Analysing textual data ininternational marketing research
Rudolf R. SinkovicsThe University of Manchester, Manchester, UK
Elfriede PenzWirtschaftsuniversitat Wien, Vienna, Austria, and
Pervez N. GhauriThe University of Manchester, Manchester, UK
Abstract
Purpose – To provide guidance for the formalised analysis of qualitative data and observations, toraise awareness about systematic analysis and illustrate promising avenues for the application ofqualitative methodologies in international marketing research.
Design/methodology/approach – Conceptually, the nature of qualitative research, globalisationand its implications for the research landscape, text-data as a source for international research andequivalence issues in international qualitative research are discussed. The methodology sectionapplies these concepts and analysis challenges to a real-world example using N*Vivo software.
Findings – A 14-step analytic design is developed, introducing procedures of data analysis andinterpretation which help to formalise qualitative research of textual data.
Research limitations/implications – The use of software programs (e.g. N*Vivo) helps tosubstantiate the analysis and interpretation process of textual data.
Practical implications – Step-by-step guidance on performing qualitative analysis of textual dataand documenting findings.
Originality/value – The paper is valuable for researchers and practitioners looking for guidance inanalysing and interpreting textual data from interviews. Specific support is given for N*Vivo softwareand its application.
Keywords Computer software, Qualitative research, International marketing, Knowledge management
Paper type Technical paper
IntroductionConstantly changing business environments in the present age of globalisation arecreating new challenges for researchers as well as companies. As a result, establishedmethodologies have to be re-evaluated in view of newer and emerging research optionsand conditions. We witness technological innovations which revolutionise the way wedeal and interact with geographically dispersed organisations and consumers (Craigand Douglas, 2001). Researchers are experiencing methodological and conceptualdifficulties when dealing with inter- or cross-national research. The application ofestablished methodologies and practices cannot always be applied successfully in thechanging international research settings (McDonald, 1985).
The selection of appropriate research strategies in international marketing researchis usually determined by the level of existing information and knowledge in the field(Churchill, 1995). The major portion of the international business and marketingliterature is geared towards precise problems, well-defined in scale and scope, whichcan be easily investigated with rigorous scientific methods. In many situations
The Emerald Research Register for this journal is available at The current issue and full text archive of this journal is available at
www.emeraldinsight.com/researchregister www.emeraldinsight.com/1352-2752.htm
Analysingtextual data
9
Qualitative Market Research:An International Journal
Vol. 8 No. 1, 2005pp. 9-38
q Emerald Group Publishing Limited1352-2752
DOI 10.1108/13522750510575426
however, the available expertise is very limited and little prior research has beencarried out that points to the underlying problems and ambiguities. In internationalmarketing research and particularly research at the marketing and entrepreneurshipinterface, the use of open, creative and flexible research designs is required. A creativeorientation in the research process will help to grasp the non-linear, sometimes chaotic,development of small and medium-sized enterprises (SMEs). This will also allowcapturing multi-dimensional phenomena in the course of data collection and analysis.Young et al. (2003) highlight this issue while discussing international entrepreneurshipand the application of theories developed in the field of international business.Johanson and Vahlne (2003) present an experiential learning-commitment mechanismfocusing on business network relationships in this context. Generally, this will supportthe provision of a much broader picture on issues of analysis. Many scholars aresuggesting that the use of exploratory research and qualitative methodologies in thesecontexts is more suitable (Ghauri and Grønhaug, 2002). Some extend this view byarguing that qualitative methodologies can help to find “meaning behind thenumbers”, particularly when many data involves narrow details that blur a clear andholistic view of the context (Denzin and Lincoln, 1998; Ruyter and Scholl, 1998).
Comparative empirical research methodology has been a subject for discussion formore than 30 years (Cavusgil and Das, 1997; Knight et al., 2003; Sekaran, 1983). Asignificant amount of effort has been devoted to address issues such as researchdesign, sampling, instrumentation, data collection and analysis, reliability andvalidity, and general application of findings (Harris, 2000; Nasif et al., 1991;Padmanabhan and Cho, 1995; Parameswaran and Yaprak, 1987; Samiee and Jeong,1994). Coviello and Jones (2004) reviewed the field of international entrepreneurshipand concluded that there is a need for dynamic research design that integratespositivist with interpretivist methodologies. In spite of this, a debate on methodologicalconsiderations is in qualitative research and the analysis of textual data is scarce.
Problem and purposeMuch has been written about the importance of interpreting qualitative data and theliterature offers a rich repertoire of methods of qualitative inquiry (Bickman and Rog,1997; Denzin and Lincoln, 1994). However, two of the most important factorsconcerning the slow adoption of qualitative research methodologies for internationalresearch have been the “lack of replicability” and the painstaking efforts required tocoordinate international research teams. Nevertheless, qualitative data, and textualdata in particular, are an “attractive nuisance” (Miles, 1979) since they offer real andfull insights into phenomena. In view of this argument and the objective to movemarketing research further away from exclusively quantitative comparative work, weadvocate formalised qualitative procedures. Formalised procedures of gathering,analysing and interpreting qualitative data are also particularly important incollaborative international work involving multiple researchers.
This paper is concerned with ways in which qualitative methodologies can beexplored in international marketing and entrepreneurship research. This issue isparticularly relevant to the emergence of computer assisted qualitative data analysissoftware (CAQDAS) which helps both researchers and marketing practitioners tofollow formalised procedures and make their investigations more solid and rigorous(Crawford et al., 2000; Marshall, 2001).
QMRIJ8,1
10
Ongoing academic debate is either of a conceptual nature looking intosoftware-related issues or applying existing qualitative methodologies in particularcontexts. Hence, we shall detail issues related to qualitative data analysis in generaland shed some light on the formalised analysis of textual data which is most widelyused in marketing research. We shall develop arguments for the notion that qualitativeresearch is on the rise, discuss certain methodological issues in the context ofinternational qualitative research and introduce an example to show how softwaresuch as N*Vivo can be used for problems with international scope. The illustrativeexample which is used in this paper is based on a project on “knowledge management”in a multinational SME context. Company managers from internationalSME-consultancies were involved in in-depth interviews and shared their views onpractices and procedures of knowledge sharing and knowledge management.Interview data generated from three different countries in three different languageswas compiled, coded and analysed following a formalised approach.
Conceptual backgroundThe nature of qualitative researchQualitative research involves the use of unstructured exploratory techniques such asgroup discussions and in-depth interviews. In contrast to quantitative techniques it ismore difficult to precisely capture phenomena with qualitative research. As a result,“qualitative research design has often been treated as an oxymoron” (Maxwell, 1997,p. 75). However, if we want to have a holistic perspective and want to obtainin-depth-knowledge about certain objects, qualitative approach is the mostappropriate:
Qualitative data are attractive [. . .] they are rich, full, earthy, holistic, “real” and their facevalidity seems unimpeachable [. . .] (Miles, 1979, p. 590).
The choice of qualitative methods over quantitative methods can be seen as a functionof a particular research purpose. According to Maxwell (1997) there are five researchpurposes for which qualitative studies are especially useful:
(1) Understanding the meaning of events, activities, situations and actions ofpeople observed.
(2) Understanding the particular context within which the participants act and theinfluence which this context has on their actions.
(3) Identifying unanticipated phenomena and influences and generating new,“grounded” theories about the latter (Strauss and Corbin, 1998a). This view is instark contrast to the rather passive acceptance that all “great” theories had beendiscovered and that the role of research lay in testing these through quantitative“scientific” procedures (Charmaz, 1988).
(4) Understanding the processes by which events and actions take place. Theability to get insights about the processes that lead to outcomes is in many casessuperior to experimental and survey research that mostly capture outcomesonly and are poor in investigating processes.
(5) Developing causal explanations.
Analysingtextual data
11
While the capability of qualitative methods to contribute to the identification of causalrelationships has long been disputed, there is now an increasing acceptance of thelegitimacy of using qualitative methods for causal inference.
Despite this encompassing view about the nature of qualitative research,discussions almost inevitably develop into a comparison of qualitative andquantitative research. Unfortunately and despite quite early attempts to bridge thishistorical antagonism (Sieber, 1973), this relationship between qualitative andquantitative research is often seen in a contrasting perspective. Rather than focusingon the benefits, the advantages and merits of each technique, the literature points todichotomy. This has not always served the qualitative marketing researchers andpractitioners well. Dichotomies such as qualitative and quantitative, positivist andnon-positivist or numerical and non-numerical tend to limit rather than open uppossibilities for researchers (Catterall, 1998). A particularly strong critique which isoften raised against qualitative research in this context is that it is lacking inreplicability, reliability and validity.
Even a somewhat more relaxed view that qualitative research can be seen assupplementary to quantitative research (Bartunek and Seo, 2002) lacks a fullappreciation of the capabilities and merits of qualitative techniques. Any perspectivewhich reinforces the view that qualitative research is “merely” preliminary to “serious”quantitative research implicitly supports the view of superiority of quantitativemethodologies over qualitative techniques and prolongs the aforementioneddichotomy. We advocate a less dichotomous paradigmatic view. As a growingnumber of researchers come to realise that qualitative methodologies offer promisingavenues for international research, attempts need to be made to integrate the twoschools of thought (Brannen, 1992) and to determine the ingredients of “good”qualitative research.
