3. Research Design
3.1 Research Paradigm
3.1.1 Qualitative Strategy
This chapter seeks to satisfy objective 2 (section 1.3.1), offering an original approach to the investigation of auto-ID innovation. A qualitative research strategy was chosen over a quantitative one due to the complexity of the research problem. The study grants the qualitative researcher the ability to focus in on a grand tour question throughout the thesis, followed by subquestions within (Anderson & Kanuka 2003, p. 35). In this instance the investigation is mainly preoccupied with the auto-ID trajectory. This type of strategy allows the researcher unlimited inquiry in areas that he/she believes is required, finding synchronicity with the systems of innovation (SI) framework underlying the study. Intrinsic to this type of research, as stated by Creswell (1998, pp. 16-17), is a commitment to collect extensive data. This researcher has chosen to conduct data analysis using categories or like themes to make sense of the large amount of data collected. In this thesis, long quoted passages also characterise the style of presentation, not only to substantiate claims made by the researcher but also to show the diverse positions held by the various auto-ID stakeholders, pointing to key pieces of evidence throughout.
3.2 Research Design
3.2.1 The Architecture
The importance of the research design cannot be overestimated. Ultimately the findings to be drawn from a thesis are dependent upon how well the research design has been constructed to respond to the central question. Nachmias and Nachmias (1976, pp. 77f) describe the research design as a plan that “guides the investigator in the process of collecting, analysing, and interpreting observations.” The naturalistic design adopted for this study has been developed specifically for the auto-ID industry being investigated. The way in which the study was composed and the chapters were organised, illustrates all the different aspects and perspectives that needed to be considered. The narrative reporting approach taken by the researcher lends itself well to this type of exploratory investigation. The process of research is not linear but rather a spiral one, where telling the story is not as straightforward as it might initially seem. All the different pieces of the puzzle need to be brought together to uncover the underlying message(s). A typical research design structure prevalent in quantitative research would have failed to present the whole story. Instead, an innovative and original approach to investigation based on a narrative reporting style (which takes into account the complexity of the questions being posed) was adopted. On first glance what may be seemingly a rather unstructured piece of work, is an intricately interwoven discussion, consciously devised to explore and understand the history, background, development and future directions of auto-ID.
Although the investigation predominantly focuses on auto-ID case studies, there is an element of historical analysis that pervades the whole thesis. The systems of innovation framework emphasises the important role of time and for this reason it has been explicitly included in the research design (see diagram 3.1 on the following page). Given that this investigation is seeking to determine the auto-ID trajectory, it is considered that past events can help to shed light on present and future trends (see section 2.7). Even the case studies are introduced in a chronological sequence. Only in this manner can one trace auto-ID innovation- from its beginnings till today- proof being gathered along the historical journey, as a way to predict the future possibilities. The exploratory nature of the research is combined at different stages with descriptive, developmental, historical and interpretive analysis. The historical nature of the research is especially reflected in chapter four when manual to automatic identification is traced. Chapter five then relies on descriptive analysis to systematically set a context for the research at large. The dimensions of auto-ID innovation are analysed in chapter six using the systems of innovation framework and embedded cases complete the case studies in chapter seven. Following is the chapter on the auto-ID trajectory where patterns are identified over a function of time. Finally, the findings and conclusion is where an interpretive style of analysis takes place.
3.2.2 The Narrative Approach
While theoretical concepts in the field of innovation underlie the thesis throughout, it is predominantly practice-driven, written for a wide-ranging audience. Apart from the Literature Review where some twenty pages are dedicated to innovation theory, the chapters are more concerned with the implementation of that theory as a framework rather than trying to make hypotheses or theoretical generalisations throughout. The approach used here is similar to that work found in Flyvbjerg (1998) and Henriksen (2002). It is in the very end of the thesis (chapters 9 and 10) that the reader can comprehend the “bigger picture”, where the major themes are drawn out and discussed in totality. Prior to that point in the thesis, thick descriptions can be found in order to gather the “facts” and provide evidence for the final conclusions. In chapters four to eight some analysis is presented but in limited detail. The reader is guided in this way and can make use of the analysis to interpret the major points that are being put forward. The fundamental building blocks are laid, step-by-step, and the reader is taken on an exploration of discovery. The role of this researcher, as narrator, is a guiding force, present and personal to some degree, but granting the reader the freedom to traverse once the evidence has been provided.
The style of presentation that has been chosen for this thesis is one that is mindful of the diverse audience that will take interest in the findings of the research. Various chapters will appeal more, or less, to specific auto-ID stakeholders. Chapters 5-7 for instance, have been written specifically for the central actor- the auto-ID companies, yet auto-ID users will also gain significantly by reading and understanding what makes auto-ID companies innovate. Similarly, auto-ID company representatives (i.e., both technology and service providers) may not care too much about the historical development of auto-ID but chapter 4 may add another perspective to their thinking when creating and introducing new products to the mass market and how they may be ultimately used. Finally, it is chapters 8 and 9 that bring all the different types of stakeholders closer together, bridging the divide, and offering a fusion of a variety of perspectives relevant to the whole audience, independent of their position or membership to a group.
In writing for such a diverse audience, the language used has had to appeal to both the academic and non-academic reader. The researcher has purposefully used a level of English that is comprehensible to the masses, without compromising on the academic nature of the research. The language is generally easy to understand and makes use of different literary devices to install reader confidence in the work, including extensive footnoting, referencing, and detailed description. What the reader can expect are long passages of description and a narrative that assists to make sense of the plethora of data gathered, as can be seen from the extensive bibliography and online resources. The writing style is not that of a distant narrator but has an almost personal and familiar quality about it. Creswell (1998, p. 170) calls this an “up-close” approach to writing.
