Study Reports on Organizational Diversity, Evolution and Classifications

Description
Study Reports on Organizational Diversity, Evolution and Classifications:- The "business case for diversity" stem from the progression of the models of diversity within the workplace since the 1960s. The original model for diversity was situated around affirmative action drawing strength from the law and a need to comply with equal opportunity employment objectives.

Study Reports on Organizational Diversity, Evolution and Classifications
Abstract This article presents a case for the construction of a formal classircation of manufacturing systems using cladistics, a technique from the biological school of classification. A seven-stage framework for producing a manufacturing cladogram is presented, along with a pilot case study example. This article describes the role that classification plays in the pure and applied sciences, the social sciences and reviews the status of existing manufacturing classifications. If organisational diversity and organisational change processes are governed by evolutionary mechanisms, studies of organisations based on an evolutionary approach such as cladistics could have potential, because as March [March JG. The evolution of evolution. In: Baum JAC, Singh JV, editors. Evolutionary dynamics of organizations. Oxford University Press, 1994. p. 39±52], page 45, states ``t here is natural speculation that organisations, like species can be engineered by understanding the evolutionary processes well enough to intervene and produce competitive organisational e€ects''. It is suggested that a cladistic study could provide organisations with a ``knowledge map'' of the ecosystem in which they exist and by using this phylogenetic and situational analysis, they could determine coherent and appropriate action for the specification of change. # 2000 Elsevier Science Ltd. All rights reserved.
Keywords: Cladistics; Manufacturing; Management; Evolution; Classification

1. Introduction Why construct a classification? This question needs to be addressed in order to understand the benefits and applications that any classification could o€er, let alone a cladistic classification. The desire to classify transcends all disciplinary boundaries whether the enti- ties under study are biological organisms, chemical el- ements or as in the case of this paper, manufacturing

systems. Carper and Snizek [1, p. 65], in their review of organisational classifications concluded that ``the most important step in conducting any form of scienti- fic enquiry involves the ordering, classification, or other grouping of the objects or phenomena under in- vestigation''. In an amusing categorisation of classifications, Good [2], a noted mathematician, provided a list which suggested five purposes for performing classifi- cation: (1) for mental clarification and communication; (2) for discovering new fields of research; (3) for plan- ning an organisational structure or machine, (4) as a check list and (5) for fun. Cormack [3] used this categ- orisation in his lecture to the Royal Statistical Society

78

I. McCarthy et al. / Omega 28 (2000) 77±95

to illustrate the role and benefits that classification o€ers research. Cormack summarised the benefits of a hierarchical classification, stating that ``the information about the entities is represented in such a way that it will suggest fruitful hypotheses which cannot be true or false, probable or improbable, only profitable or unprofitable'' [3, p. 346]. Haas, Hall and Johnson [4] discussed four advan- tages of having a realistic classification. Such a classifi- cation could (1) be strategically helpful for refining hypotheses; (2) aid in the investigation of the validity and utility of existing typologies based on logical and intuitive considerations; (3) serve as a basis for predict- ing organisational decisions or change and (4) permit the researchers to readily specify the universe from which their samples of organisations could be drawn. McKelvey [5] went further by arguing that the formulation of a classification is a necessary prerequisite for the maturation of organisation science and that, if a formal and scientific classification existed, there would be no need for contingency theory. Biologists do not need contingency theory because their classifications make it clear that one does not apply findings about reptiles to mammals when working at a specific level of the classification. The argument for creating a classification is to some extent demonstrated by the large number of typologies and classifications that have been produced by researchers from the social sciences and applied sciences and that many academic disciplines teach with reference to some form of classification. It should be noted that a typology is a description of groups, whose di€erences are identified solely accordingly to the research focus of the investigator. Existing schemes which embrace the subject of organisations include: or- ganisational strategies [6], voluntary associations [7], canning firms and farmers unions [8], general organis- ational classifications [9±11] and manufacturing-based classifications [12±25]. For a review of the above or- ganisational typologies, the reader is referred to Refs. [1,26,27]. The authors of this article sought a classification which would facilitate the storage, alignment and development of structural models of manufacturing systems. It was intended that this classification of models would provide researchers and consultants with a generic library of structural solutions for enabling manufacturing systems to maximise their operating e€ectiveness. The deficiencies of existing classifications of manufacturing systems, prohibited the realisation of the intended benefits of combining a library of ideal models (solutions) with a workable classification of manufacturing systems. This issue was discussed by McCarthy [27, p. 46], who concluded that ``previous research into developing manufacturing classifications has been based on a comprehensive understanding of

manufacturing companies, but with no reference to, or application of the science of taxonomy. This would appear to be a major shortcoming, which reduces the usefulness, stability and accuracy of the classifications. Lessons should be drawn from biological taxonomy in an attempt to stimulate further investigations into this established problem based on the disciplines and rules regularly used by the biological scientist''. Supporting the need for an organisational classification is Romanelli [28, p. 82], who states ``despite the ease with which we may identify meaningful groupings of organ- isations, no commonly accepted classification scheme has been developed''. With this stimulus, a project funded by the Engineering Physical Sciences Research Council (Grant No. GR/K97974) was initiated to investigate the feasi- bility of constructing cladistic classifications of manu- facturing systems. The remainder of this paper details the methodology, findings and conclusions of that study.

2. Introduction to the biological schools of classification There are two main principles of classification within the biological sciences: the phenetic and the phyloge- netic principles. From these two underlying principles emerge three approaches to classification, or schools of classification: phenetic, evolutionary and cladistic (refer to Fig. 1). The three schools of classification are di€er- entiated on the basis of how closely they adhere to a purely phylogenetic principle. That is, the species are classified according to how recently they share a com- mon ancestor. Phenetic classifications are non-evol- utionary and are thus at one end of the evolutionary focus scale, whilst cladistics is a purist approach to the phylogenetic principle. Evolutionary classifications are a synthesis of the phenetic and phylogenetic principles. Phylogenetic classifications have become known as cladistic classifications, because the phylogenetic prin- ciple was defended by the German entomologist Willi Hennig [29] and supporters of his ideas called the prin- ciple phylogenetic systematics, which has now evolved into the term cladism (from the Greek `klados' for branch). The cladistic school's approach to classification involves studying the evolutionary relationships between entities with reference to the common ancestry of the group. Constructing a classification using evol- utionary relationships is considered beneficial, because the classification will be unique and unambiguous. This is because evolution is actual and mankind is cur- rently unable to change evolutionary history, thus pro- viding the classification with an external reference point. With phenetic classifications there is no such

I. McCarthy et al. / Omega 28 (2000) 77±95

79

Fig. 1. Biological schools of classification.

reference point and thus in the words of Ridley [29, p. 367], ``Cladism is theoretically the best justified system of classification. It has a deep philosophic justification which phenetic and evolutionary classifications lack'' Reviews of the three schools of classification [29±31] assess the schools on their ability to produce natural and objective classifications, rather than artificial and subjective classifications. Cladistics satisfies both these criteria, as the entities within a cladistic classification will resemble each other in terms of the defining char- acters and the non-defining characters (characters not used to represent the phylogenetic relationships). Cladistics conforms to the criteria of objectivity because it represents a real unambiguous and natural property of the entity (evolutionary relationships) and thus di€erent rational people, working independently should be able to agree on a classification. There could be valid disagreements between independent investi- gators, but these will be down to assumptions and dis- agreements on the character data and not the underlying philosophy. One of the greatest strengths of the cladistic approach is that the representation of the classification (the cladogram), illustrates the data, assumptions and results, making all decisions transpar- ent. This not the case with existing organisational classifications. Section 5 of this paper presents a dis- cussion on the confusion which exists between the types of manufacturing system which are believed to exist. In summary, a cladistic classification of manufactur- ing systems would provide a system for conducting, documenting and coordinating comparative studies of manufacturing organisations. Such a system could pro- vide the consensus for formally approving, validating

and typifying the emergence of new manufacturing systems. This would help clarify the confusion on whether fractal, virtual and holonic manufacturing systems actually exist or are simply buzz words. This was an issue raised by the Engineering Physical Sciences Research Council [32]. A cladistic classification of manufacturing systems could provide knowledge and observations on the patterns of distributed character- istics exhibited by the manufacturing systems over their evolutionary development. This knowledge could lead to profitable hypotheses about the macro- and micro-evolutionary mechanisms which in¯uence manufacturing competitiveness and survival. Finally, many organisations live their lives looking forward, but to comprehend themselves they must look backwards. The resultant comprehension cannot be used to extrap- olate the future, but it does inform them of where they are and how they got there, and this information is vital for any organisation intending to embark on a journey of change.

