, theoretical papers have a higher probability of being imported into other disciplines than applied research within a discipline. We can take an example from sociology: Giddens's The constitution of society, his major work on structuration theory, according to Web of Science, has been cited in twenty-four Web of Science categories, including economics, history, geography, anthropology, religion, business, linguistics, philosophy, and medical informatics. Evidently, researchers in these fields find structuration theory useful.

Disciplines that provide tools and methods for research may also be significant exporters: the obvious example is statistics. Urata's (1990) research in Japan revealed that statistics interacted with economics, with more than 10% of citations from one to the other in both directions.

Having seen how the work of these three researchers is used in disciplines other than information science, it would be interesting to turn to the citations within information science to discover, for example, the extent to which researchers from other disciplines are reporting in information science journals, and the nature of the citations. This author has the suspicion that more work may be done in other disciplines to build on models and theories in information behaviour than is done in information science. Confirmation or refutation of that suspicion, however, would have to await further research.

Acknowledgements

My thanks to Professors Carol Kuhlthau, Reijo Savolainen and Elena Maceviciute for helpful comments on an early draft of this paper, and to the referees for their helpful suggestions.

About the author

T.D. Wilson is Professor Emeritus of the University of Sheffield. He holds a BSc degree in Economics and Sociology from the University of London and a Ph.D. from the University of Sheffield. He has received Honorary Doctorates from the Universities of Gothenburg, Sweden and Murcia, Spain. He was the recipient of the 2017 ASIST Award of Merit. He is the founder and Editor in Chief of Information Research. He can be contacted at wilsontd@gmail.com

How to cite this paper

Appendix 1: Citations to disciplines other than information science.