If there is one element of the work that really influences the narrative style, it is the level of detail the researcher goes into. On any given page the reader is given the opportunity to follow up either the original source or a related source of reference. This is different to other auto-ID works, some of which lay claim to ideas, without the adequate proof. The number of footnotes included in each chapter, for instance, grants the reader the ability to further investigate a given sub-topic which is in some way relevant to the larger question. It is this level of detail that makes this work stand out and leave its mark among other auto-ID references. Long passages of block quotes can be found throughout the thesis, granting the reader a finer understanding of the complex ideas being presented. Embedded quotes within the body are used to ensure that misrepresentation of other peoples’ ideas does not occur and to add weight to what is being conveyed. In addition, short quotes have the affect of attracting the reader’s attention to fundamental events, products, happenings and ideas that are of importance.
It follows that the method of investigation is fundamental to the design of any enquiry. As evidence suggests from previous innovation research, a case study methodology is more appropriate than the application of an experiment, survey, or interview. In reviewing literature in chapter two it was identified that the vast majority of innovation research based on evolutionary theory or the systems of innovation approach (SI) opted for the case study method. For instance, Sharp (1985) successfully used case study to investigate six high technologies (see section 2.4.1). Studies in product innovation that take the form of experiments or surveys are usually (but not always) related to consumer attitudes, sales forecasts and potential market adoption. While diffusion is considered important in this thesis, it is in the context of how it is linked to innovation. Thus, this thesis will not seek to forecast exactly how many auto-ID devices consumers will be using by 2005 but how auto-ID innovation occurs, what the auto-ID selection environment is, and what auto-ID trends are likely. Though the case study method has traditionally been considered the ‘softer’ option and has received much criticism regarding investigator carelessness, bias and the compilation of unreadable documentation, researchers continue to use the strategy successfully (Yin 1984, p. 21). The strengths of a case study method are numerous and while it does possess some weaknesses, other research methods do also. It is the responsibility of the researcher to be actively aware of these pitfalls.
3.3.1 Multiple Embedded Case Studies
As the central question explicitly implies, the case studies are all directed toward technological innovations in the auto-ID industry. The multiple embedded case study is regarded by methodology experts (Yin 1984, p. 47) as being the best in dealing with new and complex innovations, especially advanced technologies. Rather than choosing technologies from different industries (which would make comparison difficult due to the qualitative nature of this thesis) it is focussed only on the auto-ID industry (see diagram 3.2 on the following page). This is also in accordance with the systems of innovation (SI) framework at the technological systems (TS) level. The main unit of analysis is the technology (e.g. smart cards). The sub-unit of analysis is the application of the technology (e.g. telecommunications). At the first level the technology is important; at the second level the application is considered independently and then related back to the main unit of analysis. The sub-units are significant in that they offer an opportunity for more extensive analysis. Additionally, a multiple case study simply on the various auto-ID technologies would be in direct conflict with the definition given of innovation in section 1.1.4. Auto-ID technologies can only be considered innovations when they are applied to a service that can be utilised by consumers, business or government. In this sense, the sub-unit analysis is an essential part of the thesis.
The results of a multiple embedded case study are generally more compelling than that of a plain single or multiple case study design. Yin (1984, p. 48) notes that it has been particularly used by many researchers in the field of innovation. Each auto-ID technology considered in this study was carefully chosen based on its level of technical development and market maturity. The choice of cases came from the need to implement two very important qualities in the case studies, that of maximum variation and stratification with a purpose. Maximum variation ensured that different types of auto-ID devices were chosen for the study, for example, not only card devices were investigated. This aids in identifying common patterns across the auto-ID industry sector, not just isolating the card subgroup. The stratification notion facilitates comparison between auto-ID technologies by acknowledging that each auto-ID device did initially exist in an isolated subgroup before the broader industry was formed.
As has already been mentioned, the order in which the cases are presented is chronological in terms of the way one technology has impacted on the innovation and diffusion of subsequent technologies. This historical perspective helped to draw out the pattern of technical change that occurred in auto-ID since its commercial introduction. Coincidentally, the chronological manner in which the devices have been presented has also corresponded to a growing level of technological invasiveness (see diagram 3.3 below). From bar codes attached to non-living things; to magnetic-stripe cards and smart cards carried by humans; to the biometrics of humans; to RF/ID tags and transponders implanted in animals. The researcher describes this development as the “human evolution”. Auto-ID was initially developed to identify packaged goods at the checkout counter, now it is being used increasingly to monitor and track animals and humans.
The sub-unit of analysis has also been considered carefully. Two applications (i.e. the sub-unit level) are chosen for each technology (i.e. unit) based on the successful combination of technology and application (i.e. potential for widespread diffusion). For example, the smart card is well-known for its applicability in telecommunications more than it is known for its use in retail and entertainment services. On the other hand retail has become synonymous with bar code technology and financial services with magnetic-stripe card technology. As there were five auto-ID technologies chosen for the research, a maximum of ten electronic commerce (EC) applications were potentially feasible within the scope of this study. The ten applications cover a wide variety of vertical sectors specifically to address the growing pervasiveness of auto-ID technology. Within each application area (e.g. telecommunications) there can be literally thousands of product innovations, e.g., pre-paid telephone cards, subscriber identity module (SIM) cards, virtual private network (VPN) cards, cable television (CATV) cards. Several innovations are documented within each sub-unit. The number of different product innovations discussed varies dependent on the available literature and space constraints. It should also be pointed out that the innovation system studied is ‘supranational’ (i.e. global), concentrating on the technological system rather than the geographical dimension.