With respect to the nature of “good” qualitative research many authors argue that itshould be easily replicable. This implies that if qualitative data sets are presented toother analysts, they should be able to follow and perform the same analysis and arriveat the same set of conclusions (Griggs, 1987). Regrettably, while quantitativeresearchers can build on explicit rules for dealing with coding problems such asdiscussions of how to treat non-response (Marshall, 2002), explicit rules for dealingwith coding issues in qualitative studies (e.g. in the context of textual data) are hardlyavailable. Some argue that as one “truth” can never be derived from data anyway,discussions about reliability and validity are inappropriate in the context of qualitativetextual data. Others point out that even for two researchers who share an similarorientation (e.g. epistemological) and use the same theoretical concepts it will bedifficult to arrive at the same end point from the same data (Marshall, 2002).
Although aiming to be less dogmatic about these perspectives, we support the lattergroup, which advocates the introduction of some practical rules for the coding andanalysis of data. The rationale follows the belief that particularly in an internationalmarketing context the “bureaucratisation of fieldwork” (Miles, 1979) and consequentlysome sort of replicability is helpful. The generation of good knowledge across nationalborders is bound to be more difficult than in single-country setting. Internationalresearch is usually built around the efforts of a number of individuals who engage ininterviewing, transcribing and coding data. Consequently, the specific orchestration ofcollaborative efforts will have to provide for findings that are generalisable.
QMRIJ8,1
12
Globalisation and its implications for international research“International” matters have been on the rise during the past 25 years and emerged tobecome a fashion statement. Politicians, business leaders as well as academics haveattempted to stay “modern” and accordingly repeated the “global mantra” (Zander,2002). This has contributed to and reinforced the globalisation phenomenon (e.g.Buckley and Casson, 1976; Levitt, 1983; Perlmutter, 1969). Both the number ofpublications as well as the proliferation of journals within which internationalmarketing and business research is published underline the continuing relevance andtopicality of the phenomenon (Pierce and Garven, 1995). Recently, the literature hasgravitated from a strong focus on large companies towards a view that SMEs play anincreasing role in the international arena. The launch of a journal specifically devotedto international entrepreneurship illustrates this trend (Acs et al., 2003) and providesjustification for investigating the analysis of textual data in the marketing andentrepreneurship context.
The globalisation literature argues that, as far as research is concerned,technological advancements help in reaching out to even the most dispersed countrymarkets with ease. However, “advances in technology both facilitate and at the sametime render more complexity towards collection of data on a global basis” (Cateora andGhauri, 2000; Craig and Douglas, 2001). Consequently methodological and empiricaldifficulties in analysing international data are becoming more and more demanding(Brislin et al., 1973; Cavusgil and Das, 1997; Mullen, 1995). We are also witnessing thatgeographical as well as psychological distance (Hallen and Wiedersheim-Paul, 1979;O’Grady and Lane, 1996) continue to pose a challenge for research in management(Evans and Mavondo, 2002).
Market dynamics and qualitative researchEmpirical findings suggest that despite all efforts and improvements in researchdesign, the achievement of reasonable response-rates from questionnaire-type studiesis becoming increasingly problematic (Fahy, 1998; Harzing, 2000; Harzing, 1997;Schlegelmilch and Diamantopoulos, 1991). There are also practical challenges involvedwith the implementation of quantitative studies on an international scale. Thisexplains why qualitative research revenues of market research companies are catchingup with ones from quantitative research and are expected to grow further inimportance. In a study drawing on a convenience sample of market research managers,Zimmerman and Szenberg (2000) illustrate that practitioners increasingly recognisethe merits of qualitative research techniques. Managers in SMEs greatly value theversatility and ease with which they can generate insights from, for instance, smallsample interviews, and, as standard quantitative techniques become extremelysophisticated and more costly to employ in international markets, they find it beneficialto stick to more creative forms of data-generation. The analysis of textual-datagenerated from interviews proves beneficial and instrumental to overcome culturalproblems in international market research (Maclaran and Catterall, 2002). Therefore,there is a need for a deeper understanding of “how” different markets work and “why”customers from different markets behave the way they behave (Ghauri and Grønhaug,2002).
Analysingtextual data
13
Reasons for growing interest in qualitative researchAs evidenced by the changes in the distribution of qualitative research revenues, theadvancement of qualitative techniques increasingly promises new means ofunderstanding and interpreting trends in various national and cultural contexts(Craig and Douglas, 2001). Qualitative research provides insights into the meaning andthe context of consumption and purchasing. It reveals the underlying determinants ofthe purchasing process, further allowing the prediction of future developments andtrends. In addition to the environmental changes, a number of factors help to explainwhy the interest in qualitative research is set to grow in the decade ahead (see Figure 1):
. Information overload. We are witnessing escalating amounts of informationranging from large quantitative databases, scanner information, transaction dataand ad hoc research information. Managers are sometimes frowned on as being“information-junkies” while researchers, attempting to make sense out of thisinformation are critically acclaimed to be on a “pilgrimage to the ivory tower”(Simon, 1994). Despite technological breakthroughs such as data-warehousingand data-mining capabilities (e.g. offered by companies such as Oracle and IBM),the difficulties to make sense of the data are persistent. One way to deal with thechallenges is to turn to qualitative research, to see what is behind all the numbersand to find creative ways to deal with the masses of information (Livingston,1994).
. Fragmentation. According to Firat (1997), the globalisation process which hasbecome of great interest to scholars is not a uniform and “universalising”process, but a fragmented one. With the emergence of new consumer segmentsrapidly transforming in increasingly multilayered cultures, researchers andcompanies need to address these new segments with new methodologies.Qualitative approaches provide a rich set of methodologies (e.g. ethnographicissues, projective and elicitation techniques) to address these issues.
. Flexibility. Related to the problem of information overload is the lack of flexibilitywith established statistical techniques. Dependence multivariate techniques arelimited by their statistical assumptions and, to the same extent that small sample
Figure 1.Factors influencing theadoption of qualitativeresearch
QMRIJ8,1
14
sizes pose problems for quantitative analysis, large samples are problematic.Qualitative methodologies are incredibly flexible (de Ruyter and Scholl, 1998;Denzin and Lincoln, 1994) and can be applied as if tailor-made to suit theunderlying research problem without requiring large samples (Sykes, 1990).
. Professionalism. Qualitative research methods have developed to become verypowerful in providing insights and revealing meaning where problems may offerthe possibility of not only one answer. Increasingly researchers turn toqualitative methods after they experience that quantitative methods cannotprovide for answers to selected problems. Further, educators and managers arelearning more and more about qualitative methods, thus becoming increasinglyconfident about the multitude of different tools and techniques.
. Hyper reality and multimedia. The dominance of written words is quicklydiminishing as far as research methodology is concerned. While most of theinformation available for interpretation and analysis is still converted into textdata, there is an increasing interest in techniques which manage to capture thecomplexity of hyper reality and intangible elements. Audio-visual informationcan now be used for computer-aided analysis, sound can be transformed intonumerical data and post-modern researchers use pictures and collages forinterpretation (Brown, 1995a, b). Qualitative methods clearly are leading the wayin the representation of our hyper reality environments.
. Information technology. Qualitative research has been influenced by advances ininformation technology. Especially in view of the multitude of new options forcollecting qualitative data (Lee and Fielding, 1991), where, for example,teleconferencing can be used to integrate expert-opinions from remote locations.Web-interfaces can help in setting up group-discussions for new products or newmarkets. With respect to qualitative data analysis itself we witness an increasingnumber of software solutions (Miles and Huberman, 1994; Titscher, 2000;Weitzman and Miles, 1995).
Text-data as data-source for international researchQualitative data can be used for description and interpretation of social behaviours,values and norms, or structures. As indicated, we focus in this paper on textual datawhich are widely used in marketing research. Textual data encompasses “any textwhich constitutes a relevant and necessary source material for answering the questionsone is interested in” (Melina, 1998). Specifically text data can comprise of openresponses to questionnaires, newspaper editorials, commentaries, titles, articles,different kinds of reports, e.g. company annual reports, memos, etc. Numerousapproaches are offered in the literature to analyse textual data in a structured,formalised way. For a comprehensive review of these approaches see Melina (1998). Inthe context of our example (i.e. a formalised analysis of knowledge management), weshall follow the approach which has been referred to as “qualitative text analysis”(Kelle, 1997; Weitzman and Miles, 1995). We shall also illustrate the practicalitiesinvolved in analysing text-data. These entail converting the text-based string-variablesinto useful, codified information. The code-base can be achieved by “coding by hand”or the use of computers[1]. The basis for text analysis, i.e. the text data or corpus,
Analysingtextual data
15
consists of spoken texts. Taped interviews with managers in five companies have beentranscribed into written text. In developing a categorisation scheme for the materialinvolved (Berelson, 1952) a “data driven approach” was pursued. This approach hasbeen labelled “grounded-theory” approach (Strauss and Corbin, 1998a) and implicatedthe construction of a categorisation scheme a posteriori (Sinkovics and Holzmuller,2001). In contrast to the “theory driven approach”, where relevant categories aredeveloped based on a theoretical understanding of the underlying concept or construct,the categorisation scheme is constructed on the basis of the textual material alone.Hence the researcher attempts to generate the categories from the appropriate textwhich is relevant for the particular analysis.
In the context of our subsequent study a structured and systematic of codificationwill be undertaken using the computer-package N*Vivo[2] (Richards, 2000). Amultiple-rater mechanism will be employed in order to increase the validity of theactual categorisation scheme.