3. Cladistics The application of cladistics to manufacturing systems implies certain assumptions about organisational forms, their existence and diversity. Cladistic classifi- cations are produced according to how recently they share a common ancestor. This means that two manu- facturing species that share a recent and common ancestor will be placed in the same group and two manufacturing species sharing a more distant common ancestor might be placed in di€erent groups, but they would be in the same family. As the common ancestor

80

I. McCarthy et al. / Omega 28 (2000) 77±95

of two manufacturing species becomes more and more distant, they are grouped further and further apart in the classification. Eventually all organisations could be placed in a classification possibly known as the `king- dom of organisations'. For this principle of classifi- cation to apply to manufacturing organisations and their systems, investigators must agree that organis- ations evolve and that as new organisational forms emerge, it is possible to identify the distinguishing characteristics from the old organisational forms. Supporting this assumption are organisational theorists who have not produced a complete theory of organis- ational evolution, but have proposed some key con- cepts which include: organisational ecology [33,34], organisational systematics [35,36], the evolution of new organisational forms [28] and the dynamics of organis- ational speciation [37]. These concepts and the assump- tions that accompany them attempt to understand the forces which determine which organisational form is viable for a certain environment; the mechanisms which exist to preserve organisational forms and the mechanisms which are passed from one generation of organisations to another. In summary, the assumptions which govern the construction of a manufacturing cladogram are listed below: . Manufacturing systems evolve and have ancestors. This is evident by the way historians portray the advancement of manufacturing companies from prehistoric man with his tools, to ancient workshops, to the guild of craftsman, to the cottage industries and to factories which eventually became mechanised and automated. . Manufacturing systems speciate. The Ford Motor Company is described today as a lean producer, but its history demonstrates that it once was a craft shop which developed into an intensive mass produ- cer. This suggests that the Ford manufacturing plants have gone through at least two speciation events to produce new `breeds of organisation'. . Manufacturing systems are subject to the theory of natural selection. This theory consists of four basic principles: the principle of variation, the principle of heredity, the principle of natural selection and the principle of adaptation [29]. The principle of vari- ation states that there has to be variation within a population of manufacturing systems. These vari- ations need to occur and happen at random. The principle of heredity states that some manufactur- ing o€spring, on average have to resemble their parents more than resemble other members of their species. This is found when new organis- ations are born within an industry. They are more similar to organisations within that industry, than they are to organisations in other industries. This

inheritance is controlled by the organisational equivalent of genes (knowledge transfer or memes [38] or competence elements (comps) [36]), which are passed on to o€spring by chromosomes (people, communication, society) in the same form as they were inherited from the previous gener- ation [39]. If heredity were perfect, the principle of variation would not exist. The principle of natural selection suggests that manufacturing systems with a superior adaptation generate similar manufactur- ing systems (o€spring) and as long as the o€spring resemble their parent, the characters of manufac- turing systems that generate more o€spring than average will increase in frequency over time. This concept is supported by Hannan and Freeman [34] who believe that selection pressures, force organis- ations to imitate the successful organisations, the result being a reduction in organisational diversity and a net increase of a particular type of organisational form. The fourth principle, the principle of adaptation, refers to the variations in manufactur- ing systems which provide an advantage for sur- viving and existing. This is when manufacturing systems change so as to maintain existence.

3.1. The cladogram A cladogram is a tree structure capable of representing the evolutionary history of a group of manufactur- ing systems. The tree structure illustrates the relationships between the di€erent members of the group under study, according to the acquisition and polarity of characters. Fig. 2 shows a group of manufacturing species con- sisting of Ancient craft systems, standardised craft sys- tems, modern craft systems, neocraft systems and skilled large scale producers. This figure is a section from the master cladogram of automotive assembly plants (Fig. 3 and Table 1). This pilot study was undertaken to provide a worked example which would introduce the reader to cladistics and the various types of cladistic grouping that exist. The construction of this cladogram is reported in Section 4. It is important to note that this was a pioneering study and that many of the types of manufacturing system proposed in Figs. 2 and 3 will not be known to the reader. This is not because they are newly formed types of manufacturing systems, but rather that the automobile industry has not been studied using the cladistic approach. The labels given to the species shown in Figs. 2 and 3 do not conform to any codes of nomenclature for organisations, because none exist. Constructing a classification is a taxonomic process and thus by the definition of taxon- omy, groups (taxa ) are formed and are then allocated

I. McCarthy et al. / Omega 28 (2000) 77±95

81

Fig. 2. Five taxa cladogram.

Fig. 3. Automotive cladogram.

82 Table 1 Automotive cladistic characters 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

I. McCarthy et al. / Omega 28 (2000) 77±95

Standardisation of parts assembly time standards assembly line layout reduction of craft skills automation (machine paced shops) pull production system reduction of lot size pull procurement planning operator based machine maintenance quality circles employee innovation prizes job rotation large volume production suppliers selected primarily by price. exchange of workers with suppliers socialisation training (master/apprentice learning) proactive training programs product range reduction automation multiple subcontracting quality systems (procedures, tools, ISO 9000) quality philosophy (culture, way of working, TQM) open book policy with suppliers; sharing of cost data and profits ¯exible, multifunctional workforce set-up time reduction Kaizen change management TQM sourcing; suppliers selected on the basis of quality 100% inspection/sampling U-shape layout preventive maintenance individual error correction; products are not rerouted to a special fixing station sequential dependency of workers line balancing team policy (team motivation, pay and autonomy) Toyota verification of assembly line (TVAL) groups vs. teams job enrichment manufacturing cells concurrent engineering ABC costing excess capacity ¯exible automation for product versions agile automation for di€erent products insourcing Immigrant workforce dedicated automation division of labour employees are system tools and simply operate m/c's employees are system developers; if motivated and managed they can solve problems and create value product focus parallel processing (in equipment) dependence on written rules; unwillingness to challenge rules such as the economic order quantity further intensification of labour; employees are consider part of the machine and will be replaced by a machine if possible

a name (nomy = naming). Every e€ort has been made to assign labels which describe the defining characterof the system and where possible existing terms

such as craft, mass, agile and lean have been used. Thus, the labels given to the species are simply for the istics purpose of di€erentiation and communication. The in-