22.214.171.124 Literal Replication
While not apparent in the body of this thesis, a pilot study was conducted before a research effort was committed to five auto-ID technologies and ten embedded cases. Initially, a single embedded case study was chosen, that of smart cards and the applications of government and financial services. Having conducted a literature review and collected preliminary data, it became obvious to the researcher that it would be very difficult to show such ideas as selection environment and technological trajectory without looking at a larger innovation system. The single embedded case study in this instance revealed itself to be inadequate. The SI framework also called for the investigation of more than one technology. The smart card case study was therefore included along with another four technologies that could assist in a better understanding of the auto-ID trajectory. Additionally, more applications were chosen to complement the supplementary technologies being researched.
The number of case studies may seem too many to some readers and beyond the scope of this thesis. Why not two or three or four? What significance do exactly five case studies have? Some observers may consider the large number of studies to correspond to a sample such as that in a survey but this is not the case at all. To do this would be to disregard the basic principles of the case study method altogether. The motivation behind choosing five case studies was two-fold. First, it is the speculation of this researcher that results from each case study may be comparably similar given the application of the SI conceptual framework. Literal replication is evident in the cross-case findings documented by themes in chapters five to seven (see diagram 3.4 on the following page). Second, in undertaking to investigate the evolutionary paradigm of auto-ID, five case studies were deemed appropriate to ensure the overall success of the research findings. Far better a more complete list of technologies than one that was restrictive and limited, inhibiting results.
Having discussed the reasons for including five auto-ID technologies, the sub-unit must also similarly be considered. As mentioned there are ten embedded cases, covering a large spectrum of services from the perspective of consumer, business and government product innovations. The focus is primarily on business-to-consumer and government-to-consumer EC interactions. The question of choice and discretion is also applicable to the sub-unit. What was the motivation for including all those different application areas? Apart from presenting the pervasiveness of the technology itself (here the link is made between sub-unit and unit), the application is treated on its own merits as well. Though subtle, the underlying message is important, how is technology both now and in the future going to impact humans? For instance, how will auto-ID affect social and working behaviours (Hewitt 1993; Zuboff 1988)? The technological visionary, Nicholas Negroponte, (1995, p. 6) puts it well when he writes that “[c]omputing is not about computers anymore. It is about living.” As Negroponte predicted in 1995 (p. 231), “...each generation will become more digital than the preceding one.” Assuming Negroponte is right, and the way in which humans interact with one another is changing as a direct result of the impact of technology, then the application of that technology is equally relevant. It is the application that grants the technology life commercially, when a use for it is either ‘pushed’ by the manufacturer, or ‘pulled’ by the consumer. Of interest is also the increasing interconnectedness of applications that is being enabled by auto-ID devices. Never before has there been, both the capability and the potential to interlink records from different databases with such ease.
126.96.36.199 Case Study Protocol
Having defined the units of analysis above, it is now necessary to identify each of the questions to be explored in the case studies (see diagram 3.5). Each case study will contain information about the auto-ID technology itself; two examples of how the technology is applied commercially; a number of product and process innovations; factors influencing the innovation of that technology; and an exploration of the selection environment and trajectory of the technology. Due to the types of evidence used in the thesis (see 3.4.3), there was no requirement for documenting field procedures for interviews or checking for the availability of interviewees or resource requirements for surveys, etc. It is important to note here that it would have been easier to dedicate
an individual chapter to each technology but the intent was to draw out those important themes and patterns prevalent to auto-ID as a technology system (TS). Thus the chapters take the form of relevant themes, making comparison a manageable task both for the investigator and reader. Previous studies have isolated individual technologies as being separate from one another- this thesis draws the devices closer together. The level of investigation that will be undertaken is three-fold (see table 3.1 below). At the first level, the case studies (ch. 5-7) will capture the necessary patterns and trends that will aid in the prediction of the auto-ID trajectory (ch. 8). At the next level, the findings (ch. 9) will draw out the answers to those fundamental questions the thesis sought to investigate, and at the final level the conclusion (ch. 10) will offer a context of interpretation for the overall study.
Table 3.1 Levels of Investigation
Level | Explanation
1 | Case Studies --> Help to Predict Auto-ID Trajectory See diagram 3.4
2 | Findings --> So What do the Cases Actually Show?
Is the auto-ID industry a technology system (TS)?
- Do auto-ID technologies share in a common technological trajectory?
What is the typical auto-ID innovation process?
- What are the dimensions affecting the innovation of auto-ID technology?
Will one technology dominate or will several technologies co-exist?
- Are technologies undergoing a process of migration, integration or convergence?
How are auto-ID technologies impacting applications?
- What does the future hold for auto-ID?
What are the implications of auto-ID pervasiveness?
3 | Conclusion --> So What do the Findings Actually Mean and for Whom?
Are auto-ID technologies part of the evolutionary paradigm?
What are the overall implications of the findings of this thesis?
188.8.131.52 External Validity
The findings of this thesis can be generalised specifically within the domain of the auto-ID industry. To a lesser extent these findings are also applicable to the information technology sector (within which auto-ID innovations belong). At this level it is important for professionals associated with the industry to have a systems view of the dimensions affecting auto-ID devices. Often, one or two factors are studied in isolation without the awareness that industry-wide issues are much more complicated. Relationships between factors are also important. For instance, while it is technically feasible to introduce an incremental change to an auto-ID innovation, is it socially acceptable or cost-effective to do so? Industry trends can also be monitored when looking at the evolution of a technology. For instance, is it possible that one auto-ID technology will prevail over others as a dominant design?
Finally, this thesis has wider implications for the general public. At the speed with which auto-ID devices are being introduced and the new applications which require these devices, every individual will potentially be affected by the technology. While it is true that “[t]he better prepared we are to accept change, the less terrifying those changes will be” (Goodman 1995, p. 217), technical change is much more complex an issue. Yes, users must become aware of how auto-ID technologies are likely to shape their future but for the purpose of making educated decisions about the suitability of particular innovations, not to heedlessly accept every change for the sake of change. Thus, any person associated with the auto-ID industry- whether they are an engineer, manufacturer, regulator, legislator, businessman or potential/current user- will find this study extremely useful. All these persons are empowered to make decisions about tomorrow but often only have a view of their immediate dimension, be it technical for the engineer and manufacturer or economic for the businessman.