Comparability and equivalence in international qualitative researchThe issue of comparability is one of the most challenging for international marketingresearchers. Respondents from different countries are ingrained in distinct cultures,comprising of unique patterns of socio-cultural behaviours, relevant values,psychological attitudes and traits. Consequently, their response-patterns to inquiriesand expressions of agreement or disagreement over specific concepts and constructsvary significantly. This implies that responses may not be comparable across differentnational units (Brislin et al., 1973). On the other hand, where there is commonalitybetween countries, comparable constructs and concepts can be identified, althoughinevitably the development of equivalent and standardised measures will be related tosome loss of precision and accuracy in any given nation (Craig and Douglas, 2000).This methodological dilemma has been addressed by two alternative approaches in thesocial sciences, the “emic” and the “etic” schools of thought (Pike, 1966). These can beseen as two extremes on the continuum of cross-boundary, international researchmethodology. The etic school is mostly concerned with the identification andassessment of universal attitudinal or behavioural phenomena. Predominantlyquantitative methodologies are used in an attempt to establish pan-cultural or “culturefree” measures (Elder, 1976). Conversely, the emic school holds that attitudes andbehaviour are unique to a culture and best understood in their own terms. Researchersgenerally need to make a decision upfront on whether to follow the etic or the emicschool of thought[3]. The limitations as well as the specific advantages of bothapproaches have left international researchers with a strong desire to bridge theschools of thoughts, allowing for the assessment of etic constructs and measures whilein the process of identifying emic characteristics. Berry (1989) suggested a mechanismto operationalise emics and etics, arriving at a “derived etic”, which allows for usefulcomparisons.
Particularly in the context of international qualitative research, the emic/eticdiscussion is vital. Compared with domestic market research, international marketresearch is often found to be less formal and unstructured (Cavusgil, 1987). Consequentlythis raises methodological and theoretical challenges in comparing international data.Comparative research issues have a long tradition in the sociological, anthropologicaland social sciences. They have been incorporated in marketing and international
QMRIJ8,1
16
business texts with varying degrees of comprehensiveness (e.g. Craig and Douglas, 2000;Holzmuller, 1995; Jain, 1993; Usunier, 1999).
Particular attention needs to be paid to construct equivalence to ensure that thephenomena and dimensions or constructs being studied are equivalent in all nationalsettings. Similarly measure equivalence must be sought in the translations of therelevant questions in the qualitative study. These must be equivalent and henceforthcomparable (Salzberger et al., 1999).
Following a hierarchical view of relationships and stages in the empirical researchprocess (Churchill, 1995) four stages can be identified (see Figure 2):
(1) problem definition;
(2) data collection;
(3) data preparation; and
(4) data analysis.
Within all these stages equivalence issues are pertinent and have to be addressed inorder to ensure comparability and consequently reliable and valid results.
At the problem definition stage (stage 1), the equivalence of research topicsrepresents the minimum requirement for eventually meaningful comparisons acrossnational or cultural borders. This implies that the researcher must assess whether agiven concept or behaviour serves the same function in the relevant internationalcontexts. In their widely quoted example of functional equivalence, Craig and Douglas(2000) point out that bicycles in the USA are predominantly used for recreation, whilein The Netherlands, China and other countries their main function is that oftransportation. Equivalence of research topics further implies that concepts which areused in the study have to be interpreted in a similar way. Concepts such as“materialism” or “patriotism” may be decoded differently in different countries andcultures. Consequently comparative efforts will have to take these differences intoconsideration. In our example, knowledge-management will be elaborated in an
Figure 2.Equivalence in
international research
Analysingtextual data
17
international context. It will be illustrated that within the study efforts wereundertaken to maintain conceptual equivalence.
After the research problems and related constructs to measure these have beenfinalised it is important to consider the comparability of data collection procedures(stage 2). The hierarchical relationship between the stages necessitates continuousconsideration of related issues in prior and subsequent stages. Equivalence can neverbe taken for granted. As indicated by Zimmerman and Szenberg (2000), internationalresearchers are well aware of distinct cultural problems which they may face incompleting international qualitative research. In such qualitative research thesample-size is usually much smaller than in quantitative research. However, thefundamental problem persists. This is the trade-off between intra-nationallyrepresentative samples on the one hand and internationally comparable samples onthe other hand. Finally, the qualitative research needs to be administered in such a waythat no specific time-related factors affect the equivalence of the internationallycollected datasets. These may be political circumstances and climate, economicconditions or technological settings.
Following data collection, measures need to be taken to ensure equivalence in thepreparation of data (stage 3). This involves the data being handled equally andfeedback from respondents coded in a systematic and standardised way. Infocus-group discussions or expert-interviews the issue of equivalence of data handlingis particularly significant. Especially when multiple language and multiple researchersettings are involved, data preparation can become extremely cumbersome andtime-consuming.
Finally, the goal of international research lies in the comparison of internationaldata which necessitates equivalence of data (stage 4). This is the “holy grail” ofequivalence and can be seen as a function of all prior equivalence aspects.
Having discussed conceptual issues and equivalence issues in the context ofinternational qualitative research, we shall introduce an empirical research project onknowledge management to demonstrate the use of software and formalised proceduresfor checking, analysing and comparing international data in the form of textual data.
Methodology – a formalised analysis of knowledge managementKnowledge management: rationale and purpose of the studyManagers are increasingly concerned with information overload. Too muchinformation about too many things has to be absorbed quickly and in a short periodof time. In globalising markets the levels of uncertainties, aggravated by cultural andpolitical dissimilarities, add to this pressure. The literature has approached thisproblem from a number of perspectives, with knowledge and knowledge managementrecently playing a major role in the discussion of international companies (e.g. Bennettand Gabriel, 1999; Buckley and Carter, 2000, 2002; Mudambi, 2002) and SMEs(Magnusson and Nilsson, 2003; McAdam and Reid, 2001)
Considerable efforts have been made to investigate the role of knowledge resourcesin developing and maintaining sustainable competitive advantages (Davenport andPrusak, 1998; De Geus, 1988; Kim, 1993; Prahalad and Hamel, 1990; Stata, 1990) or theflows and transfer of knowledge within multinationals (e.g. Foss and Pedersen, 2002).As far as the individual level of knowledge management is concerned, there is ashortage of empirical work on employees’ willingness to share knowledge. Knowledge
QMRIJ8,1
18
sharing can be seen as a function of interpersonal relations between managers andemployees. Hence it is presumed that these are determined by situational, environmental,organisational and social factors. However, people may also be somewhat naturallyreluctant to share their most precious assets, knowledge and experience.
Consequently, this study aims to address this paucity of research onknowledge-management and interpersonal relations. A knowledge-based industrycontext is chosen to empirically investigate the company-manager-employeeinteractions. A formalised, qualitative approach will be used to identify antecedentsof successful knowledge management practices and to obtain a holistic understandingof knowledge management. The specific research questions are:
. What kind of relationships do knowledge managers perceive between acompany’s goals and culture, and knowledge management?
. How important is the motivation of employees and which motivationaltechniques are used? With what success?
A qualitative design was pursued since it was felt that a holistic perspective ofknowledge sharing, of motivational issues to share knowledge and of managers’perspectives on knowledge management within their company environment could notbe gained otherwise. We concentrated on textual data which were collected throughinterviews and used a grounded theory (Strauss and Corbin, 1998b) approach todevelop a theory. As indicated above, the purpose was to develop insights into therelationship between company, manager and employees regarding knowledgemanagement and knowledge sharing.
Instrument development and data collectionAn interview guideline was developed which covered issues discussed previously inthe literature considering the international application of the semi-structured format.To ensure functional and conceptual equivalence, the interview guideline began with aquestion on how knowledge management was seen in the company and what itinvolved. With respect to data collection we standardised the process in such a waythat equivalence of research methods, units and administration was ensured. Theinterviewers were trained to interact professionally with their respective interviewpartners and measures were taken to ensure standardised and comparable interviewsituations. Cross-cultural issues were discussed in two extensive training meetings andit was felt that the interviewers were satisfactorily equipped with the knowledgeneeded. A laddering-type interview process was encouraged to facilitate theclarification of issues, verification of interpretations of answers during the interview,and persistence in following up on emerging topics and themes arising during theinterview (Arksey and Knigth, 1999; Kvale, 1996; Lee, 1999; Rubin, 1995; Strauss andCorbin, 1998b).
At the outset 18 managers from nine international SME consulting companies wereapproached via telephone or physically during company events to participate in theproject. The companies were selected on the basis of size and their core businesses.Technology-based companies such as ones engaged in technical consulting, computerhardware and software, or electrical equipment were selected. Their potential level ofinvolvement in intra-firm knowledge management activities was assessed. Particularattention was given to international SMEs with a foothold in more than two markets,
Analysingtextual data
19
as their extended geographical reach suggested the implementation of certainprocedures regarding knowledge management. Furthermore the core business ofcompanies and their expertise in the field of knowledge management was relevant:seven companies were willing to participate, which ultimately resulted in nineinterviews conducted in three different countries, Austria, Germany and Italy. In twocompanies, more than one individual was interviewed which allowed for intra-firmcomparisons of views on knowledge management. The respondents were recruitedfrom top and middle management. They were responsible for knowledge managementin addition to other tasks. With respect to the physical context, this was standardisedto the managers’ offices, and the interviews lasted between one and two hours each.English, German and Italian was used. The interviews were tape-recorded andtranscribed afterwards.