I. McCarthy et al. / Omega 28 (2000) 77±95

83

formation content provided by the labels is considered to be a level higher than simply referring to each species, as species 1, species 2, species 3, etc. The cladograms illustrated in Figs. 2 and 3 are both clades, as they contain a set of species including the most recent common ancestor of all the members contained within that set. It is important to understand that Fig. 2 is a portion or segment of Fig. 3 and that both Figs. are clades, despite the fact that Fig. 2 is a subset of Fig. 3. This is due to research focus (establish evolutionary boundaries) and the information pre- sented. That is, Fig. 2 in its entirety and in isolation, is by definition a clade, despite the fact that Fig. 2 can be expanded to Fig. 3. If we assume that a manufac- turing researcher is only interested in the clade shown in Fig. 2 and that his specific interest is devising manu- facturing strategies for modern craft systems, neocraft systems and skilled large scale producers. Then this group of manufacturing species is known as the ingroup (the study group or the group of interest). Observations and hypotheses are made about the ingroup by comparing it with the various outgroups and most importantly with the sister group (the out- group that is genealogically the most closely related group to the ingroup). It should be noted that the ancestor of the ingroup is not the sister group, because the ancestor by definition will always be a member of the ingroup. The numbers shown on the branches of Figs. 2 and 3 denote the acquisition of characters. Character `1' (standardisation of parts) has a specific location on the tree that indicates that ancient craft systems do not possess character `1' and that standardised craft sys- tems, modern craft systems, neocraft systems and skilled large scale producers do possess character `1'. Thus, ancient craft systems are the ancestor of a new gener- ation of manufacturing systems that are based on the acquisition of character `1'. Similarly, modern craft sys- tems are a descendant of standardised craft systems as it later acquired character `2' (production time stan- dards) and character `47' (division of labour). The characters `13', `48' and `50' resulted in the formation of neocraft systems, whilst the characters `3', `16' and `32' result in the emergence of skilled large scale produ- cers.

3. 4. 5. 6. 7.

Code characters. Establish character polarity. Construct conceptual cladogram. Construct factual cladogram. Taxa nomenclature.

In order to demonstrate how a cladogram is produced, the cladogram in Fig. 3 is referred to. The cladogram is a classification of automotive assembly plants. It was produced to the conceptual level and was compiled using data from several studies of the automotive industry. These studies include the evol- ution, population density and mortality in the automo- tive industry; [44±48]; historical accounts of the industry, sometimes focusing on specific geographic regions; [49,50], to specific studies which examined the change in manufacturing techniques used within the industry [51±53]. Technical, business and financial reports produced by the automobile industry were also obtained. These documents detailed events and issues which were in¯uencing how the industry was evolving. The most significant of these documents are listed as references [54±78]. 4.1. Select the manufacturing clade The starting point is to define the clade to be studied. Such a step requires a decision which in itself is a form of classification, as the investigator must select a group of manufacturing systems which satisfy certain research objectives or interests. For example, a manu- facturing clade could be di€erentiated on the basis of the market industry into which it was born to survive, e.g. the automotive industry, electronic component manufacturers, cutting tool manufacturers, etc. Classifications based on industry di€erentiation are widely used and accepted and are di•cult to ignore. In the United Kingdom, the basic framework for analys- ing industrial activities is the standard industrial classification (SIC) [79]. The SIC is described by Price and Mueller [80] as an empirical classification which is not derived in any way from theoretical ideas on how ac- tivities should be grouped. However, it does group together organisational entities that are involved in resource exchange and transformation of a similar nature. This description of organisational activity equates to the definition of an organisational ecosys- tem as proposed by Baum and Singh [81]. A clade by definition can be equivalent to di€erent levels in the hierarchy. This is illustrated by Fig. 4, which shows how the ecological and systematic hierarchies of organisational evolution relate to each other (this figure has been adapted from [81] to include the clade level). For the purposes of this study, the automobile assembly industry (the clade) was selected, because it exists as a population of manufacturing organisations

4. Building a manufacturing cladogram The proposed framework for constructing a cladistic classification of manufacturing systems has been ident- ified and adapted from classic biological approaches to cladism [40±43]. The seven stages are listed below: 1. Select the manufacturing clade. 2. Determine the characters.

84

I. McCarthy et al. / Omega 28 (2000) 77±95

4.2. Determine the characters Once the clade has been selected, a number of di€er ent types of manufacturing system would appear to be a member of that clade (mass, lean, agile, craft, job, etc.). The complete membership of this particular clade is not yet known, because no formal or validated clades for manufacturing systems exist. It is common practice to work on existing clades within the biologi- cal sciences, because the majority of the taxonomic based research, is concerned with validating, enhancing and expanding the knowledge contained within existing cladograms. As this was a new study, a primary objec- tive of the research was to examine the evolutionary development of the entity and to identify the members of the clade. This is a process of `mining for species' and during this historical excavation, evidence is sought which will suggest the possible existence of a particular type of manufacturing system. This evidence tends to be in the form of published material or archives, which detail the existence of the manufacturing system, along with a description of its operations and defining characteristics, the location where it exists/existed and a date/period when it was first dis- covered or developed. This mining process uncovers the characters which will be used to build the cladogram. Whilst undertak- ing this exploration there are a number of steps which can be followed to help identify the final set of charac- ters which will be used to construct the cladogram. The process of determining the characters for the auto- motive cladogram consisted of two steps: character search and character selection. Character search is the task of building the initial set of characters, by simply listing known attributes possessed by automotive assembly plants. Determining which characters from this initial set should be used to construct a classifi- cation is the task of character selection. 4.2.1. Character search When searching for the manufacturing systems that constitute the clade and the characters that distinguish the species phylogenetically, it is helpful to know what to look for and what to avoid. Whereas, an attribute is a descriptive property or feature, a taxonomic charac- ter is a feature which is used in a classification. It is also important to di€erentiate between the character (the actual feature) and the character states which are a condition that this feature exhibits. For example, the character `plant layout' has numerous character states: job shop, ¯ow line, functional layout, manufacturing cells, etc. The school of classification used will contain theories which determine what is an acceptable taxonomic char- acter. For instance, in cladistics, a taxonomic character has to point to a homology between two organisations,

Fig. 4. Hierarchies of organisational evolution, adapted from [81].

(species) that make and sell a closely related set of well defined products. It is an industry which is widely known and studied and this provides benefits in terms of communicating, disseminating and validating the research. It is also a relatively young industry which has been extensively documented and this makes the investigation into phylogenetic relationships relatively easy, when compared to an industry such as the hand tool manufacturing industry, which can be traced back to prehistoric man. This is an important point, because there were no existing cladistic classifications of organ- isations which could be used as a reference or starting point, so it was important to select a study group which would satisfy and assist the research objectives in terms of information collection and results dissemi- nation. Also, the decision to study the automobile assembly industry would enable both the dissemination and exploitation of any benefits to be related to the standard industrial classification (SIC). Identifying the ancestor of a clade is a process of historical investigation where evidence is accumulated to determine the origins of a certain manufacturing type. For example, the origins of car manufacturing stem back to Karl Benz and his three-wheel auto- mobile. In terms of manufacturing systems, this would be regarded as a craft system which evolved into an early factory system and then into a mass type organisation. The process of identifying an ancestor is initially ambiguous and di•cult, both for biologists and manufacturing researchers, but the process of constructing the cladogram confirms or refutes this initial assump- tion.