3.4 Data Collection
3.4.1 Systems of Innovation Bounds
The qualitative research strategy chosen, the innovative research design spearheaded by the narrative approach, and the case study methodology adopted have all been mindful of the underlying systems of innovation (SI) conceptual framework. In this thesis, SI is not just an “add-on” framework but is intrinsic to the qualitative paradigm that has been applied. SI provides a holistic picture of the auto-ID technology system (TS). Qualitative research expert John Creswell (1998, p. 15) emphasises the requirement for exploratory research to possess this “holistic picture”. He believes this whole view “…takes the reader into the multiple dimensions of a problem or issue and displays it in all of its complexity” (Creswell 1998, p. 15). This very quality is what SI espouses as well. In fact, what attracts researchers to SI is that it is indeed a “holistic and interdisciplinary” approach which “encompasses all or most determinants of innovation” (Edquist et al. 1998, p. 20). Diagram 3.6 below illustrates the SI holistic approach to investigation. The Systems of Innovation framework “…allows for the inclusion not only of economic factors influencing innovation but also of institutional, organisational, social and political factors” (Edquist 1997, p. 17). Creswell (1998, p. 13) himself believes that it is the application of appropriate frameworks that “hold qualitative research together”. He uses the following metaphor to convey the importance of frameworks:
I think metaphorically of qualitative research as an intricate fabric composed of minute threads, many colours, different textures, and various blends of material. The fabric is not explained easily or simply. Like the loom on which the fabric is woven, general frameworks hold qualitative research together.
In diagram 3.6, each innovation dimension possesses one or more, direct or indirect relationship(s). However, the researcher is empowered with the decision of which dimensions and relationships to address in the study. All of these may not be relevant- one case study may include all the dimensions another may focus on several. The SI approach grants the researcher freedom to explore within predefined boundaries. This is advantageous considering that a major criticism of many case studies is that they are too vague and boundary-less. But there is also a fine balance that needs to be attained in the presentation of a case study; it should not be so contrived that it impedes exploration but free flowing so that all the evidence that is needed can be gathered.
The available SI literature helps the researcher to seek out specific references and sources that are relevant to the dimensions of innovation (as stated in diagram 3.6), for instance, what is meant by the factor “organisational” as opposed to “institutional” and “economic”. While SI proponents do not rigidly state a definition for each of the factors they do guide other researchers by the examples they put forward (see Edquist 1997). It should be noted, that dependent on the context of analysis, some of the key terms to describe the determinants can also be used interchangeably. For example, some researchers may describe the organisational and institutional factors in one and the same breath (even though some SI researchers clearly state their difference). This researcher conducted a literature review on SI and noted key words that described each of the determinants. She then continued to search for these key words in auto-ID specific literature and draw out any possible relationships that could be made. For example, the key words searched for three of the most important SI dimensions considered in this thesis as related to the auto-ID industry are included below:
Organisational: public organisations, policy, political bodies, regulatory agencies, organisations for higher education, technology support entities (e.g. training), patent offices, standards setting organisations, consulting agencies, knowledge production, universities, organisations with formal structures, explicit purpose, players or actors, other firms
Institutional: norms, habits, practices, routines, laws, interaction, often no specific purpose, form spontaneously, relations between groups, research and development links, consumer reactions, conflicts and cooperation, reduction in uncertainty, technical standards, rules of the game, framework conditions
Economic: infrastructure, physical infrastructure, knowledge infrastructure, standards, formal knowledge, tacit knowledge, explicit knowledge, research councils, standard setting organisations, libraries, databases, skilled/technical personnel, routine, industry associations, conferences, training centres, trade publications, research laboratories, public agencies.
Chapters 6 and 7 are where the SI approach was most prevalent in the analysis and presentation of material on auto-ID innovation. In chapter 6 whole paragraphs were dedicated to discussing the factors in each of the five case studies. Headings such as “Committees, Subcommittees and Councils”, “Public Policy”, “Clusters of Knowledge and a Growing Infrastructure”, “Setting Standards”, “A Patchwork of Statutes” and so forth, can be found to reflect direct SI concepts. In addition, chapter 7 utilises key terms from evolutionary theory to discuss the selection environment of the five auto-ID devices being studied.
Inherent to SI, given its holistic quality, is also the idea of interdisciplinarity. Given the researchers emphasis on different perspectives, this notion of bringing more than one discipline together, is also appropriate and will cater to the needs of the diverse audience this research targets. It also makes the findings of the thesis more robust and lessens the potential for researcher bias. The SI conceptual framework is also a best-fit for non-linear types of research like this exploratory investigation because it places great importance on the feedback mechanism of the players in the innovation system itself. Feedback can only take place over time as it requires a two-way interaction and SI values this exchange between stakeholders. In describing this interaction, as has been done in this thesis, the meshed research strategy is entirely appropriate. To this end, SI provides the following benefits: i) it provides a boundary for the researcher to work within, ii) it specifies the important factors that should be considered in the research giving the researcher the freedom to include or exclude determinants based on the context and case, and iii) it encourages the use of case studies, the narrative approach and innovative research designs.