In addition to the semi-structured interviews, interviewers noted contextual factorsof the interview situation. These were physical factors, the company building,company entrance, the reception and the notification to the interview partner. Theinfrastructure of the office was observed, taking note of informal spots whereemployees could meet and talk, coffee machines, open offices and the behaviour ofemployees within the building, as when having lunch together or having meetings.Visual materials, such as advertisements, rules and regulations for employees on therespective company web sites and so on were also included in the analysis aftercodification of the information as text.
Taken collectively, rich textual material was collected and subsequentlyanalysed using N*Vivo. In the following sections the different stages in theanalysis are described and illustrated. The empirical study of knowledgemanagement in international SME consulting companies serves as a demonstrationof formalised text-analysis in a cross-border context. We argue that the use ofCAQDAS provides for certain procedural advantages compared to traditionalmeans of text analysis. It is further argued that these advantages ultimatelyhelped in the formalisation of processes which otherwise would have had to becarried out by labour-intensive traditional means. In order to substantiate thisclaim, we outline the differences between qualitative text analysis by traditionalmeans and by CAQDAS in Table I.
The procedure which was pursued in this formalised approach to text analysis isillustrated in Figure 3. Using the empirical data we aimed for a holistic perspective ofthe company, manager and employee relationship for theory building. Two bi-lingualresearchers, one fluent in Italian and English, the other in German and English, wereinvolved in the coding process, which helped to safeguard against the criticism ofsubjectivity, hermeneutics and value-load. The categorisation scheme was developedin English and continually monitored, sharing ideas and concepts and updated byquestioning with the co-analyst. This procedure safeguarded against the danger of apurely uniform coding scheme as in the “etic” school of thought which would not haveallowed for the identification of country specificities. At the same time, thissafeguarded against biases and ensured equivalence of data handling.
Analysis using N*VivoN*Vivo’s group coding feature was used for analysis purposes. N*Vivo is a veryversatile and user-friendly program which on start-up introduces the Project Pad to the
QMRIJ8,1
20
researcher and provides an overview of central elements and such options as
“documents”, “nodes”, “attributes” and “sets”. Also the analytical processes of linking,
coding, modelling and searching are graphically illustrated within the pad.
The analytical steps which facilitated the formalised approach to the analysis of our
textual data using N*Vivo are illustrated in Table II. In total five core analytical
processes were identified:
(1) organising;
(2) linking;
(3) coding;
(4) searching; and
(5) modelling.
In Table II the core processes are sub-divided into several steps and discussed in terms
of advantages and problems respectively.
Figure 3.Illustration of a formalised
textual data analysisusing CAQDAS
Issues Traditional means CAQDAS
Management oflarge amountsof text data
Cards (paper)Transcripts (paper)
Storage of large, electronic documentsProvision of first reports, quickly and easily
– on sample type and size as well as– on categories (number and content)
Provision of complex picture of data and sampleInclusion of, for example, field notes in the data
Record keeping ShufflingOrganisation into pilesLosing piles sometimes
Unlimited shufflingStorage of memos, electronicallySecure storage of documents
Coding Screening of every cardor transcriptCoding of these separately
Application of standards
Automated coding and searching (auto-coding)Searching Highlighting text
sectionsCutting out of relevantsections
Use of headlines, text styles etc. to structuredocumentsBrowsing documents to show selectedsections without losing the full document
Note: Based on Marshall (2001)
Table I.Qualitative text analysisprior to use of CAQDAS
Analysingtextual data
21
N*V
ivo
step
sIl
lust
rati
on(K
Mp
roje
ct)
Ad
van
tag
esD
isad
van
tag
es/p
rob
lem
s
Org
anis
ing
proc
esse
sS
tep
1:S
ourc
eof
dat
a/se
lect
ing
dat
aty
pes
Inpu
tdata
:In
terv
iew
sw
ith
man
ager
s;d
ocu
men
tsfr
omco
mp
anie
s(e
.g.
“gol
den
rule
s”);
Pic
ture
(e.g
.il
lust
rati
ng
aco
mp
any
’sh
isto
ry)
Tec
hniq
ue:
Inte
rvie
ws
and
seco
nd
ary
rese
arch
Res
ult
:R
ich
tex
tfo
rmat
doc
um
ent
wit
hsu
b-h
ead
ing
s,li
nk
sto
doc
um
ents
and
pic
ture
s
Th
eore
tica
lsa
mp
lin
gO
pen
inte
rvie
wte
chn
iqu
eA
ssoc
iati
onte
chn
iqu
eO
ther
tex
tual
and
non
-tex
tual
mat
eria
lis
incl
ud
edA
lrea
dy
exis
tin
gm
ater
ial
can
be
inco
rpor
ated
Inte
rvie
wer
sub
ject
ivit
yS
elec
tion
ofin
terv
iew
ees
Inte
rvie
wte
chn
iqu
esar
ed
iffe
ren
tw
hen
mu
ltip
lein
terv
iew
ers
are
inv
olv
edW
hic
had
dit
ion
alm
ater
ial
isre
lev
ant?
Ste
p2:
Des
crib
ing
the
dat
aIn
putdata
:Ric
hte
xt
form
atd
ocu
men
tw
ith
sub
-hea
din
gs,
lin
ks
tod
ocu
men
tsan
dp
ictu
res
Tec
hniq
ue:
Des
crip
tion
ofin
terv
iew
s:p
roto
cols
/mem
osof
inte
rvie
wer
sab
out
the
con
tex
t,th
esi
tuat
ion
,th
ep
erso
n,
the
outc
ome,
idea
sab
out
cate
gor
ies
orco
nce
pts
Res
ult
:E
xte
nd
ing
dat
aset
wit
hd
escr
ipti
ve
tex
tual
dat
a
Asp
ects
wh
ich
mig
ht
not
be
rele
van
tat
firs
tsi
gh
tar
est
ored
tog
eth
erw
ith
the
raw
dat
aIn
crea
sed
com
par
abil
ity
ofd
ata
Hel
ps
tou
nd
erst
and
resp
ecti
ve
situ
atio
nin
wh
ich
dat
aw
ere
coll
ecte
d(f
rom
inte
rnat
ion
alp
ersp
ecti
ve)
“Noi
se”
orir
rele
van
tin
form
atio
nm
igh
tb
ein
clu
ded
(con
tinued
)
Table II.Analytic design –procedures of dataanalysis andinterpretation
QMRIJ8,1
22
N*V
ivo
step
sIl
lust
rati
on(K
Mp
roje
ct)
Ad
van
tag
esD
isad
van
tag
es/p
rob
lem
s
Ste
p3:
Ch
ang
ing
and
vie
win
gd
ata
Inpu
tdata
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
emT
echniq
ue:
Cod
eth
ed
ocu
men
tin
bro
wse
dte
xt
(sel
ecti
onm
ode)
and
edit
the
tex
tR
esult
:C
lari
fica
tion
ofre
lev
ant
tex
t
Str
uct
ure
the
doc
um
ent
and
hig
hli
gh
tre
lev
ant
sect
ion
sw
hil
ecl
eari
ng
irre
lev
ant
sect
ion
s
Con
tex
tin
form
atio
nm
igh
tg
etlo
st
Ste
p4:
Gro
up
ing
Inpu
tdata
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
emT
echniq
ue:
Pu
ttin
gto
get
her
doc
um
ents
orn
odes
inan
yn
um
ber
ofse
tsR
esult
:S
ets
(doc
um
ent
and
nod
e)
Str
uct
uri
ng
ofre
lev
ant
dat
aR
edu
cin
gte
xtu
ald
ata
tem
por
aril
yW
ron
gd
ocu
men
ts/n
odes
are
gro
up
edto
get
her
bu
tar
eh
and
led
assi
mil
ar
Ste
p5:
Sto
rin
gin
form
atio
nan
din
clu
din
gat
trib
ute
s
Inpu
tdata
:Q
uan
tita
tiv
ein
form
atio
n(e
.g.
com
pan
ysi
ze,
nu
mb
erof
emp
loy
ees)
Tec
hniq
ue:
Cre
ate
and
edit
attr
ibu
tes
from
spre
adsh
eets
orst
atis
tica
lp
ack
ages
Res
ult
:E
xte
nd
ing
dat
aset
wit
hd
escr
ipti
ve
nu
mer
ical
dat
a
Imp
orti
ng
alre
ady
exis
tin
gsp
read
shee
ts,
also
from
seco
nd
ary
mar
ket
rese
arch
Fil
teri
ng
doc
um
ents
bas
edon
attr
ibu
tes
Too
mu
chem
ph
asis
onn
um
eric
ald
ata
Eas
y-t
o-im
por
tfe
atu
reco
uld
lead
toen
orm
ous
dat
abas
ew
hic
his
dif
ficu
ltto
han
dle
Lin
king
proc
esse
sS
tep
6:L
ink
ing
doc
um
ents
and
nod
esIn
put
data
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
emT
echniq
ue:
Cre
ate
lin
ks
toot
her
doc
um
ents
orn
odes
inth
esa
me
ord
iffe
ren
tp
roje
ctor
toex
tern
ald
ata
Res
ult
:D
ocL
ink
san
dN
odeL
ink
s;li
nk
ages
Qu
alit
ativ
eli
nk
ing
En
able
sth
ere
sear
cher
toli
nk
doc
um
ents
and
nod
esp
rior
toco
din
gor
Wh
enco
din
gis
not
pos
sib
le
Dif
fere
nce
bet
wee
nli
nk
san
dco
des
mig
ht
be
un
clea
ran
dco
nfu
se(i
nex
per
ien
ced
)re
sear
cher
(con
tinued
)
Table II.