I. McCarthy et al. / Omega 28 (2000) 77±95

85

whereas in phenetic classifications, a taxonomic character contributes to the mathematical tightness of a clus- ter. To avoid searching for and selecting characters which are inappropriate Sneath and Sokal [43] describe certain kinds of characters which should be clearly dis- qualified from a taxonomic study. These are listed as inadmissible characters and include: . Meaningless characters. A character must re¯ect the internal nature of the entity, therefore, the name of a manufacturing company would not be included as a character to represent the activities of a manufac- turing system. . Logically correlated characters. Those characters which are a logical consequence of another, should be excluded. For example, if we assume that cell- based team working, requires a cellular layout, then there is a logical correlation between these two char- acters, i.e. if one character state exists, another will automatically. . Partially logical correlation's. The degree of independence is the subject of this kind of character, as a greater number of cases exist where the dependence of one character upon another is only partial. For instance the size of a workforce will be to a degree, relate to the number of machines that a manufactur- ing company has. After further investigation it could be found that the degree of dependency is small, because other factors, such as the type of technology and the type of product also in¯uence this character. Therefore, very few partially logical correlations are regarded as inadmissible. Hull [82] provides an empirical correlation to estimate the degree of independence between two characters. . Invariant characters. If a character which is normally variable, is invariable for the sample of entities under study, then it should be removed from the analysis. Such characters o€er no benefits in terms of assessing similarity. An example is the absence or presence of manufacturing technology. When con- sidering all forms of organisation, this character would vary from organisation to organisation. However, as the presence of manufacturing technol- ogy is a conforming definition for a manufacturing system, this character would not change for a popu- lation containing only manufacturing systems. The search for automotive assembly characters consisted of investigating the historical development of the car making industry by analysing the work and data of the studies cited in Section 3. The characters ident- ified, although well known, were treated as arbitrary or capricious characters, as their identification for cla- distic purposes must be confirmed. Taxonomists dis- cover characters whilst studying the entity and constructing the classification, thus many characters

are found as they come to complement the information content of the classification. This last point applies specifically to cladistics, because cladists tend to quickly eliminate characters which have no evolution- ary significance in their data sets and therefore produce classifications objectively and e•ciently. In addition to searching for characters by studying the entity, the use of reference characters was con- sidered. That is, does an exhaustive list of manufactur- ing or organisational characteristics exist and would this list help the search and selection process. To build such a list has been a common objective for many tax- onomists, but there are several problems associated with the management and use of such a list. The cost of building an exhaustive list would be high and there is no evidence that building such a list is feasible. There are many issues to manage: duplication of data, partial redundancy between characters, correlation and dependency patterns between characters. Even if such a list was available, using it might not be cost-e•cient, because the cost of selecting characters from all poss- ible characters could be prohibitive. The primary benefit of a reference list of characters, is that it provides a feel good factor and a confident starting point for researchers producing a classifi- cation. However, total reliance on a so-called exhaus- tive reference list, would be foolish and misguided, because all classifications are undertaken in situations where the complete character set is not known. To assist the search for automotive characters and to understand the significance of the characters with regards to the entity and its evolution, several categorisations of characters were identified and referred to: [4,36,83±85]. It is important that the categories do not dictate, but suggest, because the ultimate decision gov- erning character selection within a cladistic study is the existence of a synapomorphy which results in an hom- ology. Synapomorphies are characters which have a derived state and are shared by two or more taxa and thus indicate common ancestry for the manufacturing systems within this group. The distinction between homology and analogy is a fundamental concept of cladistics. A homology rep- resents `true similarity', whilst analogy is considered superficial similarity which generates noise or mislead- ing observations. An analogy is a structural grouping where a character is shared by a set of species and is derived from a common ancestor. Thus, choosing a character which is an analogy should be avoided. The relationship between analogy and homology is clearly demonstrated in Fig. 5 [29]. It is important to note the three groupings, as only monophyletic groups are included in a cladistic classification. The monophyletic groups are the groups which result in an unambiguous hierarchic arrangement, because the group contains a

86

I. McCarthy et al. / Omega 28 (2000) 77±95

Fig. 5. Homologies and analogies.

common ancestor and all its descendants and there is no con¯icting character data. Consider Fig. 3, and the characters `8' (pull procure- ment planning) and `20' (multiple subcontracting). Character `8' appears in the Toyota production system family, which includes: lean producers and agile produ- cers, whereas character `20' appears in the mass produ- cers family, which includes: pseudo lean producers, modern mass producers, European mass producers and intensive mass producers. If characters `8' and `20' are replaced with one character, say character `Z' (procure- ment policy), the structure of the cladogram would change. This is because homologies have been created between taxa which are in fact evolutionarily remote. Thus, character `Z' is an example of an analogous character because pull procurement is constrained by character `6' (pull production) and would not naturally emerge in mass producers. Similarly, it is postulated that character `20' is associated or dependent with some or maybe all of the characters on the same line-

age to the extent that it would not emerge in species which do not already exhibit character `14' (mass subcontracting by price bidding). 4.2.2. Character selection This is a screening process and in the case of cladis- tics, a character is validated if it is a synapomorphy. Thus, the selection phase in cladistics is equivalent to a test of homology. Two methods were used on the automotive study to screen characters: (1) direct test of homologies and (2) resolving character con¯icts. It should be noted that prior to building a cladogram the organisational systematist may only have a general knowledge of the ancestral links between species. Therefore, it is not obvious that a character is an analogous character at the beginning of the analysis, it is only confirmed during the construction and analysis of the cladogram. The direct test method is based on the argument that homologies and analogies tend to exist on a conti-

I. McCarthy et al. / Omega 28 (2000) 77±95

87

nuum of resemblance, where the homologies are at the high extreme resemblance end, whilst the analogies tend to exhibit only moderate resemblance [43]. Thus, even if a complete and valid historical account (`fossil record') for automotive manufacturing systems existed, the investigator would still be dependent on resem- blance based similarity. From a purist point of view, cladists argue that resemblance is not a definitive test of homology, but there is a strong case to suggest that it is a good indicator, because there are external, com- positional and structural measures which relate phenetic similarity with homology. Thus, the direct test consists of the external method, compositional method and the structural method. The external method can be applied without study- ing or knowing the internal structure of the feature. Any external characteristic of the feature is used to identify the existence of some fundamental diversity within the feature. For example, the procurement sys- tems that typically exist in lean manufacturing produ- cers tend to have subcontractors/suppliers which are located within a short distance of the assembly plant. It was common for subcontractors/suppliers in Western manufacturers to be located almost anywhere on the planet. Thus, from an external perspective only, there is a significant di€erence and the location of sub- contractors relative to the main assembly plant, could be a potential character, because no evidence of ana- logy has yet materialised. The compositional method requires the investigator to list the parts which consti- tute the considered character. This internal breakdown is then used in a comparison with other organisational species. For example, a reduction in the number of tier levels in a supply chain might be evident in service or- ganisations and retail organisations and this circum- stantial evidence could be used to guide the selection of characters for manufacturing systems. With the structural method, the focus is on how the di€erent el- ements of the character interact with each other and if there is a case for splitting a potential character into two or more characters. This decision is made purely on the basis of how the elements exist and their depen- dence with one another. Identifying and resolving character con¯icts occurs continually during stages 2±6 of the cladogram frame- work, but the final validation is a postcladogram con- struction exercise (stages 5 and 6). Once a preliminary cladogram has been constructed, it usually exhibits cer- tain character con¯icts. These con¯icts can be natural occurrences, such as parallelism or coevolution. They can also result from analogous characters, or improper coding of characters. Improper coding can be the result of analogous or imprecise definition of charac- ters states, or using the wrong polarity (i.e. confusing the derived and the primitive state), or using characters which are too general. The advantage of validating

homologies after a preliminary cladogram has been constructed is that the validity of a character is ques- tioned only if it generates a con¯ict with the others characters which are consistent and congruent with each other. Most classifications will have a consistent core, which can be identified in cladistics by running a clique analysis [86]. Any character which does not belong to the clique set should go through a thorough test of homology. It should be stressed that it is often at this stage that many characters are usually discov- ered and refined, as the phylogeny of the clade is gradually revealed and understood by the taxonomist. 4.3. Code characters Once a set of characters has been identified, along with the set of automobile assembly species which are a consequence of these characters, the relationship between the characters and the species are examined in order to allow the construction of the cladogram. A cladogram can be constructed from the character data, because a cladistic character has three properties: direction, order and polarity [87]. The coding of a character facilitates the processing of the character set. Ordering is that property of a character which refers to the possible character change sequences that can occur. The character property, direction, refers to the transition between the character states. When an inves- tigator determines the actual direction of transformation the character is said have a `polarised' state. 4.4. Establish character polarity To assess character polarity, an outgroup comparison is undertaken. This is based on the recognition that once the characteristics of the closest relative have been discovered, the information for determining which characters are primitive and which are derived is revealed. Hence, this comparison is based on the rule that for a given character with two or more states within a group, the state occurring in related groups is assumed to be primitive [88]. Any character state found only in the ingroup is considered to be derived [30]. Decisions governing the character polarity found at the outgroup node can be either decisive, with the node labelled as primitive (0) or derived (1), or equiv- ocal, with the node labelled primitive/derived (0, 1). If this method is applied to the cladogram shown in Fig. 3, the outcome would be inconclusive, because this tree has already been resolved and there are no inconsistencies in the character data. Therefore, in order to demonstrate this method, a cladogram con- sisting of taxa and characters from the automobile study is used, but the data and structure of the tree have not been resolved. This unresolved data (Table 2)