3.4.2 Construct Validity
Two main sources of evidence will be used in this study, documentation and archival records. The scope of the thesis does not allow for the collection of data from interviews. However, this study can be used to identify potential interview candidates for future studies. Initially, there was a choice of two paths this researcher could have embarked on; either to cover a broad range of auto-ID technologies and sacrifice the level of detail in each case or choose one auto-ID technology and go into intricate detail. The former was regarded as more important, considering the lack of material on the question of auto-ID and innovation, and the importance placed on the concept of selection environment. Additionally, while most methodologists (Robson 1993, p. 274) do encourage supplementary methods of evidence, they do not preclude carrying out a study based solely on one source of evidence, for instance, primary or secondary qualitative sources. And since this thesis also incorporates archival records, construct validity is present. Some advantages include: the researcher can observe changes at his/her own pace and own time and the data is in a permanent form making re-analysis possible.
3.4.3 Multiple Sources of Evidence
Traditionally books in the field of auto-ID contain a brief historical introduction about one or more technologies and give examples of applications without going into too much depth. Auto-ID books show static representations of technologies at a given point in time, however, they are useful in that they make the researcher aware of the incremental changes that have occurred over the years. They also familiarise the researcher with the more important auto-ID definitions and concepts and raise some very important issues. General computing or engineering journal articles and reports have much the same function as books but with the advantage that they are more frequent and up-to-date pieces of research. Auto-ID articles are also able to focus on aspects of the technology and are usually written by experts who have had professional experience in the field. There are also the specific auto-ID journals and magazines most of which are published by SJB services which made their debut in the mid-90s (see table 3.2 on the following page). These are excellent and reliable sources of evidence that include relevant industry contact names and telephone numbers for further investigation.
Conference proceedings are also particularly useful in a thesis such as this that is exploratory and requires empirical evidence to justify its findings. A researcher can expect to find in conference proceedings, information about the newest auto-ID innovations. Leading edge case studies and surveys are usually compiled by consultants who are at the forefront of the industry and have had real-life experiences implementing auto-ID solutions. Press releases are also crucial- though brief they are a good way of tracking developments in specific product innovations throughout the year. One criticism of press releases is that they are sometimes written by marketing employees who have the interests of the corporation at heart. Nevertheless they do indicate changes that are occurring in auto-ID. Newspaper articles about auto-ID are usually not technical in nature and are often written by reporters who do not have experience in the field. However, they do act to raise issues that are not dealt with in mainstream journals and magazines. They have predominantly reported on the social implications of the technology with a view to capturing the interests and imaginations of readers. However, while one must be careful to separate sensational material from scientific fact, it does not mean that popular material cannot be used in an investigation such as this. Often these articles are surprisingly up-to-date and offer different perspectives than would otherwise be found in scientific journals. The popular media- newspapers, radio, television- have long been used as an open forum to gauge political and social responses to technological developments before they are actually introduced.
184.108.40.206 Archival Records
The archival records used in this thesis are in the form of electronic information sources, publicly available on the Internet. Web site information is a newfound source of evidence that many researchers have simply ignored in the past. In this thesis auto-ID web sites have been chosen, after being identified in the documentation collected or by querying a variety of search engines. These web sites include: auto-ID companies such as manufacturers, value-added resellers (VARs) and system integrators (see diagram 3.7 on the following page); customers such as private organisations; auto-ID research centres such as universities; service providers; and auto-ID-related associations. By visiting these web sites, further electronic links to other sites containing valuable information were also found. Company web sites are very informative containing a lot of information that cannot be found elsewhere. Private companies, organisations, even government agencies are now placing internal articles, product technical specifications, marketing brochures, whitepapers, press releases and other auto-ID information on the Internet for wider readership and greater accessibility to employees and customers. It is estimated that approximately 1000 web sites were visited by the researcher over a period of 4 years. At least 150 of these web sites were exclusive auto-ID technology providers.
The researcher used a variety of means to identify relevant Internet sites. Initially generic searches were conducted based on the key words defined by major categories, such as “bar code”, “magnetic-stripe card” etc (see section 3.5.1 for more details on categories). A variety of popular search engines were used, such as Yahoo! and Google. The hits returned were then examined for relevance and reliability. The researcher periodically performed these searches and downloaded files in text format, HTML, Microsoft Word and Powerpoint and PDF, storing them in digital folders with meaningful names. With each periodic search performed, categories were refined and new subcategories defined. The key words used to search became more specific with each iteration, and instead of thousands of web links being returned, a few hundred would result. The guidelines for which web pages were included in the data collection were not inflexible however there were several overriding controls the researcher used. First there had to be an author of the web page or site, whether this was an individual, group or company. Second the content of the web page had to be accurate. Accuracy was established by cross-referencing similar web pages or documentation on the exact same topic(s). Third, a date on the material viewed also added weight to what was being conveyed as well as a date for the period the web site was last updated. As can be seen from the online resources section of the thesis, not all the web pages and web sites cited had a date of last update. In these instances, the researcher used her judgement to include or exclude an online source. As much as was possible, the researcher attempted to verify the authority of documents where dubious content or sourcing was suspected. Apart from the exhaustive searches that were performed on the major categories, auto-ID company marketing lists were publicly obtained from magazines such as the Automatic ID News and individual company web sites searched and further recommended links on these sites followed through. It soon becomes apparent to the researcher which sites are those considered significant to the industry (e.g. AIM Global), as numerous companies reference the same link on their respective web sites. In addition, the experienced Internet researcher who has spent thousands of hours searching for relevant online material is able to quickly discern between web content that is of value to their investigation and that which should be set aside.
3.5 Data Analysis
3.5.1 The Data Management Process
As shown above, multiple sources of evidence were used to collect data for the case studies, including documents and archival records with an emphasis on e-Research. Individual pieces of data collected were first accessed and if not in hardcopy format, printed so that a permanent record was obtained in case of digital loss. The printed matter was sorted into relevant categories and sub-categories and after some time was organized into ring-bound file folders containing like themes. It should be noted that the collection of data over the five year period in which the thesis was written was highly iterative. After each major data collection effort, piles and piles of data would be clumped together for sifting when time permitted and it made sense to place things together. Dependent on that particular cluster of data gathered, file folder sizes varied accordingly. An index was also kept and file folders were colour coded according to their relevance to the thesis topic. Files with the same colour were placed side-by-side for easy access on a large bookshelf. Computer files were also kept wherever convenient. Pictures of physical artifacts were especially scanned as they were highly valued by the researcher as tangible proof for the auto-ID trajectory.