Analysingtextual data
23
N*V
ivo
step
sIl
lust
rati
on(K
Mp
roje
ct)
Ad
van
tag
esD
isad
van
tag
es/p
rob
lem
s
Cod
ing
proc
esse
sS
tep
7:C
odin
gan
dau
toco
din
gIn
put
data
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
emT
echniq
ue:
Cre
ate
nod
es(f
ree,
tree
and
case
nod
es);
exp
lore
wh
at’s
cod
edR
esult
:N
ode
syst
em;
bro
wsi
ng
nod
esy
stem
Sav
esti
me
Incr
ease
dre
liab
ilit
y,
e.g
.b
yre
du
cin
gh
um
aner
ror
Pot
enti
alfo
ru
nex
pec
ted
insi
gh
tsth
rou
gh
re-c
onte
xtu
alis
ing
mat
eria
l
Pot
enti
alof
mec
han
ical
erro
rsD
ang
erof
sup
erfi
cial
anal
ysi
sD
e-co
nte
xtu
alis
ing
mat
eria
l
Ste
p8:
Rev
isin
gan
dre
fin
ing
Inpu
tdata
:N
ode
syst
em-b
row
ser
Tec
hniq
ue:
Del
ete,
refi
ne,
chan
ge
nod
es;
use
oth
erte
chn
iqu
esto
trac
eth
ep
roce
ss(m
emos
,li
nk
s)R
esult
:N
ewn
ode
syst
em
Kee
ps
nod
esy
stem
aliv
eb
yen
abli
ng
easy
chan
ge
ofco
din
gA
llow
sn
ewid
eas
tob
ein
teg
rate
d
Eas
yw
ayof
dea
lin
gw
ith
nod
esm
ayle
adto
an
ever
-en
din
gco
din
gp
roce
ss(w
hen
tost
opco
din
g?)
Sea
rchin
gS
tep
9:W
hat
toas
k?
Inpu
tdata
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
emT
echniq
ue:
Op
erat
ors
(nod
ean
dat
trib
ute
look
up
,te
xt,
Boo
lean
and
pro
xim
ity
sear
ch,
vec
tor
and
mat
rix
sear
ches
Res
ult
:M
atch
es
Ran
ge
from
easy
tov
ery
tail
ored
and
com
pre
hen
siv
ese
arch
Sp
eed
ofse
arch
Han
dle
larg
en
um
ber
sof
nod
esS
tore
sear
ches
inv
ario
us
way
sR
estr
ict
sear
ches
Dea
lw
ith
mu
ltip
lean
dov
erla
pp
ing
cod
esC
ond
uct
mu
ltip
lese
arch
Res
earc
her
nee
ds
tok
now
qu
esti
ons
inad
van
ceK
now
led
ge
abou
top
erat
ors
nec
essa
ry
Ste
p10
:W
her
eto
ask
it?
Inpu
tdata
:N
*Viv
od
ocu
men
ts(m
emos
),at
trib
ute
san
dn
ode
syst
em;
sets
(doc
um
ents
and
nod
es)
Tec
hniq
ue:
Usi
ng
assa
yto
olp
rior
tose
arch
;ch
oosi
ng
scop
eof
sear
ch(s
pec
ific
doc
um
ents
,n
odes
);co
mp
are
and
reru
nse
arch
esR
esult
:R
epor
ton
the
scop
eit
em;
mat
ches
As
per
Ste
p9
As
per
Ste
p9
(con
tinued
)
Table II.
QMRIJ8,1
24
N*V
ivo
step
sIl
lust
rati
on(K
Mp
roje
ct)
Ad
van
tag
esD
isad
van
tag
es/p
rob
lem
s
Ste
p11
:W
hat
tod
ow
ith
the
answ
er?
Inpu
tdata
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
em;
sets
(doc
um
ents
and
nod
es)
Tec
hniq
ue:
Col
lect
fin
din
gin
ton
odes
;st
ore
sep
arat
ely
,ap
ply
assa
yto
olR
esult
:M
atch
es
As
per
Ste
p9
As
per
Ste
p9
Mod
ellin
gpr
oces
ses
Ste
p12
:D
raw
ing
and
lin
kin
gm
odel
sIn
put
data
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
em;
sets
(doc
um
ents
and
nod
es)
Tec
hniq
ue:
N/A
Res
ult
:V
isu
alre
pre
sen
tati
onof
idea
s(m
odel
s)
Use
inea
rly
stag
eto
des
ign
the
pro
ject
and
its
dev
elop
men
tC
lari
fyn
odes
and
doc
um
ents
Cla
rify
con
cep
tC
ogn
itiv
em
apan
dca
usa
ln
etw
ork
Cat
egor
yd
evel
opm
ent
Tim
ese
qu
ence
mod
el
N/A
Ste
p13
:M
anag
ing
mod
els
Inpu
tdata
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
em;
sets
(doc
um
ents
and
nod
es);
mod
els
Tec
hniq
ue:
Del
ete
par
ts,r
efin
e,ch
ang
em
odel
sR
esult
:A
dap
ted
mod
els
“Liv
e”m
odel
sE
asy
swit
chb
etw
een
all
par
tsof
the
N*V
ivo
pro
ject
N/A
Ste
p14
:L
ayer
ing
and
gro
up
ing
item
sIn
put
data
:N
*Viv
od
ocu
men
ts(m
emos
)an
dn
ode
syst
em;
sets
(doc
um
ents
and
nod
es)
Tec
hniq
ue:
Incl
ud
ela
yer
sR
esult
:L
ayer
edm
odel
s
Sh
owd
iffe
ren
tle
vel
sof
inte
rpre
tati
onR
epre
sen
tsp
rog
ress
ive
dis
cov
ery
pro
cess
N/A
Source:
Tab
leb
ased
onco
nce
pts
adap
ted
from
Mar
shal
l(2
001)
,C
offe
eyet
al.
(199
6)an
dR
ich
ard
s(2
000)
Table II.
Analysingtextual data
25
Organising processesAfter data collection (Step 1, Table II), organisation (Step 2), changing and browsing(Step 3), grouping (Step 4) and storing information about the data (Step 5) analyticalsteps followed. Subsequent to the creation of a new project, data had to be assigned tothe project. Documents were data files which could be created directly using N*Vivo asa text editor or alternatively imported from existing data files created by wordprocessors and saved as rich text files. In the present study, the data files comprised ofnine interview transcripts and the observations made during, before and after theinterview. Original interview language was retained in the program but a commonEnglish coding scheme was developed. Additionally, memos could be included andattached to the project. These were important elements of the coding and interpretationprocess, particularly when dealing with multiple coders (see Figure 4).
Another element of the project, called attributes, helped to organise characteristicsof the documents such as company figures, information about the respondent andcharacteristics of the coding system which were descriptions of nodes. These thuscombined qualitative and quantitative data. Attributes can be established usingstatistics software (e.g. SPSS) or from any software that generates tabular output (e.g.Excel).
We included factual data about the company such as the number of employees,fields of operation and number of offices world-wide in our analysis (Figure 5).Additionally information from the interview partner was included, such as his or her
Figure 4.Document explorer
Figure 5.Using attributes todescribe documents withinN*Vivo
QMRIJ8,1
26
responsibilities regarding knowledge management and his or her position within thecompany.
Linking and coding were the next processes in the analysis. Nodes in the project padorganise the derived codes and categorisation schemes.
Linking and coding processesNext to the coding procedure, N*Vivo offered another way of outlining relationsbetween various parts of the project. Linking documents and nodes (Step 6, Table II)could serve as an alternative to coding, but it could also supplement it. The linkingfeature created links from one document to another or to external data. In the case ofthe knowledge management system employed by a particular SME, we created linksbetween manager interviews and observations, as recorded by the interviewer. Hencewe linked observations about management behaviour in the interview, guiding toursaround the company premises and management-employee interactions with the textualdata from the interview.
Coding textual data was probably the most crucial step in the analytical process(Step 7, Table II). The coding process could be seen as an ongoing interpretation andexamination of textual data from different perspectives and would also depend on thenumber of researchers involved. Two coding strategies were used: (a) a priori and (b) aposteriori categorisation of data. A priori categorisation involved the use of theory,literature or exploratory interviews with experts in the field to develop categorieswhich would subsequently be used for the analysis. A posteriori categorisation meantthat empirical indicators would be obtained directly from the data and henceforthcoded. Considering the specific paucity of relevant literature in our area, knowledgemanagement including employer to employee relationship, a posteriori coding wasapplied. This was strategy (b).
Heading styles could be used to facilitate a first, “rough” coding, which is also calledautocoding. The present project started by using the heading styles of the interviewswhich were defined in line with the interview guideline. Each of the interviews wasroughly coded into sections and a structure of textual data was established (seeFigure 6). The Node Browser within N*Vivo allowed us to inspect coded sections of thedocument.
Next each text-section was analysed with more scrutiny, following open, axial andselective coding processes, as suggested in the literature[4]. Therein concepts wereestablished and statements used to explain the phenomenon of interest. The textualdata were reduced and, as suggested by Lee (1999) or Strauss and Corbin (1998b) acertain level of abstraction was reached.