88 Table 2 Data matrix for Figs. 6±9

I. McCarthy et al. / Omega 28 (2000) 77±95

Character 1 Ancient craft (AC) Standardised craft (SC) Modern craft (MC) Neo craft (NC) Skilled large scale (SLS) Large scale (LS) Mass (M) 1 0 0 1 1 0 0/1

Character 2 1 1 0 1 1 0 0/1

Character 3 0 1 1 1 1 0 0/1

Character 4 0 0/1 0 1 1 0 0/1

is used to demonstrate the process of determining character polarity (Figs. 6±10). Fig. 6 is a possible cladogram structure for the data contained in Table 2. The nodes are labelled 1±6, whilst the species are labelled using letters (AC, SC, MC, NC, SLS, LS and M). Beginning with the charac- ter 1 from Table 2, each branch end of the cladogram is labelled with the corresponding character state (Fig. 7). Next, starting from the furthest branches (branches AC and SC) a polarity decision for node 2 is made. The nodes of the cladogram are labelled `0' if the lower node and adjacent branch are both `0', or `0' and `0, 1'. The nodes will be labelled `1' if the lower node and adjacent branch are both `1' or `1' and `0,1'. If the branches/nodes have di€erent labels, one `0' and the other `1', then the node is labelled `0, 1'. The root node (node 1) is not considered, because in order to analyse this branch another outgroup is needed. Thus, node 2 is labelled `0, 1', because the first branch (AC) is `1' and the second branch (SC) is `0'. The next stage is to identify what is termed the near- est branching structure, which occurs at node 6 (Fig. 7). The nodes of the branching structure are labelled

Fig. 7. First polarity decision using character data 1.

using the same process, but by beginning at the lowest node on the branching structure (node 4). Thus, node 4 is labelled `1', because NC is `1' and SLS is `1' (Fig. 8). Continuing towards the ingroup (M) the remaining nodes (nodes 3 and 5) are labelled, until only the out- group node (node 6) remains. Node 5 is labelled `0/1' because LS is `0' and node 4 is `1' and node 3 is labelled `0', because MC is `0' and node 2 is `0/1' (Fig. 9). The analysis for character 1 is complete when node 6 is labelled. Node 6 is found to be decisive (`0'),

Fig. 6. Determining the character polarity for mass producers and its corresponding outgroups.

Fig. 8. Second polarity decision using character data 1.

I. McCarthy et al. / Omega 28 (2000) 77±95

89

state will be decisive for the outgroup node. If the last outgroup has a di€erent character state, then the char - acter state decision will be equivocal. 4.5. Construct conceptual cladogram Various tools exist to construct cladograms which provide a `best estimate' of the evolutionary relation- ships contained within the data matrix. These tools have one of two approaches: 1. Construct the best cladogram using a specific algorithm. 2. Apply a criterion for choosing between alternative cladograms. The first approach is faster, but does not rank the trees which are considered suboptimal. The second approach provides ranking for all the trees under com- parison, but it is not able to generate exact results for matrices with more than 12 taxa, owing to compu- tational di•culties [12]. From these two approaches four methods for estimat- ing phylogeny have developed: (1) methods based on pairwise data, (2) parsimony methods, (3) Lake's method of invariants and (4) maximum likelihood phy- logenies. The parsimony method selects the shortest tree, i.e. the tree requiring the least evolutionary charac- ter changes. This method is the most popular because it has a simple rule of application which is; the longer the tree length, the worse the fit; the shorter the tree length the better the fit. The other methods vary between parsimonious and phenetic, but were developed to compare nucleotide specimens, DNA and molecular sequences. Thus, a parsimonious approach is adopted as it aims to select a best tree on an evolutionary basis rather than a phenetic basis. Also, the method is based on the tree structure rather than elements of the entity (DNA, nucleotides, molecular distances, etc.) and thus there would appear to be no limitations when applying it to a manufacturing cladogram. For a detailed account of parsimony methods, see [89]. The testing of a cladogram is essentially based on its ability to explain the phylogeny of the clade. With this aim there are two sets of problems: 1. The proposed relationships are not acceptable or not historically coherent. 2. Several con¯icting cladograms of the same length are obtained. Refusing a cladogram because it does not fit with historical evidence is a dangerous exercise as there are no general rules linking the number of characters acquired by a species and its period of existence. Very evolved species might become unfit in a later period. Once a cladogram has been produced, the first step is to map the character changes onto the tree in order
Fig. 10. Polarity decision for node 6 (outgroup node) using character data 1.

Fig. 9. Third and fourth polarity decision using character data 1.

because node 3 is `0' and node 5 is `0/1' (Fig. 10). Thus, by using the outgroup comparison a best esti- mate of the polarity was made and `0' was found to be primitive and `1' is derived for character 1. This process of assessing character polarity is made for each character. It should be noted that although this procedure plays a significant role in identifying character polarity and resolving any con¯icts that may exist in the cladogram, the final validation of character states is subject to the rule of parsimony (Section 4.5). In summary, two rules of analysis are used to con- duct an outgroup comparison: the doublet rule and the alternating sister group rule [88]. With the doublet rule, if the sister group and the first two consecutive outgroups have the same character state, then that character state is decisive for the outgroup node. Any two consecutive outgroups with the same character state are called a doublet. With the alternating sister group rule, if the character states are alternating down the cladogram, and if the last outgroup has the same character state as the sister group, then the character

90

I. McCarthy et al. / Omega 28 (2000) 77±95

to have a global view of the proposed phylogeny. It is common practice to shape test the cladogram by add- ing additional species and characters. It is important to note that adding characters and species at this stage of the framework is easier and more reliable than at the clade building stage. When examining the top section of the cladogram, the investigator should question if the acquisition could have led to a speciation, or if it is just a case of anagenesis. If a character could have potentially cre- ated a viable species, and if historical evidence of the existence of this species can be gathered, then the species should be added. The automotive cladogram was constructed using MacClade Version 3 [90]. MacClade provides an inter- action environment for exploring phylogeny and resol- ving character con¯icts. MacClade allows the user to manipulate cladogram structures and character data and to visualise the characters on each branch. Finally, MacClade provides tools for moving branches, rerout- ing clades and automatically searching for the most parsimonious tree. 4.6. Construct factual cladogram This stage involves studying real and existing manufacturing organisations in order to observe the manufacturing systems which they operate. This typically consists of plant inspections, discussions with employ- ees, assessment of planning and control procedures and assessment of documentation (annual reports, business plans and surveys, etc.). The study aims to validate the existence of the characters identified during the previous stages. It will test the validity of any proposed tree structure by ensuring that the char- acter data matrix is complete (i.e. no important histori- cal events which relate to characters have been omitted) and that the assigned polarity is correct. This stage is to an extent, validation by dissemination, because the factual data will be used to verify the con- ceptual data. The validity of any proposed tree struc- ture will also be tested by allocating existing organisations a position on the cladogram. The factual stage is undertaken because character reversal (the dropping of a character) is a possible pro- cess with manufacturing systems. This paper suggests that two forms of character reversal could occur within organisations: phylogenetic reversal and reactive rever- sal. Phylogenetic reversal is illustrated in Fig. 2(a) by character `(20±)' where by the character has been reversed naturally by the circumstances of evolution and thus is illustrated on the cladogram. Reactive character reversal occurs, because organisations realise that their current position is at the end of an inap-propriate evolutionary path and take the decision to acquire a new organisational form. This change pro-