The management of data was done mostly in hard copy format, save for the extensive bibliography and online resources that were maintained in digital format to ensure easy searching by author, date, or title. Keyword searches in the titles of entries- a form of first-level content analysis- was used to locate relevant articles. A great deal of notetaking was done on articles or studies that were considered landmark to this research. Initially these notes were scattered by author and title electronically but with each reiteration of refinement for analysis, data was placed meaningfully into categories in dedicated electronic files. A lot of thought went into categorization. First data was placed within the five major categories depicting the case studies, therein within specific vertical sectors indicating product innovations, and furthermore within specific factors of innovation as described by the SI framework. At this point the focus was on describing each individual case and its respective context within the auto-ID industry. The diagram below shows this continual process of refinement.
3.5.2 Toward Naturalistic Generalisations
Initially the narrative was divided into five chapter cases, e.g., “Bar Code”, “Magnetic-Stripe Card” etc to allow for the raw notes made by the researcher to be placed into some early draft form denoting a case study. These notes were then organized and ordered in relevance to draw out the most important issues pertaining to the given device. After continual refinement and addition to these notes, it became apparent that several overriding patterns and themes were emerging from the embedded cases, pointing to a meaningful way to respond to the thesis, “the auto-ID trajectory”. This was a result of the thorough categorical aggregation performed earlier. A cross-section of information was then taken from each of the original narratives on the cases and used to create new chapters that were based on themes which would allow for interpretation and the development of naturalistic generalisations. Such themes included: the evolution of auto-ID technology, the selection environment of auto-ID and the dynamics of auto-ID innovation. Throughout the narrative the reader will find that patterns and themes and relevant issues are depicted using tables, figures, diagrams, exhibits and other visualizations such as timelines, picture collages and displays. These techniques are very important to the overall impact of the final work ensuring that in the end the “facts” are represented.
3.5.3 Content Analysis
According to Anderson and Kanuka (2003, p. 173), “[c]ontent analysis is a research technique that does not easily fall into either qualitative or quantitative classification schemas that researchers love to fight over. It is a crossover technique that requires critical qualitative skills to assign content to any number of variables…” In this thesis however there is no cross-over. The content-analysis performed by the researcher was not rigid at all. While the formalisation of categories was given a great deal of thought to ensure that the right type of information was presented, the researcher employed a more informal method of indexing. This allowed for the iterative analysis of material. If over time, due to new developments in the industry, some categories were found to intersect within articles newly obtained by the researcher, an additional category was formed with ease. In these instances, two or more themes were found to overlap to make a new cluster. It must be stated that the categorisation technique applied to the data analysis was more about identifying complex themes than about counting the number of articles with the same key word.
While the content analysis part of this research method was used to ensure that the most relevant sources were being used in the narrative, it was the coding process, which allowed for content analysis to happen. Before being able to interpret the content the researcher first needs to be able to identify and assign a category to the data being scrutinised. This initial step of categorisation in this thesis served as the most important part of forming the direction of the narrative. Initially simplistic categories like “biometrics” were identified. However, as the research effort continued and more complex issues began to emerge, more specific categories came into fruition. For example, the important issue of “proprietary versus open standards”, in the discussion about the emerging BioAPI defacto standard (see section 220.127.116.11). With each new data collection phase and data analysis iteration the researcher was led on a more exacting path, a path that would inevitably be able to lead the researcher to an understanding of the auto-ID trajectory. Metaphorically this can be likened to an explorer who first sets out to reach a destination by taking a certain route and then during the actual exploration realises that a more appropriate course can and should be taken given current environmental factors.
18.104.22.168 Pictures as Content
When researchers speak of content analysis they are most likely referring to the practice of textual analysis using a software program like Leximancer that usually returns quantitative results based on a certain number of parameters. The use of pictures and collages in this thesis are a major contribution not only to auto-ID development but to the methodological process per se. The researcher found the graphical representation, using pictures of auto-ID artefacts, the most direct way which she could convey the trends in the industry. The pictures and collages are thought-provoking and at times the reader may even be left awestruck by developments. Faced with the reality of technological change and its speed of change, the reader is most likely to be left questioning themselves what the “auto-ID trajectory” actually means on varying scales- local, global, universal. They may even be provoked into pondering on the wider implications of information technology. Hundreds of pictures were downloaded, scanned from newspapers and magazines, or personally photographed by the researcher. They were then placed into graphical categories and brought together to reveal hidden meanings on separate palettes. For example, until now, discussion about chip implants being used for commercial purposes were considered rumours generated by Christian fundamentalists, yet pictures of these commercial chips are now available for viewing on some RF/ID web sites (see exhibit 8.6). The old adage, pictures are worth a thousand words holds true and it is the recommendation of this researcher that more qualitative researchers take advantage of the visual perspective.
Another contribution this thesis makes to knowledge is in its extensive bibliography and list of online resources. Any researcher investigating auto-ID innovation in the future can use it as a thorough reference guide or even repeat the same study with other auto-ID techniques. The search for appropriate literature was progressive throughout which made the data collection process an iterative task. Each bibliographic citation was recorded and categorised by several keywords using a database. Photocopies of actual documentation were also made of as many key sources as possible. The paper documentation was then organised into bound folders on the basis of the case study it was addressing and the major application it covered. Folders of special themes, patterns, and future possibilities for auto-ID were also created. It is estimated that over ten thousand pages have been photocopied, printed or downloaded for this research alone.