In the coding process, the researcher attempted to identify the right “container” forideas and concepts. Ultimately this quest for structure resulted in a node system, whichcould be updated throughout the coding process. A node system might have consistedof connected node groups with child and sibling nodes or of different trees. By makinga decision about the size of the node system the researcher had to find a balancebetween breadth and depth. The node system could always be seen as a function of thestage of the research process and would evolve over time. Since coding was an ongoingprocess that theoretically never stopped. Practically, the coding process only stoppedwhen the researcher had reached “theoretical saturation” and no new themes emerged
Analysingtextual data
27
(Marshall, 2002). In our study for each of the sections obtained from rough coding up to36 nodes were created. Two trees were created to facilitate management of the nodes.
The question regarding “Origin of knowledge management” for example produced18 nodes in the beginning, given that interview partners expressed their experiencedifferently or simply used different levels of abstraction. Some managers mentionedreal problems they faced in the beginning. A technology manager commented:
There was a day, when the knowledge management tool was there. We sent an e-mail toeverybody, but this did not necessarily mean that the job was finished. Quite on the contrary,it just had started.
A consultant talked more about the company’s behaviour:
The Company has been considering knowledge management as very important for a longtime.
We were also dealing with different languages and the coders used different codeswhich were aligned subsequently for reasons of equivalence.
After inspecting the nodes and discussing the node system, the two coders mergedseveral nodes following an open coding strategy. Subsequently axial coding wasapplied to inter-relate the categories and sub-categories. As a result, nodes whichexpressed the same concepts were merged and the number of child nodes was reducedto 11. In a similar manner the other nodes were initially created independently,discussed among the coders, and finally a common categorisation system wasapplied[5].
Figure 6.“Rough” coding usingheading and sub-headingstyles
QMRIJ8,1
28
Selective coding was applied to develop a theoretical understanding of the companyto manager to employee relation and its influence on knowledge sharing. Therefore, weused this feature and linked nodes with other nodes or nodes with memos that emergedduring coding. Nodes coded under the section “origin of knowledge management” werelinked to nodes coded within the section “company culture” such as the concept of“communication” which appeared several times. Regarding the managers’understanding of knowledge management, many different views were identified.They ranged from the managerial task to the inter-personal view of sharing somethingproprietary. Many managers were concerned about the loose ties between employeesand regarded knowledge management as a tool to increase contacts and facilitatecommunication.
After the right container for ideas has been identified, the researchers wanted toinspect the containers to find out and theorise about its content. With the searchprocess, N*Vivo offered a deeper analysis of textual data. This enabled researchers topose whatever question they might want to ask and the software would provide for thetext-based answer immediately.
Searching processesStarting point for project-based search were research questions relating to the textualdata. The research questions guided the researcher through the search process andfinally helped in the identification and build-up of a theory (Richards, 2000). At thisstage in the project, the data comprised documents, interview transcripts, observationsmade during interview, a refined node system, the supplementary attributes ofquantitative data and a series of memos of textual data.
To put it simply, the first question was to ask “what am I looking for in the project?”This lead to the decision whether coding-combinations, text, values of attributes, orrelations (Boolean, proximity) were looked for (Step 9, Table II). Second, “where am Ilooking for this in the project?” selected the documents and nodes that would bescreened (scope tool). Additionally the assay tool helped to point to interesting sectionswithin the documents, undertake comparisons, rerun searches and evaluate differentoutcomes. By using the assay tool certain features of documents and nodes could bechecked prior to the real search process (Step 10, Table II). Finally, the question, “whatdo I want to do with the results?” determined whether only certain sections of thedocuments were displayed or also the context, where the information found would beincluded (Step 11, Table II).
In our study we were looking for data which illustrated the company to manager toemployee relation. Additionally we were interested in different perceptions ofknowledge management, depending on the positions of consultants or technologymanagers. Therefore, nodes such as “importance of employee”, “communication”, etc.were searched and a matrix intersection between nodes and companies was produced.All documents were included into the analysis and the results were displayed as anode. Figure 7 illustrates the number of documents where the nodes were found,separated into four fields (1 ¼ E-Business, 2 ¼ Consulting, 3 ¼ Electronic Equipment,4 ¼ Computer). The importance of building up networks was only relevant forcompanies that specialised in e-business. For consulting companies employees wereimportant for knowledge management. In addition, the number of characters coded orcoding references could be displayed. Cells with high frequencies were automatically
Analysingtextual data
29
highlighted and frequency tables could be exported into SPSS-data files and used forfurther analysis.
We were also interested in potential cross-country differences. Consequently wesearched for the co-occurrence of the company’s view that communication played animportant role within an ideal knowledge management environment and companyculture. The results revealed that only for an Italian consultant the issues “knowledgemanagement” and “team orientation” were linked together. This particularinterviewees specific statement reinforced perceptions such as his SME being a“people’s company”, a “spirit of togetherness” within the company and the notion that“closer alignment with the international goals” might threaten to make the workplace“less enjoyable”. In all the other statements, knowledge management was closelylinked to issues of organisational structure, “flat hierarchy” and the conflict between“technicians and management”. It appeared that these perspectives, although probablylimited in their generality, pointed out differences pertaining to the different nationalcontexts involved in the study.
Modelling processesModels of graphical illustrations of the underlying textual data could be created atevery stage of the project using the Model Explorer. There was no automaticgeneration of models in N*Vivo. However, the researcher could build models usingdocuments, nodes, attributes and memos (Step 12, Table II). This could help to buildnodes and to facilitate the visual construction of a nodal system. Furthermore itassisted in the development of a categorisation scheme which could be used in thesubsequent coding process. Modelling was also instrumental for the conceptualisationof ideas that arose during the ongoing coding and search process. Modelling helped todesign the project but was also helpful for the illustration of the research progress.Various models which might evolve during the research process could be drawnseparately and then graphically linked together.
Figure 7.Matrix intersection
QMRIJ8,1
30
Models in N*Vivo could be seen as living entities. They were evolving, continuallychanging, refined by the researcher and updated according to the research progress(Step 13 and step 14, Table II).
The nodes concerning the “origin of knowledge management” are displayed inFigure 8, allowing visual analyses. Nodes representing the employee, for example“must activate employee” and “building up networks” as well as nodes relating to theexperience managers felt in the beginning of knowledge management were grouped.“Unclear beginning”, “first reactions þ”, “bad experience 2” were included. The nodeswere then transformed and enriched with researchers’ ideas and interpretations. Therole of management for example influenced whether knowledge management wasimplemented “top-down” or “bottom-up”. Since different layers and groupings could beused in the graphical presentation, the perspectives of different researchers could beviewed simultaneously.
ConclusionManagers were commenting much about the initial stages of implementing knowledgemanagement systems. Yet consultants, particularly those operating on an internationallevel, did not consider knowledge management to be a novel thing. They simply felt itwas important. This was in line with the finding that more international SMEsattributed a higher importance to knowledge management issues. Similarly, within themore internationalised SMEs individuals were less likely to abstain from participationin the process of knowledge sharing and dissemination. In contrast, managers oftechnical innovations within the SME context felt that knowledge managementgenerated problems. According to their view, knowledge management did not alwayshelp the employees whose careers were built on knowledge and consequently theywere not completely convinced of that the idea had significant benefits.
When asked about the company’s goals, managers mentioned that the mostimportant intellectual capital from a company’s perspective was the employees. Thisparticular node was coded 12 times throughout the interviews. It comprised statementsreferring to the enhancement of knowledge sharing on a personal level, entrepreneurialpersonality traits and individual characteristics. It was felt that knowledge
Figure 8.Model explorer
Analysingtextual data
31
management could develop into an accepted practice only if innovative employeesendorsed this idea and communication between the management and employees wasfacilitated.
Consequently, issues of “communication” and “facilitation of interaction” could beconsidered as key factors for the successful implementation of knowledge managementwithin SMEs. These issues will be important to consider in relation to future studies. Itis interesting that our finding goes beyond the scope of earlier studies in the context ofknowledge management which predominantly focused on systems and tools ofknowledge management. Taking a formalised view on textual interview data, we wereable to identify those relational dimensions and exchange-processes within theorganisation which are even more important. Hence, a holistic view of the relationshipbetween company, managers and employees was provided. We were able to transcendthe mere quantitative description of why some knowledge management systems workwhile others do not.
Implications and future outlookQualitative research methodology, including that in the growing body of work at themarketing and entrepreneurship interface, is often criticised for high levels ofsubjectivity and low reliability and validity. On a substantive level this criticism isunfair because qualitative research offers holistic perspectives on phenomena whichcannot be achieved otherwise. However, the criticism is often due to a low quality ofdocumentation and reporting of the findings and cannot be ignored. While quantitativestudies follow a rigorous organisation and presentation in how results are presented,qualitative studies are often reported in a descriptive and narrative way. Following aformalised procedure of organising a particular kind of qualitative research as in theanalysis of textual data, we attempted to overcome this criticism. Hence we presentedmethods and procedural steps which when employed may help to add to the credibilityof the findings. As suggested by Lee (1999), above and beyond sound conceptualbackground and literature, a detailed and clearly organised way of presentingempirical findings is warranted. Therefore an illustrative research example onknowledge management in an international SME context was presented and organisedaccordingly.