cess results in the organisation acquiring and reversing the necessary character states which will lead to the new organisational form. This reversal is similar to Sagasti's model of adaptive behaviour [91], which occurs due to selective pressures. Reactive reversals are not part of the phylogeny of a clade, they are a measure of a systems' lack of strategic focus.Biological organisms tend to evolve according to the rule of parsimony (smallest number of evolutionary changes), but organisations which to some extent in¯u- ence evolutionary destiny, do not always take the most parsimonious route. 4.7. Taxa nomenclature The name given to a taxa of manufacturing systems is more than a word which simply acts as a means of reference. The name given to a taxa must act as a ve- hicle for communication, be unambiguous and univer- sal. It should also indicate its position within the classification hierarchy. Je€rey [40] describes the codes of nomenclature used for plants (International Code of Botanical Nomenclature), for bacteria (International Code of Nomenclature of Bacteria) and for animals (International Code of Zoological Nomenclature). Each code di€ers in detail but certain basic features are common. For a summary of the relevant codes, discussed in an organisational context, the reader is referred to [92].

5. Applications This article began by discussing the reasons for undertaking a classification study using cladistics. Although many of the reasons presented might appear to be common sense, this does not dilute their import- ance and contribution to any serious and scientific in- vestigation into organisations. The following discussion presents possible academic and practical ap- plications of cladistics. 5.1. Understanding organisational diversity (organisational systematics) There is common agreement on the definition of the attributes of a just-in-time manufacturing system, see for instance [93, 94], but these definitions are su•- ciently vague to cause confusion with the terms ¯exible manufacturing systems, agile manufacturing systems, world class manufacturing systems and lean manufac- turing systems. This problem has been identified by many researchers and is summarised by the following quote: ``( F F F) the diversity involved in the manufactur- ing industry is such that it is unlikely that all industry types should be aiming for the same procedures, pol-

I. McCarthy et al. / Omega 28 (2000) 77±95

91

icies and culture. Yet there has been very little research which tries to identify what the term world class (WC) means for certain industry types. This leaves the cur- rent apparently poor performers with inadequate infor- mation to decide whether they are really not of WC standard, and, if not, insu•cient appropriate guidance to determine how to achieve the WC goals to which most would agree to aspire''. [95, p. 43]. Despite the need for knowledge on the evolution of new organisational forms, as described in Section 1 of this paper, no theoretical consensus exists for organis-ing and supporting the vast number of empirical stu- dies which examine industrial and organisational diversity. Using a systematic and comparative method such as cladistics, permits an assessment of the general- ity of the attributes of complex systems [96]. Cladistic classifications and the desire to develop a theory of or- ganisational di€erences could play a significant role in explaining the processes by which the practices and structures of organisations and organisational forms persist and exist over time. 5.2. Understanding organisational ecology Where as the first application was concerned with creating a systematic system of organisational diver- sity, this discussion suggests that cladistic classifi- cations could provide the comparative index which might assist the creation of theories which focus on or- ganisational processes (e.g. replication, mutation, recombination, learning, entrepreneurship, competition and natural selection) and organisational events (e.g. birth, death, transformation, speciation and extinc- tion). Cladistics could be coupled with functional stu- dies which seek to ascertain an overall measure for complexity, stress resistance, mortality index etc. in an ecosystem. A functional study of organisations would aim to forecast environmental/market changes (the rate of new product introduction, service mechanisms, supply relationships, etc.) and forecasts on which man- ufacturing species will dominate, compete and survive such market and economic conditions. Functional stu- dies and cladistics are viewed as complementary disci- plines by many biologists and philosophers [97], since their results describe di€erent properties of species (re- spectively, their identity and their strategy for survi- val). The goal of functionalists is to develop a catalogue of knowledge, related to a classification, for identifying strategies for survival. An example of such a classification is the CSR model of Philip Grime from the NERC unit of the University of She•eld [98]. The CSR model, models the environment along two dimen- sions: stress and disturbance. Stress is a limitation put on the resources necessary for the organisations to sur- vive. In biological terms, stress is the lack of nutrients, the lack of light, cold temperatures, etc. In manufac-

turing terms, examples of stress are unreliable sourcing mechanisms, lack of skilled labour, lack of finance, machine breakdowns, etc. Disturbance is a serious en- vironmental event which happens occasionally. Examples of disturbances in biology are fire, frost, earthquakes, etc. In manufacturing, disturbances are strikes, fire, the loss of a market. If several organis- ations exist in a perfect environment with no stress and no disturbance, they tend to be competitors (C). Competitors are merciless and compete to be the tal- lest, biggest, etc. If stress appears in the environment, stress tolerators (S) tend to take the lead over competi- tors, whose strategy for survival is not appropriate. If disturbance is high, ruderals (R) are better adapted and dominate the environment. Competition is the dominate functional type studied and documented in business studies and in manufacturing management, but it would be interesting and possibly beneficial to develop policies for creating manufacturing systems which are tolerators or ruderals.

5.3. Understanding and achieving organisational change

( F F F ) an attempt was made to identify a general implementation sequence. However, similar to the observation made by Im and Lee [99], a general implementation pattern for the JIT practices could not be established [94, p. 8].

The first two applications were academic in nature, but the deliverables from such applications could pro- vide organisations with new tools and knowledge which could help them to be proactive in the manipu- lation of their evolution. Since cladistics is a classifi- cation method which ties its definition of similarity to naturally occurring change processes, the result is that the information contained within a cladogram is useful for identifying standard change sequences. A clado- gram could also provide a framework or index for positioning and benchmarking studies [100]. The analysis of a cladogram goes further than a simple specification of a change sequence. It indicates: the sequence of steps required to transform an organis- ation to a certain state, along with the characteristics which must be dropped (the `unlearning' steps). If there is agreement that the cladogram has been con- structed according to the rules of parsimony, the physi- cal and financial cost of the identified change route would be minimised. The tree-like nature of a cladogram could be com- pared to a map, which once constructed provides or- ganisations with an unambiguous and precise definition of the starting point of the change journey.

92

I. McCarthy et al. / Omega 28 (2000) 77±95

If the journey is a mimetic process then it will also provide a definition of the destination. 5.4. Strategy Despite the popularity of ¯exible manufacturing systems, managers su€er from inadequate frame - works to help incorporate ¯exibility into their stra- tegic planning [101, p. 7]. A cladogram provides a snapshot of the evolutionary history of a company. Thus, it can be used by managers to check that their vision for the future is consistent with their understanding of the past. Cladistics also provides an interesting measure of stra- tegic excellence, through the principle of parsimony. Strategic management is a discipline which was under close scrutiny in the eighties and many researchers questioned if a correlation could be found between the practice of strategic management and organisational performance, usually defined as profitability. Although some researchers confirmed the existence of such a cor- relation [102±104], many others found no correlations whatsoever, [105±109]. Strategic management is con- cerned with the long term sustainability of profits and thus strategic excellence can be di•cult to define, because assessments may need to view a decade of financial loss before capturing the benefits of a well- articulated strategy. If there is agreement with the statements that ``( F F F ) successful firms have followed more than one route to successful redesign.'', ``Too often, ( F F F), pieces are missing from the strategies and structures firms create in the process of redesign'' [110, p. 129], then the prin- ciple of parsimony could o€er a legitimate definition of strategic excellence. Researchers can easily question, a posterior, how parsimonious the strategy of a firm was. The Toyota Motor Company demonstrates a remark- able record of excellent strategic practices, with the highly focused introduction of the Toyota production system [111] and its subsequent evolution toward lean production. Cladistics could be used to develop a set of performance measures which would govern the stra- tegic decision making process within companies.