The research design for this study was chosen to complement the corresponding complexity of the central thesis. It is highly structured, ensuring that a complete investigation is carried out in response to the question “what is the auto-ID trajectory?” The researcher chose a qualitative approach that would allow for the exploration of interdisciplinary perspectives. Central to the research design are the multiple embedded case studies that showcase the auto-ID selection environment, featuring devices and prominent applications. A narrative style of presentation is used throughout to provide substantiation using thick descriptive passages and to assist the reader to interpret themes so as to guide them toward a conclusion. The aim of the researcher is to reveal what was previously hidden in piecemeal observations. The systems of innovation (SI) framework underlying the research effort, has been used to give a holistic perspective on the future of auto-ID by examining the past. The data collection, which was extensive, made use of documentation and archival records with an emphasis on online sources, providing multiple sources of evidence. High-level processing of data was performed by the researcher and numerous data analysis techniques were used to draw out findings. Following is a historical review of auto-ID and chapters 5-7 featuring the actual case studies.
 “Qualitative research is an enquiry process of understanding” (Creswell 1998, p. 15).
 Not all previous innovation studies using the same strategy have awarded the space necessary to convey their conceptual framework, specific research questions, methods of data collection and data analysis techniques. In fact, very few auto-ID studies have actually stipulated the use of any sort of research design.
 See Anderson and Kanuka (2003, p. 33): “[a] historical paradigm attempts to detail the objective reconstruction of historical events.”
 Descriptive accounts help describe “typical patterns of interaction” (Anderson & Kanuka 2003, p. 175).
 This is a distinct ontological quality that is common in narratives.
 According to Yin (1984, p. 23) a case study “investigates a contemporary phenomenon within its real-life context”. See also Simpson, M. et al. (2003, p. 14). It is an ideal methodology to use when a holistic multi-perspective analysis is required (Feagin et al. 1991).
 See ‘Findings and conclusions of ISE case studies on public technology procurement’ (Edquist et al. 1998). The authors discuss the importance of the case study method on pages 9-24.
 Using a quantitative approach would severely limit the results and scope for analysis. Instead “[a]n exploratory study takes a very broad look at the phenomenon under study. Attention is not as focused as in a study to test a hypothesis. The purpose is to gather information, so that a description of what is going on can be made” (Bouma 1993, p. 90).
 One of the greatest strengths that the case study method possesses is in its flexibility. “Generally speaking, in experimental designs, any failure to carry out the pre-specified design has serious implications… Surveys call for considerable and detailed pre-planning before you start the survey proper… Case study, however, is defined solely in terms of its concentration on the specific case, in its context” (Robson 1993, pp. 148f).
 Chapter five especially will be dedicated to tracing the historical perspective of identification, from manual to automatic. In fact, the whole study has a historical element attached to it. Similarly, in the research conducted by Edquist et al. (1998, p. 17) “…case studies within the sub-project employed a historical approach, and many covered processes of technological development spanning several decades.” Industry dynamics happen over time, thus history is very important.
 Refer to section 2.7.1, especially the paragraph which discusses the changes happening from devices one carries, to devices one wears, to devices that encourage ubiquity.
 The term is derived from numerous sources. However two phrases ‘Human Metaphor’ and ‘New Age Systems’ from Andersen Consulting (1991) have influenced the researcher the most.
 One criticism of past research, as already mentioned in the literature review, is that equal space is not allotted to each auto-ID technology, giving an overall unbalanced perspective. This thesis hopes to rectify this situation by giving the same space to each case study.
 There were also questions surrounding the contribution of such a study since a gap in the literature had been filled by the publication of Smart Card Innovation (Lindley 1997).
 Edquist et al. (1998, p. 9) further explains that “…cases are selected for the purpose of generalising findings to ‘theory’, after testing to determine whether the results of initial cases can also be found to occur in comparable cases.”
 Edquist, following Yin, has rightly observed, “[c]ase study research and analysis is commonly confronted with objections to its limited capacity for generalisations… what are often (wrongly) demanded are statistical generalisations- i.e., statements of frequency with respect to relationships among variables occurring within a ‘representative sample’ that can be readily generalised to a larger ‘population’ or ‘universe’. Case studies are not, however, properly intended to make statistical generalisations. Rather, case studies rely on analytical generalisations, in which “the investigator is trying to generalise a particular set of results to some broader theory” (Yin 1994:36)” (Edquist et al. 1998, p. 9). See also Tellis (1997, p. 2).
 This is not to say that the SI framework is limiting in the sense that it pre-empts results. But on the contrary it allows for exploration, with the added benefit of empowering the researcher to compare results from different cases.
 We should note that “[r]eplication logic in multiple case study analyses is similar to that used in multiple-experiment research designs, where if similar results are obtained from the repetition of critical experiments, replication is said to have occurred” (Edquist et al. 1998, p. 23). See also Hersen and Barlow (1976); Yin (1981, 1994, p. 51).
 If one was to contemplate the exclusion of any one case study, which auto-ID technology would be omitted? To remove bar codes for instance, would impact on the comparison with RF/ID tags, perhaps even to the extent of rendering the RF/ID case study quite meaningless overall. Thus, all five auto-ID technologies have been included to cater for the diversity of the industry.
 The manner in which the case studies are designed also allow for additional auto-ID technologies and applications to be studied by any researcher. For instance, a case study on a new auto-ID technology that may emerge in the future could be added without disrupting the structure of the research.
 Furthermore “[a]s computers from your desktop to the collar of your shirt become networked, nothing less than a new medium of human communications is emerging, one that may prove to surpass all previous revolutions- the printing press, the telephone, television, the computer- in its impact on our economic and social lives. This is, in fact, a paradigm shift (Tapscott 1998, p. 24).”