Computer software programs such as N*Vivo help and support researchers inmaking the analytical process of coding and analysing textual data more accessible forother researchers. Further formalised procedures will help to ensure that issues ofequivalence are addressed in the international context. Particularly when manyresearchers are involved, the formalisation and continuous interaction of computersystem and researcher encourage the “questioning” and challenging of thefundamental assumptions made while coding. Therefore this interaction will lead tobetter research results particularly at the marketing and entrepreneurship interface.
Formalised processes promise to make the qualitative inquiry based of textual datamore logical and replicable. This is encouraging since the replicability of researchfindings is generally considered to be the “holy grail” of scientific research in that itensures scientific knowledge by continually challenging it. Within this paper, we focuson one specific form of qualitative analysis – the analysis of textual data. Theargument of following formalised procedures and their suggested merits may notnecessarily be transferable to other qualitative methods. Nevertheless we hope that a
QMRIJ8,1
32
more formalised approach to the analysis of textual data will help to crossparadigmatic borders on different types of inquiry and alleviate certain prejudices inthe adoption of these methodologies.
Notes
1. Although some researchers express concern related to the potential theoretical andmethodological costs of computer use in qualitative research (Coffeey et al., 1996), we believethat the danger of methodological biases and distortions arising from the use of certainsoftware packages is overemphasised in the discussion (Kelle, 1997). In fact, the applicationof computer software makes life easier for the researcher. Moreover, it formalises the way inwhich a researcher can look at the text body. Hence, qualitative software has the potential toincrease the reliability of research findings by making the process of analysis moresystematic and transparent.
2. We reviewed and ranked a number of different software packages (Atlas.ti, C-I-Said,Ethnograph, HyperResearch, N*VIVO, Nudist and MaxQDA) by looking at developers’homepages. We decided to use N*Vivo because of particular coding features andgroup-features, which were deemed necessary for our research.
3. In relationship to the empirical study within this paper the approach was to aim forcomparisons but allow for cultural idiosyncrasies. A coding scheme was therefore developedwhich was not purely uniform.
4. Open coding is usually used for the discovery of categories and the identification of newconcepts. Axial coding applies categories and concepts to empirical data. Here, categories arerelated to their subcategories and intersections of related categories are identified. Theobjective of axial coding is to add depth to categories. Finally, selective coding is the processwhere categories are integrated and refined in order to build a theory (Step 8, Table II).
5. It has to be noted that national specificities were retained, i.e. nodes were only merged intoone common categorisation system, provided that the underlying concepts were equivalent.
References
Acs, Z., Dana, L.-P. and Jones, M.V. (2003), “Toward new horizons: the internationalisation ofentrepreneurship”, Journal of International Entrepreneurship, Vol. 1 No. 1, pp. 5-12.
Arksey, H. and Knight, P. (1999), Interviewing for Social Scientists, Sage Publications, ThousandOaks, CA.
Bartunek, J.M. and Seo, M.-G. (2002), “Qualitative research can add new meanings to quantitativeresearch”, Journal of Organizational Behavior, Vol. 23 No. 2, pp. 237-42.
Bennett, R. and Gabriel, H. (1999), “Organisational factors and knowledge management withinlarge marketing departements: an empirical study”, Journal of Knowledge Management,Vol. 3 No. 3, pp. 212-25.
Berelson, B. (1952), Content Analysis in Communication Research, Free Press, Glencoe, IL.
Berry, J.W. (1989), “Imposed etics-emics-derived etics: the operationalization of a compellingidea”, International Journal of Psychology, Vol. 24, pp. 721-35.
Bickman, L. and Rog, D.J. (1997), Handbook of Applied Social Research Methods, SagePublications, London.
Brannen, J. (Ed.) (1992), Mixing Methods: Qualitative and Quantitative Research, Avebury,Aldershot.
Analysingtextual data
33
Brislin, R.W., Lonner, W.J. and Thorndike, R.H. (1973), Cross Cultural Research Methods, JohnWiley & Sons, New York, NY.
Brown, S. (1995a), Postmodern Marketing, Routledge, London.
Brown, S. (1995b), “Postmodern marketing research: no representation without taxation”,Market Research Society. Journal of the Market Research Society, Vol. 37 No. 3,pp. 287-310.
Buckley, P.J. and Carter, M.J. (2000), “Knowledge management in global technology markets;applying theory to practice”, Long Range Planning, Vol. 33 No. 1, pp. 55-71.
Buckley, P.J. and Carter, M.J. (2002), “Process and structure in knowledge management practicesof British and US multinational enterprises”, Journal of International Management, Vol. 8No. 1, pp. 29-48.
Buckley, P.J. and Casson, M. (1976), The Future of the Multinational Enterprise, PalgraveMacmillan, London.
Cateora, P.R. and Ghauri, P.N. (2000), International Marketing (European ed.), McGraw-HillPublishing Company, London.
Catterall, M. (1998), “Academics, practitioners and qualitative market research”, QualitativeMarket Research: An International Journal, Vol. 1 No. 2, pp. 69-76.
Cavusgil, S.T. (1987), “Qualitative insights into company experiences in international marketingresearch”, The Journal of Business & Industrial Marketing, Vol. 2 No. 3, pp. 41-54.
Cavusgil, S.T. and Das, A. (1997), “Methodological issues in empirical cross-cultural research:a survey of the management literature and a framework”, Management InternationalReview, Vol. 37 No. 1, pp. 71-96.
Charmaz, K. (1988), “The grounded theory method: an explication and interpretation”,in Emerson, R.M. (Ed.), Contemporary Field Research: A Collection of Readings, WavelandPress, Prospect Heights, IL.
Churchill, G.A.J. (1995), Marketing Research – Methodological Foundations, 7th ed., The DrydenPress, Fort Worth, TX.
Coffeey, A., Holbrook, B. and Atkinson, P. (1996), “Qualitative data analysis: technologies andrepresentations”, Social Research Online, Vol. 1 No. 1, available at:www.socresonline.org.uk/1/1/4.html
Coviello, N.E. and Jones, M.V. (2004), “Methodological issues in international entrepreneurshipresearch”, Journal of Business Venturing, Vol. 19 No. 4, pp. 485-508.
Craig, C.S. and Douglas, S.P. (2000), International Marketing Research, 2nd ed., John Wiley &Sons, Chichester.
Craig, C.S. and Douglas, S.P. (2001), “Conducting international marketing research in thetwenty-first century”, International Marketing Review, Vol. 18 No. 1, pp. 80-90.
Crawford, H.K., Leybourne, M.L. and Arnott, A. (2000), “How we ensured rigour in a multi-site,multi-discipline, multi-researcher study”, Forum: Qualitative Social Research, Vol. 1 No. 1,available at: www. qualitative-research. net/fqs-texte/1-00/ 01-00crawfordetal-e.pdf
Davenport, T.H. and Prusak, L. (1998), Wenn Ihr Unternehmen Wußte, Was Es Alles Weiß. DasPraxisbuch zum Wissensmanagement, (H. Hohlein, Trans.), Verlag Moderne Industrie,Landsberg/Lech.
de Geus, A.P. (1988), “Planning as learning”, Harvard Business Review, Vol. 66 No. 2, pp. 70-4.
de Ruyter, K. and Scholl, N. (1998), “Positioning qualitative market research: reflections fromtheory and practice”, Qualitative Market Research, Vol. 1 No. 1, pp. 7-14.
QMRIJ8,1
34
Denzin, N.K. and Lincoln, Y.S. (Eds) (1994), Handbook of Qualitative Research, Sage Publications,Thousand Oaks, CA.
Denzin, N.K. and Lincoln, Y.S. (Eds) (1998), Strategies of Qualitative Inquiry, Sage Publications,Thousand Oaks, CA.
Elder, J.W. (1976), “Comparative cross-national methodology”, Annual Review of Sociology, Vol. 2No. 1, pp. 209-30.
Evans, J. and Mavondo, F.T. (2002), “Psychic distance and organizational performance:an empirical examination of international retailing operations”, Journal of InternationalBusiness Studies, Vol. 33 No. 3, pp. 515-32.
Fahy, J. (1998), “Improving response rates in cross-cultural mail surveys”, Industrial MarketingManagement, Vol. 27 No. 6, pp. 459-67.
Firat, A.F. (1997), “Educator insights: globalization of fragmentation – a framework forunderstanding contemporary global markets”, Journal of International Marketing, Vol. 5No. 2, pp. 77-86.
Foss, N.J. and Pedersen, T. (2002), “Transferring knowledge in MNCs: the role of sources ofsubsidiary knowledge and organizational context”, Journal of International Management,Vol. 8 No. 1, pp. 49-67.
Ghauri, P.N. and Grønhaug, K. (2002), Research Methods in Business Studies: A Practical Guide,2nd ed., Financial Times Prentice-Hall, London.
Griggs, S. (1987), “Analysing qualitative data”, Market Research Society. Journal of the MarketResearch Society, Vol. 29 No. 1, p. 15.
Hallen, L. and Wiedersheim-Paul, F. (1979), “Psychic distance and buyer-seller interaction”,Organisajon, Marked og Samfund, Vol. 16 No. 5, pp. 308-24.
Harris, S. (2000), “Reconciling positive and interpretative international management research:a native category approach”, International Business Review, Vol. 9 No. 6, pp. 755-70.
Harzing, A.-W. (1997), “Response rates in international mail surveys: results of a 22-countrystudy”, International Business Review, Vol. 6 No. 6, pp. 641-65.
Harzing, A.-W. (2000), “Cross-national industrial mail surveys: why do response rates differbetween countries?”, Industrial Marketing Management, Vol. 29 No. 3, pp. 243-54.