Cladistics, as with all classifications, is a method for systematically organising knowledge about a popu- lation of entities. It is a process for studying diversity and attempting to identify and understand laws and re- lationships which explain the evolution and existence of the variety groups. Its intellectual and practical value is derived from this ability to explain. This article suggests that cladistics is a novel and appropriate approach for producing an organisational classification, because unlike the best phenetic classifi- cations and the multitude of subjective classifications, cladistics has an underlying philosophy (evolution) and accompanying rules and procedures. Cladistics uses evolutionary relationships to identify and form groups, because evolution is the process which accompanies the changes which materialise to produce di€erent or- ganisational forms. The resulting classification and the knowledge contained within, provide insights into organisational diversity. These insights include: observing the patterns and events which accompany the organis- ational change and observing the most parsimonious route between di€erent organisational forms. This fundamental, but important insight could result in organisational cladograms being used as a tool within a change framework, for achieving successful organisational design and change. Thus, regardless of the industrial sector, organisations could use clado- grams as an evolutionary analysis technique for deter- mining `where they have been and where they are now''. This evolutionary analysis could be used to for- mulate coherent and appropriate action for managers who are organisational architects and planners.

References
[1] Carper WB, Snizek WE. The nature and types of organisational taxonomies: an overview. Acad Manage Rev 1980;5(1):66±75. [2] Good IJ. Categorisation of classification. In: Mathematics and computer science in medicine and bi- ology. London: H.M.S.O, 1965. p. 115±28. [3] Cormack RM. A review of classification. Proceedings of the Royal Statistical Society 1971;3:321±67. [4] Haas J, Hall R, Johnson N. Toward an empirically derived taxonomy of organisations. In: Bovers R, editor. Studies on behaviour in organisations. Athens, GA: University of Georgia Press, 1966. p. 157±80. [5] McKelvey B, Guidelines for the empirical classification of organisations. Adm Sci Q. 1975;20:509±25. [6] Chrisman J, Hofer C, Boulton W. Toward a system for classifying business strategies. Acad Manage Rev 1988;13(3):413±28. [7] Gordon CW, Babchuk N. A typology of voluntary organisations. Am Sociol Rev 1958;24:22±3. [8] Emery FE, Trist EL. The casual texture of organisational environments. Human Relat 1965;18:21±32.

6. Summary Although classification is an habitual process which all humans do, the use of classifications in organis- ational science has not reached the same status as the classifications which exist in physics, chemistry and bi- ology. This paper has sought to describe and justify the benefits of organisational classifications and in par- ticular cladistic classifications of manufacturing sys- tems.

I. McCarthy et al. / Omega 28 (2000) 77±95 [9] Thompson JD. Organisations in action. New York: McGraw-Hill, 1967. [10] Perrow C. Organisational analysis: a sociological review. Belmont, CA: Brooks/Cole, 1970. [11] Van Ripper PP. Organisations: basic issues and proposed typology. In: Bowers RV, editor. Studies on behaviour in organisations. Athens: University of Georgia Press, 1966. [12] Constable CJ, New CC. Operations management, a systems approach through text and cases. John Wiley &Sons, 1976. [13] Wild R. The techniques of production management. London: Holt, Reinhart and Winston, 1971. [14] Johnson LA, Montgomery DC. Operation research in production planning, scheduling and inventory control. New York: John Wiley & Sons, 1974. [15] De Toni A, Panizzolo R. Repetitive and intermittent manufacturing: comparison of characteristics. In: Integrated manufacturing systems, vol. 3. MCB University Press, 1992. p. 23±37 (No. 4). [16] Schmitt TG, Klastorin T, Shtub A. Production classification system: concepts, models and strategies. Int J Prod Res 1985;23(3):563±78. [17] Ingham H. Balancing sales and production: models of typical business policies. Management Publication, 1971 [ch 1±2]. [18] Wild R. Production and operations management. Cassel Ed, 1989 [ch 1]. [19] Aneke NAG, Carrie AS. A comprehensive ¯owline classification scheme. Int J Prod Res 1984;22(2):282±97. [20] Burbidge JL. International Seminar On Group Technology, Final report. Turin International Centre, Turin, Italy, 1970. [21] Frizelle GDM. OPT in perspective. In: Advanced manufacturing engineering, 1. Butterworth and Co, 1989. [22] Barber KD, Hollier RH. The use of numerical analysis to classify companies according to production control complexity. Int J Prod Res 1986;24(1):203±22. [23] Woodward J. In: Industrial organisation, theory and practice. Oxford University Press, 1980. p. 22±49. [24] Burbidge JL. The principles of production control, 4th ed. Plymouth, UK: MacDonald & Evans, 1962. [25] Hitomi K. Manufacturing systems engineering (a unified approach to manufacturing technology and pro- duction management), 2nd ed. London: Taylor and Francis, 1996. [26] Spencer MS, Cox JF. An analysis of the product±process matrix and repetitive manufacturing. Int J Prod Res 1995;33(5):1275±94. [27] McCarthy IP. Manufacturing classifications: lessons from organizational systematics and biological taxon- omy. Int J Manuf Technol Manage Ð Integrated Manuf Sys 1995;6(6):47±8. [28] Romanelli E. The evolution of new organizational forms. In: Annual review of sociology, 17. Annual Reviews, 1991. p. 79±103. [29] Ridley M. Evolution. Blackwell Scientific Publications, 1993. [30] Wiley EO, Siegel-Causey D, Brooks DR, Funk VA. The compleat cladist Ð a primer of phylogenetic pro-

93

[31] [32]

[33] [34] [35]

[36]

[37]

[38] [39]

[40] [41]

[42] [43]

[44]

[45]

[46]

[47]

[48] [49] [50] [51] [52]

cedures. In: Special Publications No 19. The University of Kansas Museum of Natural History, 1991. De Queiroz K. Systematics and the Darwinian revolution. Philos Sci 1988;55:238±59. EPSRC (1996), Blueprint Ð The control design and production newsletter of EPSRC, Issue No. 9, July 1996. Hannan MT, Freeman J. The population ecology of organisations. Am Sociol Rev 1977;83:929±84. Hannan MT, Freeman J. Organisational Ecology. Cambridge, MA: Harvard University Press, 1989. Baum JAC. A population perspective organizations: a study of diversity and transformation in child care ser- vice organisations. Ph.D. dissertation, Faculty of Management, University of Toronto, 1989. McKelvey B. Organisational systematics: taxonomy, evolution, classification. Berkeley: University of California Press, 1982. Lumsden CJ, Singh JV. The dynamics of organizational speciation. In: Singh JV, editor. Organizational evol- ution: new directions. Newbury Park, CA: Sage, 1990. p. 145±63. Brodie R. The virus of the mind: the new science of the meme. Integral Press, 1995. McCarthy IP. The development of a manufacturing classification using concepts from organisational sys- tematics and biological taxonomy. Ph.D. dissertation, University of She•eld, UK, 1995. Je€rey C. Biological nomenclature, 3rd ed. Systematics Association, Chapman and Hall, 1977. Forey PL, Humphries CJ, Kitching IJ, Scotland RW, Siebert DJ, Williams DM. Cladistics: a practical course in systematics. Oxford: Clarendon Press, 1992. Minelli A. Biological systematics the state of the art. Chapman & Hall, 1994. Sneath P, Sokal R. Numerical taxonomy, the principles and practices of numerical classification. Freeman, 1973. Rao HV, Reddy M. University manuscript. Density and organizational mortality in technologically hetero- geneous industries. Emory University, GA, USA, 1992. Hannan MT, Freeman J. Organizations in industry: strategy, structure and selection. Oxford University Press, 1995. Scott WR. Organizations: rational, natural and open systems, 3rd ed. Englewood Cli€s, NJ: Prentice Hall, 1992. Hannan, Carroll, Dundon, Torres. Organizational evolution in multinational context: automobile manufac- turers in Belgium, Britain, France, Germany, and Italy. Am Sociol Rev 1995;88:234±53. Cusumano MA. The Japanese automobile industry. Cambridge, MA: Harvard University Press, 1985. Flink JJ. The automobile age. Cambridge, MA: MIT Press, 1988. Laux JM. The European automobile industry. New York: Twayne, 1992. Rae JB. The American automobile manufacturers: the first forty years. Philadelphia: Chiltern, 1959. Hounshell DA. From the American system to mass