 Internal validity will not be addressed here given this study is exploratory and descriptive in nature. See Yin (1984, p. 36).
 See Soy (1996, p. 1).
 Research that is conducted on any topic should be purposeful. The findings of a study must have some value to someone, somewhere. See Action Research and Organisational Development (Cunningham 1993).
 In Edquist (et al. 1998, pp. 15-21), the researchers describe case study analysis using “themes” because the degree of uniformity among the individual case studies conducted for the European countries happened to be low. In the case studies conducted for auto-ID in this thesis, broad headings denote the description of a dimension, and in each dimension some themes are highlighted more than others. The characteristics to be found in each of the auto-ID innovations should be uniform as all technologies are part of the same industry but deviations are not so much a methodological flaw as they are a reflection of uniqueness.
 It is important to keep in mind, “[w]hile the extremes of tight pre-structured case study designs and loose emergent ones can be justified in different circumstances, in practice most case study work is likely to fall somewhere between these extremes” (Robson 1993, p. 149).
 See Tellis (1997, p. 8) for a brief explanation on each type of evidence.
 The online resources presented (after the bibliography), could assist any researcher to do a detailed study using auto-ID stakeholders as interviewees. This database of contacts is not readily available in print. It offers any auto-ID researcher a substantial advantage for his/her investigation.
 A classic example of this is the Australia Card debate of 1987. All forms of media were heavily involved in the debate; from front page headlines, to letters to the editor, to polls taken during current affairs shows, to talk-back radio comments. See Smith (1989) who gave a first-hand account of the Australia Card: [and] the story of its defeat, especially chapter 8 titled ‘[t]he day of the media’ and chapter 9 titled ‘[t]he role of the press’. In the latter, Smith (p. 150) writes: “[i]f one accepts the Australia Card as a matter of great importance, the bringing about of its demise… must rank as foremost in the achievements of ordinary people. And the events of September would not have occurred but for the part played by the media, and in particular the press… in my opinion the role of the press was paramount… The newspapers responded to the groundswell of public opinion.”
 At each web site it is commonplace to find a myriad of press releases dating back to the mid-90s, a historical overview of the technology, product-specific technical information, case studies of the latest applications, as well as customer testimonials. The information in these web sites has been largely ignored by researchers in the field but should be given attention based on their individual merit. It is here that many references were gathered with respect to the auto-ID trajectory.
 The rise of company e-business solutions has meant that valuable information can now be accessed via the Internet. In order to attract customers, many companies are choosing to invest in the publication and maintenance of a web site that is both informative and can enable a flow of information from anywhere in the world. What may have started off as an extension of a marketing strategy has now become something more encompassing, i.e., the practice of customer relationship management (CRM).
 “In particular, the e-researcher’s and subject’s assumptions and values in relation to the role of the Net will be a part of the research process. Thus, the Net and the research results are indivisible” (Anderson & Kanuka 2003, p. 34). This researcher believes the Net can aid in the identification of the technological trajectory of any artefact. In particular the role of the Internet in exploring new trends and granting qualitative researchers the ability to forecast should be explored by methodology experts further. In the field of technological innovation especially, the Internet is an invaluable data gathering tool. Patent databases can be searched, government policies reviewed, standards bodies referenced and academic research laboratories consulted for future directions, among many other capabilities.
 “The qualitative e-researcher interacts with the research using an in-depth inductive process and an emerging design that is identified during the research process” (Anderson & Kanuka 2003, p. 34).
 See Hewson et al. (2003, pp. 11f).
 Creswell (1998, p. 153) states that “…analysis consists of making detailed description of the case and setting…”
 “In one interpretive form, qualitative content analysis involves the unstructured reading and rereading of the text with the researcher developing a narrative or interpretation that eventually reveals the meanings within the text” (Anderson & Kanuka 2003, p. 176).
 See Robson (1993, p. 393) on the “pattern matching” technique. These often help to explain and simplify findings illustrating the main points. If used correctly these data analysis techniques are excellent ways to convey findings, particularly in a thesis such as this which is for the greater part descriptive. The reader is attracted to the exhibits and is quickly able to make a judgement on what is being described by the researcher. See also Miles and Huberman (1984).
 Creswell (1998) should be cited here as a major influence in the approach I took for the data analysis, complimenting the work of Robson (1993). “In categorical aggregation, the researcher seeks a collection of instances from the data, hoping that issue-relevant meanings will emerge. In direct interpretation, on the other hand, the case study researcher looks at a single instance and draws meaning from it without looking for multiple instances. It is a process of pulling the data apart and putting them back together in more meaningful ways. Also, the researcher establishes patterns and looks for a correspondence between two or more categories… Finally, the researcher develops naturalistic generalisations from analysing the data, generalisations that people can learn from the case either for themselves or for applying it to a population of cases… To these analysis steps I would add description of the case, a detailed view of aspect about the case- “the facts”…” (Creswell 1998, p. 154).
 “Content analysis stands or falls by its categories… without clearly formulated problems for investigation and with vaguely drawn or poorly articulated categories… Since the categories contain the substance of the investigation, a content analysis can be no better than its system of categories” (Berelson 1952, p. 147).
 “Qualitative content analysis thereby allows us to work with the meanings that underlie the content rather than directly with the content we are studying” (Anderson & Kanuka 2003, p. 177).
 “The process of demarcating and labelling a variable in content analysis is often referred to as coding. The challenge for coders is to reliably and consistently identify and qualify each instance of the object or variable they are looking for in the content” (Anderson & Kanuka 2003, p. 174).
 Other aspects of reliability can only be appreciated within the actual case studies themselves. For instance, looking at the case study protocol and then looking at each auto-ID case (by theme) to see whether or not the protocol was followed.