Holzmuller, H.H. (1995), Konzeptionelle und Methodische Probleme in der InterkulturellenManagement- und Marketingforschung, Schaffer-Poeschel, Stuttgart.
Jain, S.C. (1993), International Marketing Management, 4th ed., Wadsworth Publishing, Belmont,CA.
Johanson, J. and Vahlne, J.-E. (2003), “Business relationship learning and commitment in theinternationalization process”, Journal of International Entrepreneurship, Vol. 1 No. 1,pp. 83-101.
Kelle, U. (1997), “Theory building in qualitative research and computer programs for themanagement of textual data”, Social Research Online, Vol. 2 No. 2, available at:www.socresonline.org.uk/2/2/1.html
Kim, D.H. (1993), “The link between individual and organizational learning”, Sloan ManagementReview, Vol. 35 No. 1, pp. 37-50.
Knight, G.A., Spreng, R.A. and Yaprak, A. (2003), “Cross-national development and validation ofan international business measurement scale: the coiscale”, International Business Review,Vol. 12 No. 5, pp. 581-99.
Kvale, S. (1996), Interviews: An Introduction to Qualitative Research Interviewing, SagePublications, Thousand Oaks, CA.
Analysingtextual data
35
Lee, R.M. and Fielding, N. (1991), Using Computers in Qualitative Research, Sage Publications,London.
Lee, T.W. (1999), Using Qualitative Methods in Organizational Research (Organizational ResearchMethods Series), Sage Publications, Thousand Oaks, CA.
Levitt, T. (1983), “The globalization of markets”, Harvard Business Review, Vol. 61 No. 3,pp. 92-102.
Livingston, G. (1994), “‘Fishing’ the quantitative pool yields qualitative insights”, MarketingNews, Vol. 28 No. 12, p. H29.
McAdam, R. and Reid, R. (2001), “SME and large organisation perceptions of knowledgemanagement: comparisons and contrasts”, Journal of Knowledge Management, Vol. 5No. 3, pp. 231-41.
McDonald, M.H.B. (1985), “Methodological problems associated with qualitative research: someobservations and a case analysis of international marketing planning”, InternationalStudies of Management & Organization, Vol. 15 No. 2, pp. 19-40.
Maclaran, P. and Catterall, M. (2002), “Analysing qualitative data: computer software and themarket research practitioner”, Qualitative Market Research: An International Journal,Vol. 5 No. 1, pp. 28-39.
Magnusson, J. and Nilsson, A. (2003), “To facilitate or intervene? A study of knowledgemanagement practice in SME networks”, Journal of Knowledge Management Practice,Vol. 4, available at: www.tlainc.com/articl48.htm.
Marshall, H. (2001), Rigour or Rigidity? The Role of CAQDAS in Qualitative Research,Association for Qualitative Research (AQR), Melbourne, 5-7 July, available at:www.latrobe.edu.au/aqr/offer/papers/HMarshal.htm
Marshall, H. (2002), “What do we do when we code data?”, Qualitative Research Journal, Vol. 2No. 1, pp. 56-70.
Maxwell, J.A. (1997), “Designing a qualitative study”, in Bickman, L. and Rog, D.J. (Eds),Handbook of Applied Social Research Methods, Sage Publications, London, pp. 69-100.
Melina, A. (1998), “Computer-assisted text analysis methodology in the social sciences”,Arbeitsbericht Nr. 97/07, Zentrum fur Umfragen, Methoden und Analysen (ZUMA),Mannheim..
Miles, M.B. (1979), “Qualitative data as an attractive nuisance: the problem of analysis”,Administrative Science Quarterly, Vol. 24 No. 4, pp. 590-601.
Miles, M.B. and Huberman, A.M. (1994), Qualitative Data Analysis: An Expanded Sourcebook,Sage Publications, Thousand Oaks, CA.
Mudambi, R. (2002), “Knowledge management in multinational firms”, Journal of InternationalManagement, Vol. 8 No. 1, pp. 1-9.
Mullen, M.R. (1995), “Diagnosing measurement equivalence in cross-national research”, Journalof International Business Studies, Vol. 26 No. 3, pp. 573-96.
Nasif, E.G., Al-Daeaj, H., Ebrahimi, B. and Thibodeaux, M.S. (1991), “Methodological problems incross-cultural research: an updated review”, Management International Review, Vol. 31No. 1, pp. 79-91.
O’Grady, S. and Lane, H.W. (1996), “The psychic distance paradox”, Journal of InternationalBusiness Studies, Vol. 27 No. 2, pp. 309-33.
Padmanabhan, P. and Cho, K.R. (1995), “Methodological issues in international business studies:the case of foreign establishment mode decisions by multinational firms”, InternationalBusiness Review, Vol. 4 No. 1, pp. 55-73.
QMRIJ8,1
36
Parameswaran, R. and Yaprak, A. (1987), “A cross-national comparison of consumer researchmeasures”, Journal of International Business Studies, Vol. 18 No. 1, pp. 35-49.
Perlmutter, H.V. (1969), “The tortuous evolution of the multinational corporation”, ColumbiaJournal of World Business, Vol. 4 No. 1, pp. 9-18.
Pierce, B. and Garven, G. (1995), “Publishing international business research: a survey of leadingjournals”, Journal of International Business Studies, Vol. 26 No. 1, pp. 69-89.
Pike, K.L. (1966), Language in Relation to a Unified Theory of the Structure of Human Behavior,Mouton, The Hague.
Prahalad, C.K. and Hamel, G. (1990), “The core competence of the corporation”, Harvard BusinessReview, Vol. 68 No. 3, pp. 79-91.
Richards, L. (2000), Using N*vivo in Qualitative Research, 2nd ed., QSR International Pty,Bundoora.
Rubin, H.J. (1995), Qualitative Interviewing: The Art of Hearing Data, Sage Publications,Thousand Oaks, CA.
Ruyter, K.D. and Scholl, N. (1998), “Positioning qualitative market research: reflections fromtheory and practice”, Qualitative Market Research: An International Journal, Vol. 1 No. 1,pp. 7-14.
Salzberger, T., Sinkovics, R.R. and Schlegelmilch, B.B. (1999), “Data equivalence in cross-culturalresearch: a comparison of classical test theory and latent trait theory based approaches”,Australasian Marketing Journal, Vol. 7 No. 2, pp. 23-38.
Samiee, S. and Jeong, I. (1994), “Cross-cultural research in advertising: an assessment ofmethodologies”, Journal of the Academy of Marketing Science, Vol. 22 No. 3, pp. 205-17.
Schlegelmilch, B.B. and Diamantopoulos, A. (1991), “Prenotification and mail survey responserates: a quantitative integration of the literature”, Journal of the Market Research Society,Vol. 33 No. 3, pp. 243-55.
Sekaran, U. (1983), “Methodological and theoretical issues and advancements in cross-culturalresearch”, Journal of International Business Studies, Vol. 14 No. 2, pp. 61-73.
Sieber, S.D. (1973), “The integration of fieldwork and survey methods”, American Journal ofSociology, Vol. 78 No. 6, pp. 1335-59.
Simon, H. (1994), “Marketing science’s pilgrimage to the ivory tower”, in Laurent, G., Lilien, G.L.and Pras, B. (Eds), Research Traditions in Marketing, Kluwer Academic Publishers,Dordrecht, pp. 27-43.
Sinkovics, R.R. and Holzmuller, H.H. (2001), “National differences in materialism – usingalternative research strategies to explore the construct”, Journal of International ConsumerMarketing, Vol. 13 No. 2, pp. 103-34.
Stata, R. (1990), “Organizational learning: the key to management innovation”, SloanManagement Review, Vol. 30 No. 3, pp. 63-74.
Strauss, A.L. and Corbin, J.M. (1998a), Basics of Qualitative Research: Grounded TheoryProcedures and Techniques, Sage Publications, Thousand Oaks, CA.
Strauss, A.L. and Corbin, J.M. (1998b), Basics of Qualitative Research: Techniques andProcedures for Developing Ground Theory, 2nd ed., Sage Publications, Thousand Oaks,CA.
Sykes, W. (1990), “Validity and reliability in qualitative market research: a review of theliterature”, Journal of the Market Research Society, Vol. 32 No. 3, pp. 289-338.
Titscher, S. (2000), Methods of Text and Discourse Analysis, Sage Publications, London.
Usunier, J.-C. (1999), Marketing across Cultures, 2nd ed., Prentice-Hall, New York, NY.
Analysingtextual data
37
Weitzman, E.A. and Miles, M.B. (1995), Computer Programs for Qualitative Analysis, SagePublications, Thousand Oaks, CA.
Young, S., Dimitratos, P. and Dana, L.-P. (2003), “International entrepreneurship research: whatscope for international business theories?”, Journal of International Entrepreneurship,Vol. 1 No. 1, pp. 31-42.
Zander, U. (2002), “When Muhammed goes to the mountain: globalization, cathedrals ofmodernity and a new world order”, in Havila, V., Forsgren, M. and Hakansson, H. (Eds),Critical Research on Multinational Corporations, Pergamon Press, Oxford, pp. 153-77.
Zimmerman, A.S. and Szenberg, M. (2000), “Implementing international qualitative research:techniques and obstacles”, Qualitative Market Research: An International Journal, Vol. 3No. 3, pp. 158-64.
QMRIJ8,1
38