94

I. McCarthy et al. / Omega 28 (2000) 77±95 production. Baltimore: Johns Hopkins University Press, 1984. Womack JP, Jones DT, Roos D. The machine that changed the world. New York: Macmillan Publishing, 1990. Fiat Group. Financial overview, January 30, 1998. Fiat Group. Annual report, 1996. Fiat Group. Report of the Board of Directors on oper- ations in the first half of 1997. Ford Motor Company. Annual report, 1965. Ford Motor Company. Annual report, 1975. Ford Motor Company. Annual report, 1985. Ford Motor Company. Annual report, 1995. General Motors. A look at General Motors today, 1996. General Motors. What drives General Motors. Annual report, 1996. General Motors. The EV1 electric vehicle, teamwork in action. Annual report, 1995. Honda. Annual report, 1995. Mercedes-Benz. Annual report, 1995. Mitsubishi Corporation. The principles that define Mitsubishi Corporation. Annual report, 1996. Mitsubishi Corporation. Annual report, 1995. Nissan. Even higher customer satisfaction. Annual report, 1995. Peugeot Motor Company Plc. Annual review, 1995. Peugeot Motor Company Plc. Statement of accounts and annual report, 1995. Peugeot Motor Company Plc. Annual review, 1996. Peugeot Motor Company Plc. Statement of accounts and annual report, 1996. Renault SA. Annual report, 1995. Toyota. Here's how we are getting better and even bet- ter. Annual report, 1996. Toyota. You ain't seen nuthin' yet! Annual report, 1995. Volkswagen AG. Annual report, 1996. Volkswagen, AG. Annual report, 1995. Volvo. Annual report, 1996. [86] Quicke DLJ. Principles and techniques of contemporary taxonomy. Chapman and Hall, 1993. [87] Swo€ord DL, Maddison WP. Reconstructing ancestral states under Wagner parsimony. Math Biosci 1987;87:199±299. [88] Watrous LE, Wheeler QD. The out-group comparison method. Syst Zool 1981;30:1±11. [89] Felsenstein J. Parsimony in systematics: biological and statistical issues. Ann Rev Ecol System 1983;14:313±33. [90] Maddison WP, Maddison DR. MacClade Version 3. Analysis of phylogeny and character evolution. MA, USA: Sinauer Associates, 1992. [91] Sagasti F. A conceptual and taxonomic framework for the analysis of adaptive behaviour. General systems, vol. XV, 1970. p. 151±60. [92] McCarthy IP, Leseure M, Ridgway K, Fieller N. Building manufacturing cladograms. International Journal of Technology Management 1997;13(3):269±86. [93] Stevenson W. Production/operations management, 4th ed. Homewood, IL: Irwin, 1993. [94] Hum S, Ng Y. A study on just-in-time practices in Singapore. Int J Oper Prod Manage 1995;15(6):5±24. [95] Hendry LC. World class in the make-to-order sector. MESELA '97 Conference, 22±24 July, 1997, Loughborough, ISBN 1 86058 0661, 1997. p. 41±6. [96] de Pinna M. Concepts and tests of homology in the cladistics paradox. Cladistics 1991;7:367±94. [97] Gri•ths P. Cladistics and functional explanation. Philo Sci 1994;61:206±27. [98] Grime P. The C±S±R model of primary plant strategies: origins, implications and tests ch 14. In: Gottlieb LD, Kain SK, editors. Plant evolutionary biology. London: Chapman and Hall, 1988. [99] Im J, Lee S. Implementation of just-in-time systems in US manufacturing firms. Int J Prod Res 1989;28(6):5± 14. [100] Camp R. Benchmarking, the search for industry best practices that lead to superior performance. Milwaukee, WI: ASQC Quality Press, 1989. [101] Suarez F, Cusumano M, Fine C. An empirical study of ¯exibility in manufacturing. In: Sloan management review, 1995. p. 25±32. [102] Armstrong J. Strategic planning improves manufacturing performance. In: Long-range planning, 1991. p. 127±9. [103] Powell T. Strategic planning as competitive advantage. In: Strategic Manage J, 1992. p. 551±8. [104] Waalevwijn P, Segaar P. Strategic management: the key to profitability in small companies. In: Long-range planning, 1993. p. 24±30. [105] Grinyer P, Norburn D. Planning for existing markets: perceptions of executives and financial performance (pt. 1). J R Stat Soc A 1975;138:70±81. [106] Kallman E, Shapiro H. The motor freight industry: a case against planning. In: Long-range planning, 1978. p. 81±95. [107] Kudla J. The e€ects of strategic planning on common stock returns. In: Acad Manage J, 1980. p. 5±32. [108] Leontiades M, Tezel A. Planning perceptions and planning results. In: Strategic Manage J, 1980. p. 65±79.

[53]

[54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] [65] [66] [67] [68] [69] [70] [71] [72] [73] [74] [75]

[76] [77] [78] [79] Gibson JL, Ivancevich JM, Donnelly JR. Organizations: behaviour, structure, processes, 7th ed. Homewood IL: Irwin, 1991. [80] Price JL, Mueller CW. Handbook of organisational measurement. Marshfield, MA: Pitman, 1986. [81] Baum JAC, Singh JV. Evolutionary dynamics of organizations. Oxford University Press, 1994. [82] Hull DL. The natural system and the species problem. In: Sibley CG, editor. Systematic biology. Proceedings of an International Conference Conducted At The University of Michigan, June 14±16, 1967. p. 56±61. [83] Pugh D, Hickson D, Hinings C, Turner C. Dimensions or organizational structure. Adm Sci Q 1968;13:65±105. Sells S. [84] Toward a taxonomy of organizations. In: Cooper W, Leavitt H, Shelly M, editors. New perspec- tives in organizational research. New York: Wiley, 1964. p. 515±32. Warriner C, editor. Empirical taxonomy of organiz- ations: [85] problematics in their development. Presented at the Roundtable Discussion, Annual Meeting of the American Sociological Association, Boston, 1979.

I. McCarthy et al. / Omega 28 (2000) 77±95 [109] Rue L, Fulmer R. Is long-range planning profitable? In: Academy of Management Proceedings, 1973. p. 66±89. [110] Miles R, Coleman H, Douglas C. Keys to success in corporate redesign. Calif Manage Rev 1995;37(3):128± 45.

95

[111] Monden Y. Toyota production system: practical approach to production management. Industrial Engineering and Management Press, Institute of Industrial Engineers, 1983.



doc_321018770.docx